The comprehensiveThe comprehensive prompt playbookprompt playbook for performance marketers

Table of contents Introduction 03 How to get started 04 Chapter 1: Monitor 06 Find what needs attention across channels Chapter 2: Diagnose 12 Understand why performance changed Chapter 3: Pace 19 Keep budgets and performance targets on track Chapter 4: Allocate 25 Decide where budget can create the most value Chapter 5: Optimize search 31 Cut wasted spend and capture better demand Chapter 6: Optimize social and video 37 Improve creative and audience performance Chapter 7: Plan 45 Build new campaigns from previous performance Chapter 8: Measure outcomes 51 Link ad spend to on-site and revenue outcomes Chapter 9: Act 57 Execute, test, and measure changes Chapter 10: Communicate 64 Turn performance data into clear decisions

Introduction For performance marketers, the challenge isn’t simply monitoring the latest numbers. It’s knowing what deserves attention, understanding why performance changed, and deciding what to do next. AI can accelerate that work, but only when it uses current data and follows a disciplined process. This playbook is built around that process. Supermetrics for Claude and Supermetrics for ChatGPT connect your marketing data directly to the AI tools you already use. You can investigate performance in plain English, monitor budgets, find optimization opportunities, and create evidence-backed campaign plans without repeatedly exporting and preparing data. Treat your AI tool like a capable new member of your performance team. Give it clear targets, consistent definitions, relevant business context, and firm approval boundaries. This playbook gives you the prompts to do exactly that. 3

How to get startedHow to get started Before diving into the workflows, the first step is to connect your marketing data to Claude or ChatGPT. The installation process depends on whether you're new to Supermetrics or are an existing user. Already use Supermetrics?Already use Supermetrics? 1. Connect Supermetrics to Claude or ChatGPT. 2. Sign in and authenticate your Supermetrics account. 3. Start using the prompts in this playbook. New to Supermetrics?New to Supermetrics? 1. Start a free Supermetrics trial. 2. Connect your marketing data sources. 3. Connect Supermetrics to Claude or ChatGPT. 4. Start using the prompts in this playbook. Start a Supermetrics for Claude free trialStart a Supermetrics for Claude free trial Start a Supermetrics for ChatGPT free trialStart a Supermetrics for ChatGPT free trial Connect SupermetricsConnect Supermetrics to Claudeto Claude 1. Open Customize. 2. Select Connectors. 3. Search for Supermetrics. 4. Add Supermetrics and sign in. Connect SupermetricsConnect Supermetrics to ChatGPTto ChatGPT 1. Open Settings. 2. Select Plugins. 3. Hit Browse plugins. 4. Search for Supermetrics. 5. Install plugin and sign in. Once the Supermetrics connector is active, you are ready to start pulling your performance data and using the prompt workflows in this playbook. Note that if you are a new user, you may need to connect your marketing data sources to Supermetrics. For more instructions, check out our support documentation. 4

Calculation rulesCalculation rules Before using the prompts, add these calculation rules to your Claude project instructions or ChatGPT custom instructions. They help the AI calculate performance metrics consistently and avoid common errors when combining data. Calculate CTR, CPC, conversion rate, CPA, and ROAS from the underlying totals rather than averaging reported ratios. State the denominator used for conversion rate. Don’t sum reach across overlapping entities or treat frequency and impression share as additive. Don’t combine monetary values from different currencies unless they have been converted using a stated exchange-rate source and date. 5

01 C H A P T E R 1 Monitor Find what needs attention across channels Performance marketers can’t investigate every movement across every campaign. They first need to separate material changes from normal variation and low-volume noise. These prompts use Supermetrics to review the latest performance data, compare results with relevant periods and targets, and surface the channels, campaigns, or metrics that deserve attention. Use them to decide where to investigate next. Diagnose why performance changed in Chapter 2 before deciding what to change. 6

C H AP T E R 1 — M O N I TO R PROMPT 01 Build a cross-channel performance overview Use Supermetrics to retrieve performance data for: Sources and accounts: [INSERT] Current and comparison periods: [INSERT] Primary business outcome: [INSERT] Cross-channel measurement source: [GA4 / CRM / PLATFORM DATA BY CHANNEL / OTHER] Currency: [INSERT] Retrieve campaign name, date, impressions, clicks, spend, primary outcomes, and attributed value where available. Calculate CTR, CPC, conversion rate, cost per outcome, and ROAS from the underlying totals. Return: Channel | Spend | Primary outcomes | Attributed value | Cost per outcome | ROAS | Change versus comparison Then summarize the largest changes, the channels driving the overall result, and any important difference between scale and efficiency. Use the stated measurement source for cross-channel totals. If outcome definitions differ or could overlap, show each channel separately rather than creating a blended total. Flag material gaps in data, currency, time zone, or attribution settings. Finish with the headline result and the areas that need investigation. 7

C H AP T E R 1 — M O N I TO R PROMPT 02 Run a daily performance check Use Supermetrics to retrieve the latest complete daily data for: Sources and scope: [INSERT] Primary KPIs and targets: [INSERT] Materiality threshold: [INSERT, IF KNOWN] Minimum spend or outcome threshold: [INSERT] Expected conversion lag: [INSERT, IF KNOWN] Compare the latest complete day with the same weekday last week, the previous seven-day average, and the supplied targets. Flag material changes in spend, delivery, traffic, outcomes, revenue, CPA, or ROAS. Also flag campaigns with unusually low delivery, meaningful spend without outcomes, or missing or implausible data. Ignore normal daily variation and percentage changes driven by low volume. Treat recent outcome data as provisional when conversion lag applies. Return: Priority | Channel or campaign | Metric | Latest result | Benchmark | Variance | Why it matters | Next check Use High, Medium, or Low priority based on likely business impact. If nothing material requires attention, state: No material exceptions found in the latest complete data. Make it recurring: If your AI tool supports scheduled runs with access to Supermetrics, schedule this check for a time after your source data has refreshed. Use the same accounts, KPIs, targets, and alert thresholds on every run. Ask it to report only material exceptions (or confirm that none were found) and review the first few runs to refine the thresholds. You can also adapt the prompt for a weekly check using the latest complete week. 8

C H AP T E R 1 — M O N I TO R PROMPT 03 Audit campaign naming conventions Use Supermetrics to audit: Platforms, accounts, and campaigns: [INSERT] Required naming format and allowed values: [INSERT] Approved exceptions or legacy campaigns: [INSERT, IF ANY] Fields that must be consistent across channels: [INSERT] Retrieve campaign names, IDs, platform, status, and recent spend. Check each name against the supplied format. Identify missing fields, invalid values, inconsistent abbreviations, and names that cannot be parsed reliably. Return: Platform | Campaign ID | Current name | Issue | Reporting impact | Proposed name | Priority Prioritize active campaigns whose names affect reporting or campaign identification. Use campaign IDs to distinguish similarly named campaigns. Propose a new name only when the correct values are supported by the supplied rules or other verified information; otherwise mark the field Needs confirmation. Finish with the recurring naming problems and any rule that needs clarification. Do not rename campaigns without explicit approval. PROMPT 04 Check performance against KPI targets Use Supermetrics to compare performance with: Sources and scope: [INSERT] Period: [INSERT] KPI targets: [INSERT] Target-to-date: [INSERT, IF AVAILABLE] Minimum data threshold: [INSERT] Return: Scope | KPI | Actual | Target to date | Absolute variance | Percentage variance | Status Classify each result as Meeting or exceeding target, Below target, or Insufficient data. Apply the correct direction to each KPI: lower CPA is better, while higher ROAS, outcomes, and revenue are better. Calculate CPA and ROAS from the underlying totals. If only a full-period volume or revenue target is supplied, state the current result without assuming linear progress and flag it for pacing analysis. Finish with the gaps that matter most commercially, any stronger areas offsetting them, and targets that can’t yet be assessed reliably. 9

