To measure marketing performance, connect a business goal to a specific customer action, choose a small set of defined KPIs, and review the evidence at a pace that fits the decision. Efficiency metrics such as cost per action and reported return on ad spend (ROAS) show what happened under a stated measurement setup; they do not, by themselves, prove that marketing caused the result. Use incrementality experiments when the question is whether a campaign generated outcomes that would not otherwise have occurred.
Start with the decision, not a dashboard
First decide what marketing is meant to do: build awareness, create qualified demand, convert existing demand, retain customers, or contribute to revenue. Then specify the audience and the customer action that would indicate progress. Google recommends a clear KPI structure, an owner for measurement, and measures that cover relevant performance and brand outcomes, rather than a universal list of numbers (Think with Google’s CMO guide).
Choose one primary KPI for the objective and a few supporting indicators that help explain changes. If the desired outcome cannot be measured directly, a proxy KPI can be useful—but label it as a proxy and explain what it does and does not represent. For example, a campaign intended to generate leads might use qualified lead submissions as its outcome, while click-through activity may help diagnose the path to that outcome rather than stand in for business impact.
- Define the action: name the event that counts, such as a lead submission, purchase, download, or in-app purchase.
- Set the scope: state the audience or channel, date range, and relevant cost basis.
- Document the calculation: specify numerator and denominator where applicable, plus the attribution model and reporting window.
- Assign ownership: identify who checks the data and what decision follows if the KPI moves.
Choose metrics that answer the question
Metrics are useful when their definitions fit the decision. The following groups are examples, not a required dashboard; select only those that correspond to the campaign objective.
#1 Best Overall
Business outcomes
Track the result the business values—such as revenue or customer acquisition—and state its time period and scope. A revenue figure without its date range, channel coverage, and associated cost basis may be difficult to compare or act on.
Efficiency
Cost per action (CPA) relates campaign cost to a defined action. For Google Analytics’ All channels performance report, cost per key event is total ad cost divided by selected key events. The “key event” must represent an action that matters to the business, not simply an event that is easy to count (Google Analytics: All channels performance report).
Return
In that same Google Analytics report, ROAS is revenue for selected key events divided by total ad cost. This is a report-level revenue-to-ad-cost ratio, not net profit: it does not itself subtract product, fulfillment, overhead, or other costs. State the report’s definition and scope rather than treating every platform’s ROAS as interchangeable.
Funnel and brand outcomes
Use lead submissions, purchases, downloads, or in-app purchases when those actions align with the goal. If the objective is awareness or consideration, include brand measures and define how they are collected; a performance-only view can miss progress against a branding objective.
Rank #3
Incremental outcomes
Conversion lift and incremental conversions address whether advertising produced additional outcomes beyond what would have happened without exposure. They answer a different question from reported conversions or attributed revenue; do not label attributed results as incremental unless the measurement supports that conclusion (Think with Google’s measurement guidance).
Make sure the measurement setup can support the KPI
Before interpreting campaign performance, verify that data is being collected and that the important customer actions are recorded and classified correctly. Google’s setup guidance covers creating a property, collecting events, configuring conversions, tagging campaign URLs, and importing external or offline data (Google Analytics: Set up Analytics for a website and/or app).
Rank #4
- Check collection: confirm the Analytics property is collecting website or app data for the relevant period.
- Record meaningful actions: ensure the events needed for the KPI are being collected.
- Mark important actions: Google Analytics lets any collected event be marked as a key event when it measures an action important to business success. The Advertising section can then provide pathing and attribution reports for credited touchpoints (Google Analytics: About key events).
- Tag campaign links: use campaign information in URLs so campaign traffic can be identified in reporting.
- Check campaign data access: the All channels report requires a data-collecting property and at least one key event. Campaign data appears when the property is linked to an active-spending Google Ads, Search Ads 360, or Display & Video 360 account, or campaign data is imported (Google Analytics: All channels performance report).
Use the measurement method that fits the question
Attribution, marketing mix modeling (MMM), and experiments are related but not interchangeable. Google’s guidance presents them as complementary methods and notes there is no one-size-fits-all measurement solution; the right mix depends on the decision, channels, available data, and organizational capacity (Think with Google’s modern measurement playbook).
| Method | Question it helps answer | What to keep in mind |
|---|---|---|
| Attribution | Which observed touchpoints received credit along a recorded path to an important action? | Credit depends on the selected model. It is not, by itself, proof that a credited touchpoint caused the action. |
| Marketing mix modeling (MMM) | How should contribution be assessed across a broader channel mix, including traditional and digital activity? | It has its own assumptions and role; use it alongside other methods rather than as a substitute for every question. |
| Incrementality experiment | Did advertising generate conversions that would not have happened otherwise? | Conversion-lift tests are designed to examine additional outcomes, not simply reassign credit among observed touchpoints. |
Google Analytics’ attribution guidance frames questions such as, “What roles did referrals, searches, and ads play in key events?” and “How much time passed between a customer’s initial interest and their purchase?” Use the model, eligible channels, and lookback window that fit the question, and disclose them when comparing results (Google Analytics: Attribution paths).
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errorsBest Value
Interpret reported results with care
- Distinguish credit from cause: attribution allocates credit under a model; an incrementality test is intended to assess additional outcomes.
- Expect product differences: Google says Google Analytics and Google Ads can attribute key events differently, so the same date range may show different values in the two products (Google Analytics: All channels performance report).
- Read cost at the report’s scope: in the All channels report, when a campaign is included after filtering, the cost used for cost per key event and ROAS is the full campaign cost, not a subdivided amount.
- Use a decision-matched review cadence: check data often enough to make the relevant decision, but interpret results in the context of the defined reporting window and attribution settings.
- Do not assume a universal target: the cited guidance provides definitions and examples, not a cross-industry ROAS threshold. A useful target depends on the business outcome, margins, costs, and measurement scope.
Build a recurring test-and-learn practice around the measures: generate hypotheses, test them, and scale tactics that have demonstrated value. The Think with Google guide describes this as “Creating a test-learn-improve cycle by building a learning culture and a continuous practice of generating and testing hypotheses and scaling only tactics that have been proven to work.”
A deeper reference for marketing measurement
For readers who want a broader treatment of marketing measurement, Pearson lists Marketing Metrics: The Manager’s Guide to Measuring Marketing Performance, 4th edition, by Neil Bendle, Paul W. Farris, Phillip Pfeifer, and David Reibstein. Pearson describes it as covering measurement across marketing investments, including brand, social, and omnichannel measurement (Pearson book listing).
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.