C H AP T E R 1 — M O N I TO R PROMPT 05 Find the campaigns driving the overall result Use Supermetrics to retrieve campaign-level data for: Sources and accounts: [INSERT] Current and comparison periods: [INSERT] Primary business outcome and measurement source: [INSERT] Primary efficiency KPI: [INSERT] Minimum spend or outcome threshold: [INSERT] Establish how the overall account or channel result changed. Then identify the campaigns that contributed most to that movement. Return: Campaign | Spend change | Outcome change | Efficiency change | Contribution to net outcome change | Why it matters | Confidence Rank campaigns by absolute business impact, not percentage change alone. Separate major contributors from small-volume movements, new or paused campaigns, and campaigns without enough data. For additive metrics such as spend, conversions, or revenue, show each campaign’s absolute change. Calculate its percentage contribution to the net change only when the result is meaningful. Explain when contributions are negative or exceed 100% because improvements in some campaigns offset declines elsewhere. If the net change is zero or close to zero, don’t calculate contribution percentages. Show the absolute positive and negative movements instead. CPA and ROAS aren’t additive. Explain how changes in spend, outcomes, or conversion value affected the overall ratio rather than assigning each campaign a percentage share of the CPA or ROAS movement. Finish with: Campaigns supporting the overall result Campaigns weakening it Campaigns requiring further investigation 10

C H AP T E R 1 — M O N I TO R PROMPT 06 Produce the weekly cross-channel performance report Create a weekly report for: Reporting and comparison periods: [INSERT] Sources and scope: [INSERT] Primary business outcome and KPI targets: [INSERT] Cross-channel measurement source: [INSERT] Audience: [INSERT] Use Supermetrics to refresh the latest complete data. Reuse analysis already completed for these periods rather than repeating it. Use this structure: Performance summary: The headline result in no more than three sentences. Scorecard: Channel, spend, primary outcomes, attributed value, CPA or ROAS, and change versus comparison. What changed: The movements with the greatest business impact. What needs attention: The campaigns or issues requiring investigation. Position against targets: Actual performance versus target. Next steps: The investigations or decisions required this week. State the latest complete date and any data issue that could change the interpretation. Use the stated measurement source for cross-channel totals and keep incompatible attribution views separate. 11

02 C H A P T E R 2 Diagnose Understand why performance changed Knowing that performance changed is only the starting point. Before adjusting budgets, bids, targeting, or creative, you need to understand what produced the movement. These prompts use Supermetrics to break top-line results into their component metrics and trace the change from channel level to the campaigns and segments behind it. Use them to separate supported explanations from hypotheses that need more evidence. 12

C H AP T E R 2 — D I AG N O S E PROMPT 01 Diagnose a performance change Use Supermetrics to investigate: Observed change and primary KPI: [INSERT] Scope: [ACCOUNT / CHANNEL / CAMPAIGN / MARKET] Current period and comparison: [INSERT] Outcome or value definition: [INSERT] Sources: [INSERT] Check that the periods are complete and comparable, allowing for conversion lag and any changes in attribution, currency, or metric definitions. Break the KPI into the component metrics that can change it. Start at channel or account level, then retrieve deeper breakdowns only for the segments contributing most to the movement. Determine whether the result came from a change within existing segments, a shift in spend or volume between segments, or both. Return: Driver | Evidence | Estimated contribution or effect | Affected scope | Confidence | Next check Finish with the deepest supported chain: Headline result → Component metric → Segment → Underlying driver Label anything not supported by the available data as a hypothesis. 13

C H AP T E R 2 — D I AG N O S E PROMPT 02 Diagnose an increase in CPA Use Supermetrics to investigate: Sources and scope: [INSERT] Current and comparison periods: [INSERT] Conversion action: [INSERT] CPA target or materiality threshold: [INSERT, IF KNOWN] Minimum spend or conversion threshold: [INSERT] Expected conversion lag: [INSERT, IF KNOWN] Retrieve spend, clicks, conversions, CPC, conversion rate, and CPA. Calculate CPA from total spend and conversions. First, compare the increase with normal variation across recent comparable periods. Account for conversion lag, volume, and business impact. If no material increase is established, say so and specify when to check again. Investigate sooner if the potential impact is substantial or tracking looks unreliable. If the increase is material, determine whether CPC rose, conversion rate fell, spend shifted toward higher-CPA segments, or conversion mix or tracking changed. Start with the channels and campaigns driving the movement, then inspect the relevant auction, keyword, audience, device, market, placement, or landing-page data. Return: Driver | Evidence | Estimated effect on CPA | Affected scope | Confidence | Next check Finish with the main supported driver and any material secondary contributors. Label unverified explanations as hypotheses. 14

C H AP T E R 2 — D I AG N O S E PROMPT 03 Investigate whether competition drove a CPC increase Use Supermetrics to investigate: Source, accounts, and campaigns: [INSERT] Current and comparison periods: [INSERT] Primary outcome and efficiency target: [INSERT] Minimum data threshold: [INSERT] Auction insights or other competitor data: [ATTACH OR IDENTIFY SOURCE, IF AVAILABLE] Retrieve spend, impressions, clicks, CPC, conversion rate, conversions, CPA, and available impression- share, rank, and Quality Score data. Compare complete, like-for-like periods and check whether the CPC increase exceeds normal variation. If auction insights are supplied, examine changes in competitor presence, overlap, and relative visibility for the affected campaigns. Check whether changes in your bids, budgets, targeting, keyword mix, or ad quality offer another explanation. Return: Campaign or segment | CPC change | Competitive evidence | Other possible driver | Outcome impact | Confidence | Next check State whether increased competition is a supported explanation, a plausible hypothesis, or unsupported by the available data. Identify the campaigns where further investigation would matter most. Do not estimate competitors’ spend from auction insights or treat your impression share as market-wide share of voice. 15

C H AP T E R 2 — D I AG N O S E PROMPT 04 Diagnose a decline in ROAS Use Supermetrics to investigate: Sources and scope: [INSERT] Current and comparison periods: [INSERT] Revenue or conversion-value definition: [INSERT] Currency: [INSERT] Retrieve spend, clicks, conversions, conversion value, CPC, conversion rate, average value per conversion, and ROAS. Determine whether ROAS declined because traffic costs rose, conversion rate fell, value per conversion decreased, spend shifted toward weaker segments, or measurement changed. Start with channel and campaign, then inspect the deeper dimension most likely to explain the movement. Separate performance changes within campaigns from changes in campaign or product mix. Keep platform-attributed value and GA4 revenue separate unless a validated method makes them comparable. Return: Driver | Evidence | Effect on cost, conversions, or value | Affected scope | Confidence | Next check Finish with the primary cause, material secondary causes, and any measurement issue affecting confidence. 16

C H AP T E R 2 — D I AG N O S E PROMPT 05 Diagnose a decline in conversion volume Use Supermetrics to investigate: Sources and scope: [INSERT] Current and comparison periods: [INSERT] Conversion action: [INSERT] Trace the change through: Impressions → Clicks → Sessions or landing-page visits → Conversions Retrieve spend, CTR, CPC, conversion rate, and the delivery metrics needed to locate the largest loss. For search, include budget and impression-share data where relevant. For social, include reach, frequency, CPM, and placement. Use GA4 to check whether the decline also appears in sessions and on- site outcomes. Clicks and sessions use different definitions, so compare their direction and scale rather than treating them as a perfectly matched funnel. Return: Stage | Current result | Comparison | Change | Main contributing segment | Estimated effect on conversions | Confidence | Next check Finish by stating whether the evidence points to lower demand, reduced delivery, weaker ad response, poorer post-click conversion, or measurement failure. 17

C H AP T E R 2 — D I AG N O S E PROMPT 06 Determine whether a change is real or caused by tracking Use Supermetrics to investigate: Observed change and metric: [INSERT] Period: [INSERT] Advertising and analytics sources: [INSERT] Expected conversion events: [INSERT] Check data freshness, missing dates or metrics, the point when reporting changed, conversion lag, and any change in event names, conversion definitions, or attribution settings. Compare the movement across independent sources and related metrics. Check whether spend and clicks continued while conversions stopped, and whether the gap is materially outside the normal historical range. Return: Check | Evidence | Interpretation | Confidence | Next technical check Classify the result as Likely performance change, Likely tracking problem, Combination of both, or Insufficient evidence. Advertising platforms and analytics tools won’t report identical totals. Focus on sudden, unusually large, or inconsistent differences rather than expecting a perfect match. Finish with what the evidence confirms, which decisions may be affected, and what remains uncertain. 18

03 C H A P T E R 3 Pace Keep budgets and performance targets on track Spending according to plan doesn’t guarantee that you’ll hit the desired business result. You need to understand how quickly the budget is being used, what that spend is likely to produce, and whether the target remains realistic. These prompts use Supermetrics to compare actual spend with the budget plan, forecast the likely end-of-period result, and identify constraints affecting delivery. Use them to spot pacing risks before they become difficult to correct. 19

C H AP T E R 3 — PAC E PROMPT 01 Check budget pacing Use Supermetrics to retrieve daily spend for: Sources and scope: [INSERT] Budget period and total budget: [INSERT] Channel or campaign allocations: [INSERT, IF AVAILABLE] Planned spend by date: [INSERT, IF AVAILABLE] Currency and latest complete date: [INSERT OR DETERMINE] Exclude incomplete days. Calculate actual spend, remaining budget, and the daily spend required to use the budget by period end. Where a spend plan exists, compare actual with planned spend to date. Forecast remaining spend from a recent representative run rate, adjusted for weekday patterns, campaign schedules, promotions, and known delivery changes. Add that forecast to actual spend to project the final total. Return: Channel or campaign | Budget | Planned spend to date | Actual spend | Pacing variance to date | Required daily spend | Projected final spend | Status Classify each as Overspend risk, On pace, Underspend risk, or Insufficient data based on the projected final spend and available plan. State the run rate and adjustments used. If no reliable spend plan exists, mark pacing variance to date as unavailable rather than assuming an even spend curve. Only assess campaign-level pace when a corresponding allocation exists, and treat shared budgets as one allocation. Finish with the total budget position, the largest risks, and any campaigns needing a delivery investigation. 20

C H AP T E R 3 — PAC E PROMPT 02 Forecast end-of-period spend and results Use Supermetrics to forecast: Sources and scope: [INSERT] Forecast period and budget: [INSERT] Primary outcome and target: [INSERT] Efficiency target: [INSERT] Relevant historical periods: [INSERT] Currency and latest complete date: [INSERT OR DETERMINE] Retrieve current daily performance and comparable historical data. Account for recent run rate, weekday patterns, seasonality, conversion lag, and material changes in CPC, conversion rate, or value per conversion. Forecast final spend, outcomes, attributed value, CPA or ROAS, and variance from budget and target. Return: Metric | Actual to date | Baseline forecast | Reasonable range | Target | Variance | Confidence State the main assumptions and explain why the historical periods are comparable. Keep observed results separate from forecast assumptions and avoid unsupported precision. Classify target attainment as On track, At risk, Requires unusually strong performance, Outside the historical range, or Insufficient data. Finish with the expected result, main uncertainty, and the earliest signal that should trigger an update. 21

C H AP T E R 3 — PAC E PROMPT 03 Identify what is constraining campaign delivery Use Supermetrics to investigate delivery constraints for: Sources and campaigns: [INSERT] Period: [INSERT] Primary outcome and efficiency target: [INSERT] Retrieve spend, budget, delivery, traffic, outcomes, and efficiency metrics. For paid search, include impression share lost to budget and rank where available. For paid social, include reach, frequency, CPM, and recent changes in delivery. Determine whether each material campaign appears constrained by budget, rank or auction competitiveness, limited demand, audience saturation, weak ad response, or another issue. Return: Campaign | Delivery change | Likely constraint | Evidence | Business impact | Confidence | Next check Distinguish direct evidence from proxy signals. Spending the daily budget doesn’t prove more efficient demand exists, while high frequency alone doesn’t prove saturation. Treat unavailable impression-share data as missing, not zero. Finish with campaigns that have credible headroom, those constrained for other reasons, and those where the cause remains unclear. 22

C H AP T E R 3 — PAC E PROMPT 04 Calculate what is required to hit the target Use Supermetrics to calculate what is required for the rest of the period. Sources and scope: [INSERT] Period and latest complete date: [INSERT] Total and remaining budget: [INSERT OR CALCULATE] Outcome target: [INSERT] CPA or ROAS target: [INSERT] Retrieve actual spend, outcomes, and value to date. Calculate the remaining outcome gap, days remaining, required daily spend and outcomes, and the CPA or ROAS required over the remaining period. Return: Requirement | Result to date | Remaining gap | Required run rate | Recent range | Best comparable result | Feasibility Classify feasibility as Consistent with recent performance, Achievable with improvement, Requires unusually strong performance, Outside the historical range, or Insufficient data. Show whether the targets conflict. For example, spending the full budget may be incompatible with the efficiency target at the required outcome rate. State any assumption that depends on additional spend retaining current efficiency. Finish with a direct verdict on whether the budget, outcome, and efficiency targets appear mutually achievable. 23

C H AP T E R 3 — PAC E PROMPT 05 Refresh a pacing forecast Update this forecast: Scope and sources: [INSERT] Original forecast and date: [INSERT] Latest complete date: [INSERT] Budget and targets: [INSERT] Use Supermetrics to compare actual performance since the forecast with its expected spend, outcomes, and efficiency. Identify which assumptions changed and whether conversion maturity explains any apparent miss. Return: Metric or assumption | Previous forecast | Actual or current evidence | Updated forecast | Change | Reason Update the final spend, outcomes, CPA or ROAS, variance from budget and target, reasonable range, and confidence level. Revise the forecast only when the evidence has materially changed. Treat one anomalous day or immature conversion data as uncertainty rather than a new trend. Finish with the updated baseline, main changed assumption, remaining uncertainty, and whether the pacing risk has increased or decreased. 24

04 C H A P T E R 4 Allocate Decide where budget can create the most value The campaign with the best average CPA or ROAS isn’t necessarily the best place for the next pound or dollar. It may have limited demand left to capture, or its efficiency may decline as spend increases. These prompts use Supermetrics to assess your current allocation, identify credible headroom, and model the likely effect of adding, removing, or moving budget. Use them to compare the expected value of different options before making a live change. 25

C H AP T E R 4 — AL L O C AT E PROMPT 01 Evaluate the current budget allocation Use Supermetrics to evaluate: Sources and scope: [INSERT] Current and historical periods: [INSERT] Primary outcome and efficiency target: [INSERT] Total budget and currency: [INSERT] Campaign roles and constraints: [BRAND / PROSPECTING / RETARGETING / OTHER] Retrieve spend, outcomes, attributed value, CPA, and ROAS by channel and campaign. Include qualified outcomes or realized revenue where connected data allows. Assess current contribution, historical efficiency, performance at different spend levels, recent stability, and delivery headroom. Compare campaigns with similar roles rather than treating brand, prospecting, and retargeting as interchangeable. Return: Channel or campaign | Role | Current spend | Spend share | Outcome share | Efficiency | Evidence of headroom | Assessment Classify each as Potential to test higher spend, Allocation appears reasonable, Possible reduction candidate, Investigate further, or Insufficient evidence. Compare outcome shares only when the definitions are consistent and non-duplicative. Finish with the best-supported parts of the allocation, credible opportunities, and areas where the evidence is weak. 26

C H AP T E R 4 — AL L O C AT E PROMPT 02 Decide where to place incremental budget Use Supermetrics to allocate: Additional budget and period: [INSERT] Eligible channels or campaigns: [INSERT] Primary outcome and efficiency guardrail: [INSERT] Maximum change by campaign: [INSERT, IF ANY] Currency, constraints, and sources: [INSERT] Retrieve current and historical performance. Assess outcome volume, stability, performance at comparable spend levels, delivery headroom, and downstream value where available. Estimate how much additional budget each option could absorb and the likely marginal CPA or ROAS. Where the data can’t support that estimate, recommend a staged test. Create: Conservative option: Favor confidence and efficiency. Recommended option: Best fit for the stated objective. Growth option: Accept more uncertainty for additional volume. Return: Option | Channel or campaign | Current budget | Additional budget | Incremental outcome range | Marginal CPA or ROAS | Confidence Allocate only the amount the evidence suggests can be absorbed within the efficiency guardrail. If the full amount can’t be allocated responsibly, show the remainder as unallocated and explain why. Do not apply the allocation without explicit approval. 27

C H AP T E R 4 — AL L O C AT E PROMPT 03 Reduce budget with the least expected impact Use Supermetrics to model: Current budget and required reduction: [INSERT] Period and eligible activity: [INSERT] Campaign roles and objectives: [INSERT FOR EACH CAMPAIGN] Protected activity or minimum commitments: [INSERT] Primary outcome and efficiency guardrails: [INSERT] Currency and sources: [INSERT] Retrieve current and historical spend, outcomes, efficiency, and downstream value where available. Estimate where budget can be removed with the smallest expected outcome loss. Consider each campaign’s stated role, absolute contribution, performance at lower spend levels, learning or delivery risks, and conversion lag. Use marginal performance where possible rather than average CPA or ROAS. Do not infer a campaign’s role from its name alone. If roles are missing or unclear, flag the affected campaigns and ask for that context before recommending cuts to them. Return: Channel or campaign | Role | Current budget | Reduction | New budget | Expected outcome loss | Marginal efficiency of removed spend | Range | Confidence Ensure the reductions total the required amount and protect all stated commitments. Then estimate the effect on total outcomes and blended efficiency, keeping incompatible attribution views separate. Finish with the recommended reduction, expected business impact, most sensitive assumption, and main risk. Do not apply the reduction without explicit approval. 28

C H AP T E R 4 — AL L O C AT E PROMPT 04 Model a proposed budget reallocation Use Supermetrics to model: Budget removed from: [INSERT CHANNEL OR CAMPAIGN AND AMOUNT] Budget added to: [INSERT CHANNEL OR CAMPAIGN AND AMOUNT] Period: [INSERT] Primary outcome and efficiency guardrails: [INSERT] Currency and sources: [INSERT] Confirm that the amount removed equals the amount added. Retrieve performance at comparable spend levels for both sides of the move. Estimate the outcomes lost and gained, the net business effect, and the change in blended CPA or ROAS. Check delivery headroom, campaign role, conversion lag, possible learning disruption, and downstream value. Return: Campaign | Current budget | Proposed budget | Expected outcome change | Marginal CPA or ROAS | Range | Main assumption | Confidence Classify the proposal as Test as proposed, Test a smaller move, Evidence doesn’t support the move, or Insufficient evidence. Finish with the expected net effect and the evidence that would confirm or challenge the recommendation. Do not apply the reallocation without explicit approval. 29

C H AP T E R 4 — AL L O C AT E PROMPT 05 Recommend and stress-test a complete budget allocation Use Supermetrics to recommend how to allocate: Total budget and period: [INSERT] Eligible channels or campaigns and their roles: [INSERT] Primary objective, outcome target, and efficiency guardrails: [INSERT] Minimum, maximum, or committed allocations: [INSERT] Currency, constraints, and sources: [INSERT] Retrieve current and historical performance. Build the allocation using business contribution, marginal efficiency at comparable spend levels, delivery headroom, downstream value, and the supplied constraints. Return: Channel or campaign | Role | Current budget | Proposed budget | Change | Expected outcome range | Marginal CPA or ROAS | Confidence Ensure the proposed allocation respects the supplied constraints. If the full budget can’t be deployed within the efficiency guardrails, state the conflict and show how much can be allocated credibly. Stress-test it under a base case, a higher-cost case, and a lower-conversion or value case. Use changes drawn from relevant historical volatility where available. For each case, show expected outcomes, blended efficiency, and target status. Finish with the recommended allocation, reasonable result range, main downside risk, and any change that should be introduced in stages. Do not apply the allocation without explicit approval. 30

05 C H A P T E R 5 Optimize search Cut wasted spend and capture better demand Paid-search performance depends on what people search for, which keywords match those queries, and how relevant your ads and landing pages are. Account-level metrics can hide the terms and keywords creating waste or limiting growth. These prompts use Supermetrics to analyze search terms, keywords, match types, Quality Score, impression share, and ad performance. Use them to find wasted spend, uncover valuable demand, and create better ads based on previous results. 31

C H AP T E R 5 — O P T I M I Z E S E ARC H PROMPT 01 Find the biggest paid-search optimization opportunities Use Supermetrics to review: Source, accounts, and campaign types: [INSERT] Current and comparison periods: [INSERT] Primary conversion and efficiency target: [INSERT] Currency and minimum data threshold: [INSERT] Retrieve campaign, ad-group, keyword, and relevant search-term performance. Include spend, traffic, conversions, conversion value, CTR, CPC, conversion rate, CPA, ROAS, impression-share metrics, Quality Score, and match type where available. Identify wasted spend, strong areas with room to grow, major contributors to weak account performance, and activity worth protecting. Account for conversion lag and possible tracking issues. Return: Priority | Campaign, ad group, or keyword | Issue or opportunity | Evidence | Recommended action or test | Potential impact | Confidence Classify each as Protect, Test expansion, Reduce or exclude candidate, Improve, or Investigate further. Rank opportunities by absolute business impact rather than extreme ratios on small volumes. Finish with the five actions or tests most likely to improve the account. 32

C H AP T E R 5 — O P T I M I Z E S E ARC H PROMPT 02 Audit search terms and find negative-keyword candidates Use Supermetrics to retrieve search-term performance for: Source and scope: [INSERT] Period: [INSERT] Business, products, audiences, and markets: [INSERT] Primary conversion and efficiency target: [INSERT] Minimum data threshold: [INSERT] Historic approved negatives and reasons: [INSERT OR RETRIEVE, IF AVAILABLE] Retrieve search terms, matched keywords, match types, campaign, ad group, spend, conversions, CPA or ROAS, and existing negatives where available. Find terms that are irrelevant, show consistently weak commercial intent, or have spent materially without acceptable outcomes. Group related queries into themes. Use historic approved negatives to understand the team’s exclusion rules, but assess each new term against its current context. Before proposing a negative, check whether the term converted elsewhere, whether conversion lag applies, whether an existing negative covers it, and whether the exclusion could block valuable variants. Return: Search term or theme | Campaign | Spend | Conversions | CPA or ROAS | Reason | Negative match type | Application level | Recommendation Classify each as Add as negative, Review manually, Keep active, or Insufficient evidence. Finish with the spend associated with the strongest candidates. Label it as spend under review, not guaranteed savings. Do not add negatives without explicit approval. 33

C H AP T E R 5 — O P T I M I Z E S E ARC H PROMPT 03 Find new keyword and match-type opportunities Use Supermetrics to identify valuable search terms that could become keywords. Source and scope: [INSERT] Period: [INSERT] Product or offer: [INSERT] Primary conversion and efficiency target: [INSERT] Markets or languages: [INSERT] Minimum data threshold: [INSERT] Retrieve search terms, matched keywords, match types, campaign structure, impressions, clicks, spend, conversions, conversion value, CPA, and ROAS. Find commercially relevant terms that perform consistently, aren’t already covered effectively, and could benefit from tighter bidding, messaging, or landing-page alignment. Group close variants by intent. Return: Proposed keyword | Search intent | Existing coverage | Match type | Campaign or ad group | Impressions or clicks | Spend | Conversions | CPA or ROAS | Confidence Separate strong candidates from ideas worth testing, terms already covered, and terms without enough evidence. Treat past search-term performance as evidence for testing, not a forecast. Do not add or edit keywords without explicit approval. 34

C H AP T E R 5 — O P T I M I Z E S E ARC H PROMPT 04 Diagnose Quality Score and impression-share losses Use Supermetrics to investigate: Source and scope: [INSERT] Period: [INSERT] Primary outcome and efficiency target: [INSERT] Retrieve keyword-level performance, Quality Score and its components, impression share lost to rank or budget, spend, conversions, CPA, and ROAS where available. Identify material opportunities linked to budget constraints, rank or auction competitiveness, expected CTR, ad relevance, or landing-page experience. Return: Keyword or ad group | Main issue | Supporting metrics | Business impact | Recommended action or test | Priority | Confidence Treat Quality Score as a diagnostic signal, not the goal. If only the current score is available, don’t use it to explain historical changes. Treat missing impression-share data as unavailable rather than zero. Prioritize issues likely to affect cost, visibility, or conversion performance. Route budget-related opportunities through the pacing and allocation workflows before recommending more spend. Finish with the highest-value improvements and the evidence needed to judge whether they worked. 35

C H AP T E R 5 — O P T I M I Z E S E ARC H PROMPT 05 Create new search ads from previous performance Use Supermetrics to analyze previous search ads and create new variants. Source and scope: [INSERT] Period: [INSERT] Product, offer, target keywords, and search intent: [INSERT] Primary conversion: [INSERT] Number of variants: [INSERT] Landing page, approved claims, and brand requirements: [INSERT] Ad format and character limits: [INSERT] Retrieve available headlines, descriptions, spend, CTR, conversions, conversion rate, CPA, and ROAS. Compare ads with similar intent, targeting, landing pages, and delivery conditions. Identify message angles, offers, proof points, calls to action, and structures associated with stronger click and conversion performance. Treat responsive search ad asset findings as directional unless controlled evidence supports a stronger conclusion. Create meaningfully different variants. For each, provide: Performance insight | Hypothesis | Search intent | Headlines | Descriptions | Landing-page alignment Use one main messaging hypothesis per variant. Use only claims and offers supplied in the brief, and follow the stated brand and platform requirements. Finish with a testing plan covering the control, variable, primary KPI, guardrails, and evidence required to choose a winner. Do not publish the ads without explicit approval. 36

06 C H A P T E R 6 Optimize social and video Improve creative and audience performance A rise in paid-social CPA could point to creative fatigue, audience saturation, higher auction costs, or weaker post- click performance. Looking at the account-level result won’t tell you which explanation is most likely. These prompts use Supermetrics to compare performance at campaign, ad-set, audience, and ad level. Use them to identify the ads driving results, detect fatigue, assess audience headroom, and turn previous performance into new creative tests. 37

C H AP T E R 6 — O P T I M I Z E S O C I AL AN D V I D E O PROMPT 01 Identify your strongest and weakest paid-social ads Use Supermetrics to retrieve: Sources and scope: [INSERT] Period and optional comparison period: [INSERT] Primary conversion and efficiency target: [INSERT] Currency and minimum data threshold: [INSERT] Retrieve performance by campaign, ad set, and ad. Include objective, format, spend, impressions, reach, frequency, clicks, CTR, CPM, conversions, conversion value, CPA, and ROAS where available. Compare ads within the same platform and with similar objectives, audiences, placements, active periods, and conversion definitions. Return: Ad | Campaign or ad set | Spend | Outcomes | CPA or ROAS | Scale | Performance versus comparable ads | Recommended action or test | Confidence Classify each as Protect, Test more scale, Refresh candidate, Reduce candidate, or Insufficient data. Account for uneven platform delivery. Finish with the ads driving the most value, those consuming the most inefficient spend, and promising ads that need a fairer test. 38

C H AP T E R 6 — O P T I M I Z E S O C I AL AN D V I D E O PROMPT 02 Detect creative fatigue Use Supermetrics to investigate: Sources and scope: [INSERT] Period and time breakdown: [INSERT] Primary conversion: [INSERT] Minimum data threshold: [INSERT] Retrieve time-series performance by ad, including spend, reach, frequency, CPM, CTR, conversion rate, CPA, and ROAS where available. Look for performance weakening as exposure grows. Compare each ad with similar ads running at the same time. Check whether changes in audience, placement, budget, optimization, offer, or landing page provide a better explanation. Use historical patterns rather than a universal frequency threshold. Return: Ad | Performance trend | Reach and frequency trend | Comparison with similar ads | Likely explanation | Recommended response | Business impact | Confidence Classify each finding as Likely fatigue, Possible audience saturation, Broader auction or post-click issue, Normal variation, or Insufficient evidence. Finish with the ads that need refreshing, those that remain effective, and the highest-value next test. 39

C H AP T E R 6 — O P T I M I Z E S O C I AL AN D V I D E O PROMPT 03 Analyze audience performance and saturation Use Supermetrics to analyze: Sources and scope: [INSERT] Current and comparison periods: [INSERT] Primary conversion and efficiency target: [INSERT] Currency: [INSERT] Retrieve performance by ad set or the closest audience-level dimension. Include spend, reach, frequency, CPM, CTR, conversions, CPA, and ROAS where available. Assess business contribution, efficiency, changes in reach and frequency, and whether higher spend produced proportional additional reach and outcomes. Compare prospecting, retargeting, and customer audiences separately. Return: Audience or ad set | Role | Spend | Reach | Frequency | CPA or ROAS | Saturation signal | Recommended action or test | Confidence Classify each as Test more scale, Refresh creative first, Test broader targeting, Reduce candidate, or Insufficient evidence. Treat ad-set performance as evidence about the full setup, not the audience alone. Account for creative, placement, objective, and optimization differences. Infer audience overlap only when the source supports it. Finish with the strongest audiences, the clearest saturation risks, and the next test for each priority audience. 40

C H AP T E R 6 — O P T I M I Z E S O C I AL AN D V I D E O PROMPT 04 Review and refresh customer audiences Use Supermetrics to review: Ad platforms and current audiences: [INSERT] CRM, analytics, and customer-data sources available: [INSERT] Audience purpose: [RETARGETING / CUSTOMER MATCH / SUPPRESSION / OTHER] Eligibility, consent, and exclusion rules: [INSERT] Refresh frequency and lookback period: [INSERT] Retrieve available aggregate data on audience size, membership freshness, match rate, reach, spend, and outcomes. Compare current lists with the stated rules and identify audiences that may be stale, too broad, too narrow, or missing important exclusions. Recommend which audiences to keep, refresh, split, combine, or test. For each recommendation, specify the inclusion and exclusion rules, source data needed, destination, expected benefit, and how to measure the result. Return: Audience | Purpose | Current issue | Proposed rule or refresh | Data required | Destination | Measurement plan | Priority Flag rules that cannot be checked with the available data. Do not include personal records in the response or assume that access to campaign reporting also grants access to CRM profiles or audience- list management. Finish with the highest-priority audience changes and the setup or approval needed to make them. Do not create, upload, or sync lists without explicit approval 41

C H AP T E R 6 — O P T I M I Z E S O C I AL AN D V I D E O PROMPT 05 Find patterns in high-performing social creative Use Supermetrics to analyze: Sources and scope: [INSERT] Period: [INSERT] Primary conversion and business objective: [INSERT] Minimum data threshold: [INSERT] Retrieve ad-level performance and available creative fields, including copy, headline, CTA, format, spend, CTR, conversions, CPA, and ROAS. State which fields and assets are accessible. Compare ads with similar objectives, audiences, placements, offers, and active periods. Look for repeatable patterns in hooks, customer problems or benefits, offers, proof, calls to action, copy structure, format, and visuals where accessible. Return: Creative pattern | Comparable ads using it | Performance evidence | Where it worked | Limitation | Confidence Separate patterns linked to stronger conversion performance from those that improve clicks without improving downstream results. Treat findings based on one ad as hypotheses, and don’t infer visual content from filenames or URLs. Finish with up to five principles and the next creative hypothesis each one suggests. 42

C H AP T E R 6 — O P T I M I Z E S O C I AL AN D V I D E O PROMPT 06 Create new paid-social ads from previous performance Use Supermetrics to analyze previous ads and create new paid-social concepts. Platform and scope: [INSERT] Historical period: [INSERT] Product or offer and target audience: [INSERT] Campaign objective and primary conversion: [INSERT] Number of concepts: [INSERT] Landing page, approved claims, and brand requirements: [INSERT] Target placements, formats, and available assets: [INSERT] Retrieve relevant ad performance and accessible creative fields. Identify messages, hooks, offers, and formats associated with stronger conversion performance. Note concepts showing fatigue or lacking enough delivery to judge. Create meaningfully different concepts based on the underlying findings rather than rewriting a previous winner. For each concept, provide: Performance insight | Hypothesis | Audience | Placement and format | Hook | Primary text | Headline | CTA | Visual direction | Landing-page alignment Use one main creative hypothesis per concept. Separate the historical finding from the new interpretation, use only approved claims and offers, and follow the platform requirements. Then provide: Test | Control | New variant | Variable | Primary KPI | Guardrail | Evidence required | Decision rule Use the primary business outcome to choose a winner. Treat CTR as a supporting metric when the objective depends on conversions, revenue, or lead quality. Do not publish the ads without explicit approval. 43

C H AP T E R 6 — O P T I M I Z E S O C I AL AN D V I D E O PROMPT 07 Diagnose paid-video performance Use Supermetrics to analyze paid-video campaigns for: Ad platforms and scope: [INSERT] Current and comparison periods: [INSERT] Campaign objective and primary outcome: [INSERT] Target and minimum data threshold: [INSERT] Video formats, placements, and creative IDs: [INSERT, IF AVAILABLE] Retrieve spend, impressions, views, view rate, cost per view, reach, frequency, video completion metrics, clicks, and downstream outcomes where available. State which metrics the source provides and keep different platforms’ view definitions separate. Compare campaigns and videos with similar objectives, formats, audiences, placements, and delivery periods. Identify whether weak performance appears in delivery cost, viewing and completion, click response, or downstream outcomes. Check whether changes in budget, targeting, placement, or measurement better explain the result before attributing it to the creative. Return: Campaign or video | Objective | Spend | Viewing and completion | Downstream result | Main finding | Recommended test | Confidence Finish with the videos worth protecting, those needing a new creative test, and any delivery or measurement issue to investigate. Judge performance against the campaign’s primary objective. Do not treat views or view-through conversions as proof of incremental business impact. 44

07 C H A P T E R 7 Plan Build new campaigns from previous performance Campaign planning doesn’t need to start with a blank page. Previous campaigns contain useful evidence about the audiences, messages, offers, and formats associated with stronger results. These prompts use Supermetrics to retrieve that evidence before developing new ideas. Use them to extract transferable lessons, adapt successful campaigns for other platforms, forecast likely outcomes, and build a campaign brief grounded in previous performance. 45

C H AP T E R 7 — P L AN PROMPT 01 Extract lessons from previous campaigns Use Supermetrics to identify lessons for the next campaign. Sources and accounts: [INSERT] Historical period: [INSERT] Objective and primary business outcome: [INSERT] Markets or audiences: [INSERT] Currency and minimum data threshold: [INSERT] Retrieve campaign performance, including spend, outcomes, efficiency, audience, landing page, format, and accessible creative fields. Group campaigns by comparable objectives, products, audiences, markets, offers, and periods. Identify approaches that created business value at meaningful scale, worked consistently, or drove engagement without downstream results. Return: Finding | Supporting campaigns | Evidence | Context | Implication for the next campaign | Confidence Distinguish repeatable findings from results driven by one campaign or unusual conditions. Keep different attribution views separate and analyze only creative elements you can access. Finish with what to repeat, what to avoid, and what still needs testing. Include up to five findings, returning fewer if the evidence is weak. 46

C H AP T E R 7 — P L AN PROMPT 02 Generate campaign concepts from previous results Use the findings from Prompt 1 or retrieve the required history through Supermetrics. Product or offer and target audience: [INSERT] Objective, primary conversion, and target platforms: [INSERT] Historical period and sources: [INSERT] Number of concepts: [INSERT] Landing page, approved claims, and brand or legal requirements: [INSERT] Audience research and available assets: [INSERT] Identify evidence about messages, offers, hooks, formats, calls to action, channel fit, and post-click performance. Treat audience motivations as hypotheses unless customer research supports them. Create meaningfully different concepts. Return: Concept | Audience insight | Central idea | Performance evidence | Channel fit | Test hypothesis | Main risk For each concept, provide the proposition, hook, CTA, recommended channels, formats, and landing-page direction. State which elements come from historical evidence and which are new creative decisions. Use only supplied features, offers, claims, and customer evidence. Rank the concepts by expected business value, strength of evidence, and usefulness of the learning they could produce. 47

C H AP T E R 7 — P L AN PROMPT 03 Adapt a successful campaign for another platform Use Supermetrics to adapt: Original campaign and platform: [INSERT] Target platform, audience, and market: [INSERT] Target-platform role and objective: [INSERT] Primary conversion and budget: [INSERT] Historical period: [INSERT] Placements, formats, landing page, and brand requirements: [INSERT] Retrieve the original campaign’s performance, accessible creative fields, audience information, and post- click outcomes. Identify what appears to have made the campaign valuable, separating: The central message or offer The creative execution The audience and targeting The platform’s delivery conditions The landing-page contribution Then retrieve relevant historical performance from the target platform. Identify comparable campaigns and the formats, messages, audiences, and calls to action associated with stronger results there. Return: Element | Role in original campaign | Keep, adapt, or replace | Target-platform execution | Evidence Create the adapted concept, including the audience, message angle, hook, copy, headline, CTA, format, visual direction, and landing-page recommendation. State the main transfer assumption and test hypothesis. If target-platform evidence is limited, label the adaptation as a test. Do not create or publish the campaign without explicit approval. 48

C H AP T E R 7 — P L AN PROMPT 04 Forecast a new campaign using comparable campaigns Use Supermetrics to forecast: Proposed campaign, objective, and channel: [INSERT] Audience, market, budget, and duration: [INSERT] Primary conversion and CPA or ROAS target: [INSERT] Launch period, currency, sources, and historical period: [INSERT] Find the three to five most comparable historical campaigns based on objective, offer, audience, market, budget, season, format, and landing page. Return: Historical campaign | Similarities | Important differences | Comparability | Performance Use the strongest benchmarks to create downside, baseline, and upside scenarios for delivery, traffic, conversions, CPA, conversion value, and ROAS. Return: Metric | Downside | Baseline | Upside | Target | Main assumption If no credible benchmark exists, present planning scenarios rather than a forecast. Keep historical results separate from assumptions and avoid unsupported precision. Finish with the expected outcome range, whether the target appears realistic, the main uncertainty, and the earliest result that should trigger a forecast update. 49

C H AP T E R 7 — P L AN PROMPT 05 Build an evidence-backed campaign brief Create a campaign brief for: Selected concept, product, and audience: [INSERT] Objective, primary conversion, and targets: [INSERT] Channels, market, budget, and period: [INSERT] Landing page and measurement source: [INSERT] Approved claims, brand requirements, and constraints: [INSERT] Use the analysis from Prompts 1–4. Use Supermetrics to refresh any important figure or benchmark that may be outdated. Build the brief using: Objective and audience: Define the business outcome and audience insight. Proposition: Set out the message, offer, hook, and CTA. Performance evidence: Show which historical findings informed the plan and where transferability is uncertain. Channel and creative plan: Define each channel’s role, formats, creative direction, and landing-page experience. Budget and timing: Show the allocation and schedule. Measurement: Define KPIs, guardrails, attribution source, and reporting cadence. Test plan: State the hypotheses, evidence required, and decision rules. Risks and requirements: List key assumptions, missing information, tracking needs, and approvals required before launch. Keep supplied facts, performance findings, and new hypotheses clearly separated. 50

08 C H A P T E R 8 Measure outcomes Link ad spend to on-site and revenue outcomes Advertising platforms show the results they attribute to your campaigns. Analytics and CRM platforms show what happens after someone reaches your site or enters your funnel. Looking at either view alone can produce an incomplete picture. These prompts use Supermetrics to connect paid-media data with site behavior, lead quality, and revenue. Use them to test whether platform-reported success is supported downstream while keeping different attribution methods clearly separated. 51

C H AP T E R 8 — M E AS URE O UTC O M E S PROMPT 01 Build a paid-to-site performance view Use Supermetrics to retrieve: Advertising sources and accounts: [INSERT] GA4 property: [INSERT] Period: [INSERT] Primary GA4 outcome and revenue metric: [INSERT] Currency and approved join keys: [INSERT] Retrieve advertising spend, clicks, platform outcomes, and attributed value. From GA4, retrieve campaign, landing page, sessions, engaged sessions, outcomes, and revenue. Aggregate the datasets to the highest reliable shared level before joining them. Use validated campaign IDs or UTM values, not similar campaign names. Report matched spend, unmatched traffic, duplicate matches, and material differences in time zone, currency, or tagging. Return: Channel or campaign | Matched spend | Ad clicks | Sessions | Click-to-session rate | Engaged sessions | GA4 outcomes | GA4 revenue | Cost per outcome | GA4-based ROAS | Match coverage Calculate ratios from underlying totals. Keep platform-attributed outcomes separate from GA4 outcomes. Finish with the strongest sources of on-site value, where the platform and site views disagree, and any tracking issue that limits a decision. 52

C H AP T E R 8 — M E AS URE O UTC O M E S PROMPT 02 Investigate the gap between ad clicks and site sessions Use Supermetrics to investigate: Advertising sources and accounts: [INSERT] GA4 property and period: [INSERT] Campaign identifiers: [INSERT] Acceptable variance: [INSERT, IF KNOWN] Retrieve ad clicks and spend alongside GA4 sessions by date, campaign, device, and landing page. Calculate click-to-session rate as GA4 sessions ÷ ad clicks, plus the absolute and percentage differences. Compare the result with the normal historical range when no threshold is supplied. Return: Segment | Ad clicks | GA4 sessions | Difference | Click-to-session rate | Spend affected | Severity | Next check Identify whether the largest gaps align with tagging problems, redirects, page-load issues, device or landing-page patterns, consent coverage, or a sudden reporting change. Treat these as hypotheses unless the retrieved data confirms them. Clicks and sessions won’t match exactly, and the rate can exceed 100%. Treat spend as affected by the discrepancy, not automatically wasted. Finish with where the gap appears abnormal and the next technical check required. 53

C H AP T E R 8 — M E AS URE O UTC O M E S PROMPT 03 Find landing pages weakening paid performance Use Supermetrics to analyze: Advertising sources and accounts: [INSERT] GA4 property: [INSERT] Current and comparison periods: [INSERT] Primary outcome: [INSERT] Campaign-to-page join key and minimum threshold: [INSERT] Retrieve paid clicks and spend alongside GA4 sessions, engagement, outcomes, and revenue by landing page and device. Assign spend to a page only when a validated URL or identifier supports the mapping. If a campaign sends traffic to several pages, don’t duplicate its full spend across them. Return: Landing page | Channel or campaign | Device | Matched spend | Sessions | Engagement rate | Conversion rate | Cost per outcome | Revenue | Recommended action or test | Confidence Identify pages with substantial traffic but weak engagement, strong engagement but poor conversion, or efficient results at meaningful volume. Highlight differences between platform-reported and on-site performance. Account for audience, offer, device, and campaign objective. Treat the page as a possible contributor unless a controlled test supports a causal conclusion. Calculate page-level cost per outcome and ROAS only when spend can be assigned reliably at that level. Otherwise, report the GA4 landing-page metrics without allocating spend or calculating page-level efficiency. Finish with the landing-page investigations or tests likely to create the greatest business impact. 54

C H AP T E R 8 — M E AS URE O UTC O M E S PROMPT 04 Connect paid campaigns to lead quality and revenue Use Supermetrics to evaluate downstream performance. Advertising and analytics sources: [INSERT] CRM or revenue source: [INSERT] Period and expected lead-to-revenue lag: [INSERT] Qualified outcome and revenue definition: [INSERT] Currency and approved join key: [INSERT] Retrieve campaign-level spend, leads, qualified leads, opportunities, customers, pipeline, and realized revenue where available. Validate the join and report matched and unmatched spend, leads, customers, and revenue. Use mature acquisition cohorts where possible and label recent outcomes as incomplete. Keep personal information out of the output. Return: Channel or campaign | Spend | Leads | Qualification rate | Opportunities | Customers | Lead-to-customer rate | Pipeline | Revenue | CAC | Business-value ROAS | Match coverage Calculate business-value ROAS from matched realized revenue and spend. Keep pipeline separate from realized revenue. Identify campaigns whose top-of-funnel efficiency isn’t supported by lead quality or revenue, as well as campaigns with higher acquisition costs but stronger downstream value. Finish with the budget or targeting decisions that could change under the downstream view and the attribution limits affecting them. 55

C H AP T E R 8 — M E AS URE O UTC O M E S PROMPT 05 Reconcile attribution and check for double counting Use Supermetrics to compare: Advertising sources and GA4 property: [INSERT] Period and campaign scope: [INSERT] Conversion action, GA4 event, and revenue metric: [INSERT] Source of truth for cross-channel totals: [INSERT, IF AGREED] Currency and decision this comparison should inform: [INSERT] Retrieve platform-attributed outcomes and value alongside the corresponding GA4 outcomes and revenue. State differences in conversion definitions, attribution models and windows, view-through treatment, reporting dates, time zones, and currencies. Check whether the same conversion could be credited by more than one platform. Where transaction or event IDs and a valid matching method are available, identify duplicate records and calculate a deduplicated total. Otherwise, describe the overlap risk without estimating the number of duplicates. Return: Channel or campaign | Platform outcomes | GA4 outcomes | Difference | Platform value | GA4 revenue | Overlap or tracking issue | Decision impact Calculate percentage differences only when the measures are comparable, and state the denominator. Identify where the views agree, where a channel’s contribution changes materially, and where a discrepancy suggests a tracking issue. Finish with the measurement view best suited to the decision and any cross-channel total the data supports. Do not sum platform-attributed conversions or revenue into a supposedly deduplicated total. 56

09 C H A P T E R 9 Act Execute, test, and measure changes Analysis only creates value when it leads to a better decision. But moving from a finding to a live campaign change requires a clear hypothesis, agreed success criteria, and safeguards around execution. These prompts help you turn an insight into a controlled test, prepare an approved change, and monitor what happens next using Supermetrics data. Where campaign actions are supported, require explicit approval before applying, rolling back, or expanding any live change. 57

C H AP T E R 9 — AC T PROMPT 01 Turn a finding into a testable optimization Use Supermetrics to check whether this finding still holds: Finding: [INSERT] Proposed change: [INSERT] Sources and affected scope: [INSERT] Primary business outcome: [INSERT] Constraints: [INSERT] Retrieve the latest complete data. Summarize the evidence, likely business impact, and main uncertainty. Design a test and return: Hypothesis | Control | Treatment | Unit of analysis | Scope | Test design | Primary KPI | Guardrails | Main confounders Change one major variable where practical. Choose the strongest feasible design, such as a platform experiment, matched comparison, or before-and-after test. State what the design can and cannot establish. Finish with the recommended test, why it is worth running, and the main risk. Do not implement it without explicit approval. 58

C H AP T E R 9 — AC T PROMPT 02 Design an incrementality test Use Supermetrics to assess a proposed lift test: Campaigns, channels, and decision to inform: [INSERT] Primary business outcome and measurement source: [INSERT] Proposed test period and budget: [INSERT] Available test methods or platform studies: [INSERT, IF KNOWN] Minimum commercially meaningful lift: [INSERT, IF KNOWN] Conversion lag and known constraints: [INSERT] Retrieve comparable historical spend, reach, outcomes, and conversion volume. Recommend the strongest feasible treatment and control design, such as a platform lift study or geographic holdout. Specify the unit of assignment, what each group will experience, the primary metric, guardrails, and how you will check that the groups are comparable. Assess whether the available volume and test period are likely to support a useful result. Check for audience overlap, other campaigns reaching the control group, seasonality, and changes planned during the test. Confirm eligibility before recommending a platform study; if the data cannot support a reliable calculation of sample requirements, state what is missing. Return: Question | Test design | Treatment and control | Primary outcome | Minimum useful lift | Duration | Main risks | Feasibility Finish with a recommendation: Run the test, Revise the design, or Insufficient evidence to plan it. State what decision the result could support. Do not treat attributed conversions as incremental lift or claim that a simple before-and-after comparison proves causality. 59

C H AP T E R 9 — AC T PROMPT 03 Define success before making the change Use Supermetrics to establish the baseline and success criteria for: Proposed change and test design: [INSERT] Scope, control, and unit of analysis: [INSERT] Baseline period: [INSERT] Primary outcome and expected effect: [INSERT] Minimum commercially meaningful improvement: [INSERT, IF KNOWN] Acceptable downside: [INSERT, IF KNOWN] Planned implementation date and test-duration constraints: [INSERT] Known constraints: [INSERT] Retrieve complete baseline data for the treatment and control or comparison groups. Define the primary KPI, diagnostic metrics, guardrails, conversion-maturity period, required evidence, and earliest reliable evaluation date. Use the supplied commercial thresholds. If none are provided, propose suitable thresholds based on the baseline and business objective. Label them as assumptions that require confirmation before the test begins. Return: Metric | Baseline | Success threshold | Guardrail | Required volume or evidence | Maturity date | Evaluation date | Data source Then define what will count as: Success Promising but immature No meaningful impact Negative impact Inconclusive Account for seasonality, promotions, conversion lag, and changes in budget or audience composition. Calculate sample requirements only when the available data supports a reliable calculation. Otherwise, state what information is missing. If the test design can’t support a causal conclusion, state the strongest conclusion it can support. 60

C H AP T E R 9 — AC T PROMPT 04 Prepare and execute an approved campaign change Prepare this change for review: Change ID and approval owner: [INSERT] Platform and account: [INSERT] Entity names and IDs: [INSERT] Current and proposed values: [INSERT] Implementation time: [INSERT] Rollback condition and action: [INSERT] Execution mode: [CONNECTED TOOL / MANUAL] Use Supermetrics to refresh the supporting evidence. If an action tool is available, use it only to verify the current setting at this stage. Return: Change ID | Account | Entity and ID | Setting | Current value | Proposed value | Evidence | KPI | Rollback condition | Execution method Confirm that the entities and values match the approved recommendation, the campaign is eligible, and no material change has invalidated the plan. Stop and request approval for this exact change. After approval, recheck the current values and apply only the approved settings. Stop if the state has changed materially or the action may already have been submitted. After execution, return: Entity and ID | Setting | Previous value | Applied value | Status | Verification If no supported action tool is available, provide exact manual steps and state that the change has not been made. 61

C H AP T E R 9 — AC T PROMPT 05 Monitor a change after launch Use Supermetrics to monitor: Change ID and implementation details: [INSERT] Affected group and control: [INSERT] Primary KPI, guardrails, and success criteria: [INSERT] Conversion lag, maturity date, and evaluation date: [INSERT] Pre-authorized rollback action: [INSERT, IF ANY] Confirm that the change remains active, then retrieve the latest complete data for the affected and comparison groups. Return: Metric | Baseline or control | Latest result | Change | Threshold | Data maturity | Status Separate: Implementation, delivery, and tracking checks Early directional signals Mature outcome data Guardrail breaches Check for unrelated changes that could invalidate the comparison. Do not declare a winner before the evaluation point unless a guardrail is breached or the change is causing material harm. If a guardrail is breached, state the evidence, business risk, urgency, and whether the rollback condition has been met. Apply a rollback only if that exact action was authorized in advance. Otherwise, request approval. Finish with the current status, urgent issues, data maturity, and next review date. 62

C H AP T E R 9 — AC T PROMPT 06 Measure whether the change worked Use Supermetrics to evaluate: Change and implementation date: [INSERT] Evaluation period: [INSERT] Affected group and control or baseline: [INSERT] Primary KPI, guardrails, and success threshold: [INSERT] Sources: [INSERT] Retrieve mature pre-change, post-change, and control data where available. Compare delivery, primary outcomes, efficiency, and guardrails. When a control exists, compare the change in the affected group with the change in the control, not only the raw before-and-after result. Return: Metric | Baseline or control | Result | Change | Success threshold | Status Where the design allows, calculate uncertainty or statistical significance. Keep statistical reliability separate from commercial importance. Check for unequal exposure, audience or budget changes, seasonality, concurrent campaign changes, tracking issues, and immature conversions. Recommend one decision: Roll out, Roll out cautiously, Keep testing, Revise and retest, Revert, No meaningful impact, or Inconclusive. Explain the evidence, expected business impact, confidence, and next step. State what the test established, what it did not establish, and whether the hypothesis was supported. Do not roll out or revert without explicit approval. 63

10 C H A P T E R 1 0 Communicate Turn performance data into clear decisions Stakeholders rarely need every available metric. They need to know what changed, why it matters, what the evidence supports, and whether they need to make a decision. These prompts use Supermetrics to ground updates, explanations, decision memos, and retrospectives in the latest performance data. Use them to adapt the same evidence for different audiences without oversimplifying the result or presenting an unsupported explanation as fact. 64

C H AP T E R 1 0 — C O M M UN I C AT E PROMPT 01 Answer a stakeholder’s performance question A stakeholder has asked: [INSERT QUESTION] Audience and decision: [INSERT] Sources and scope: [INSERT] Period and benchmark: [INSERT] Currency: [INSERT] Identify the minimum evidence needed, then retrieve it through Supermetrics. Clarify any definition that could materially change the answer. If the question asks why performance changed, use a completed diagnosis or state that further investigation is needed. Respond with: Direct answer: One or two sentences. Evidence: Current result, benchmark, difference, and business relevance. Interpretation: What the evidence supports and what remains uncertain. Decision or next step: What the stakeholder should decide or what information is still needed. Use language suited to the audience and include only evidence that affects the decision. 65

C H AP T E R 1 0 — C O M M UN I C AT E PROMPT 02 Write an executive performance update Create an executive update for: Audience and reporting period: [INSERT] Comparison and business targets: [INSERT] Sources and scope: [INSERT] Decision or approval required: [INSERT, IF ANY] Currency: [INSERT] Use Supermetrics to refresh the latest complete data and reuse relevant monitoring, diagnosis, or pacing analysis. Write approximately 200 words using: Headline: The most important business result. Performance: What changed and its commercial impact. What drove it: Only causes supported by completed analysis. Outlook: Position against budget and outcome targets. Include a forecast only if one has been completed. Decision or next step: What the reader needs to decide or do. Add: KPI | Actual | Target or benchmark | Variance | Status Include only material channels or campaigns. State the latest complete date and any measurement issue that could change the interpretation. 66

C H AP T E R 1 0 — C O M M UN I C AT E PROMPT 03 Explain why a performance target was missed Use Supermetrics to explain: Target and actual result: [INSERT OR RETRIEVE] Period, scope, and audience: [INSERT] Sources and original plan or forecast: [INSERT] Currency: [INSERT] Compare target and actual results, planned and actual spend, forecast assumptions, KPI drivers, and channel or campaign contributions. For additive outcomes such as revenue or conversions, quantify each material driver’s contribution to the gap. For CPA or ROAS, explain how spend, conversions, and conversion value affected the ratio. Return: Driver | Evidence | Contribution or effect | Controllability | Confidence Then write: What happened: State the absolute miss and business effect. Why it happened: Explain the main supported drivers. What we learned: Identify the planning or performance assumption that proved inaccurate. What happens next: State the response and how its effect will be measured. Use neutral language and separate evidence from speculation. Include What was not responsible only when the data directly disproves a plausible explanation. 67

C H AP T E R 1 0 — C O M M UN I C AT E PROMPT 04 Create a budget decision memo Create a memo for: [INSERT BUDGET DECISION] Decision-maker and deadline: [INSERT] Current and proposed allocation: [INSERT] Total budget and period: [INSERT] Primary objective and efficiency guardrails: [INSERT] Constraints, currency, and sources: [INSERT] Use Supermetrics to refresh the relevant allocation analysis. Assess business contribution, efficiency at comparable spend levels, delivery headroom, downstream value, and forecast uncertainty. Write a one-page memo using: Decision required Recommendation Evidence: Current and proposed budget, expected outcome and efficiency ranges, and confidence. Expected impact: Baseline and downside cases. Alternative: The strongest credible alternative. Risks: Assumptions that could weaken the recommendation. Implementation and measurement: Whether to stage the change, plus the KPI, guardrails, evaluation period, and decision rule. Keep only information that affects the decision. Label modeled outcomes as scenarios unless comparable-spend evidence supports a forecast. Do not apply the budget change without explicit approval. 68

C H AP T E R 1 0 — C O M M UN I C AT E PROMPT 05 Create a campaign retrospective Create a retrospective for: Campaign, objective, and audience: [INSERT] Period, planned budget, and expected outcomes: [INSERT] Sources, primary business outcome, and measurement view used for the verdict: [INSERT] Conversion or revenue maturity date: [INSERT] Audience for the retrospective: [INSERT] Use Supermetrics to retrieve mature results from the relevant advertising, analytics, and downstream sources. Keep incompatible attribution views separate. Return: Metric | Target or forecast | Actual | Variance | Interpretation | Data source Then write: Verdict: Did the campaign achieve its primary business objective? What worked and what did not: Identify the strongest and weakest contributors and quantify their impact. What drove the result: Separate supported drivers from hypotheses. What we learned: State transferable lessons and where they apply. What to repeat, change, or test: Identify the highest-value next actions. What remains uncertain: Identify gaps requiring another test or better measurement. Judge the campaign against its primary business outcome. Use CTR, engagement, and other intermediate metrics to explain the result, not replace it. Finish with the three most important implications for the next campaign. 69

From insight toFrom insight to actionaction Good performance work isn’t just about moving faster. It’s about grounding every decision in current data, a clear hypothesis, and evidence you can defend. Used together, the workflows in this playbook give you a repeatable way to move from: Monitor → Diagnose → Pace → Allocate → Optimize →Monitor → Diagnose → Pace → Allocate → Optimize → Plan → Measure → Act → CommunicatePlan → Measure → Act → Communicate With Supermetrics, you can bring your marketing data directly into the AI tool you already use, so you can spend less time moving data around and more time acting on it. Analyze yourAnalyze your marketing datamarketing data with Supermetricswith Supermetrics for Claudefor Claude Get started Analyze yourAnalyze your marketing datamarketing data with Supermetricswith Supermetrics for ChatGPTfor ChatGPT Get started 70

Learn more at supermetrics.com

Flipso