How to Measure a Performance Marketing Agency: Incrementality, Attribution and the Numbers That Matter
How to Measure a Performance Marketing Agency: Incrementality, Attribution and the Numbers That Matter
To measure a performance marketing agency, judge it on incremental business outcomes against a baseline agreed before the work starts, not on platform-reported ROAS. The core numbers are customer acquisition cost, contribution margin and payback period by market, checked with incrementality tests such as geo lift, conversion lift or holdouts. Marketing mix modeling adds a cross-channel view once spend runs across several channels.
Most agency reports lead with return on ad spend as Google Ads or Meta Ads Manager shows it. That figure counts conversions the platform can link to an ad, including many that would have happened anyway. It measures credit rather than cause, and a brand deciding whether to renew an agency needs numbers that separate the two.
The methods for doing this are now widely available, and two of the most capable are free. Google and Meta each publish an open-source marketing mix model, and both platforms offer lift studies that compare exposed and unexposed audiences. The harder work is agreeing in the contract which numbers count, how often they’re reviewed and who owns the data.
Why Does Platform-Reported ROAS Overstate an Agency’s Impact?
Platform-reported ROAS overstates agency impact because each ad platform credits itself for conversions it touched, not conversions it caused. Branded search, retargeting and returning customers inflate the figure most, since those buyers were already close to purchasing.
Each platform also reports in its own silo. A customer who sees a Meta ad, then searches the brand on Google and buys, can be counted by both platforms. Added together, platform-reported conversions can exceed the orders recorded in the store or CRM.
Analytics tools reduce the double counting but don’t solve the causal problem. According to Google Analytics Help (2026), GA4’s data-driven attribution uses a counterfactual approach to estimate which touchpoints most likely drove a key event. It still only sees the journeys it can track.
An agency that optimizes only to platform ROAS has an incentive to move budget toward branded search and retargeting, where reported returns look highest. That isn’t necessarily bad faith. It does produce reports that improve each month while new-customer growth stays flat.
What Is the Difference Between Attribution and Incrementality?
Attribution assigns credit for a conversion across the touchpoints a customer passed through. Incrementality measures how many conversions wouldn’t have happened without the ads. Measuring it requires an exposed group and a control group that didn’t see the ads.
Attribution is useful for daily optimization, such as deciding which keyword, ad set or creative is pulling its weight inside a channel. It can’t say whether the channel as a whole is worth its budget, because it has no record of what unexposed customers did.
According to Google Ads Help (2026), Conversion Lift compares conversions between a group that sees the ads and a group that doesn’t. Tests can be built on users or on geography, and the geography version supports offline sales data. Google grants access through account representatives, and Meta offers its own Conversion Lift product.
Which Measurement Methods Should a Brand Use, and When?
A brand should combine four measurement methods, each answering a different question. Platform attribution guides in-channel optimization, GA4 covers cross-channel paths, incrementality tests prove cause and marketing mix modeling sets budgets. Few brands need all four in their first year with an agency.
| Method | What it answers | Cost and effort | When to use it |
|---|---|---|---|
| Platform attribution (Google Ads, Meta Ads Manager) | Which campaigns, ads and keywords the platform credits with conversions | Free and built in | Daily and weekly optimization inside one channel |
| GA4 data-driven attribution | How credit splits across channels on tracked user paths | Free, but needs clean tagging and well-defined key events | Monthly cross-channel reviews |
| Incrementality tests (geo lift, conversion lift, holdouts) | How many conversions the ads actually caused | Low tool cost; the real cost is media held back from the control group and several weeks of runtime | Before scaling a channel, and periodically for each major channel |
| Marketing mix modeling (Google Meridian, Meta Robyn) | How each channel contributes to sales over time, including offline and brand effects | Open-source software; needs analyst time and a long history of weekly spend and sales data | Quarterly or annual budget planning once spend runs across several channels |
How Do Geo Lift Tests and Holdouts Work?
A geo lift test changes advertising in a set of regions and compares sales there with matched regions where nothing changed. A holdout does the same with part of the audience, withholding ads from a randomly chosen group.
Geo tests suit brands that sell offline or through marketplaces, where user-level tracking is weak. In a compact market such as Singapore, the small number of distinct regions makes clean geo splits harder. Audience-level conversion lift or a structured on-and-off test is often more practical there.
When Is Marketing Mix Modeling Worth It?
Marketing mix modeling is worth it when a brand spends across several channels and wants to set budgets by channel. It works on aggregate data, so it isn’t affected by cookie loss or consent rates.
According to Google for Developers (2026), Meridian is Google’s open-source MMM, and its GeoX tool turns geo experiment results into model priors. According to Meta Marketing Science (2024), Robyn is an open-source MMM that uses ridge regression and calibrates against geo tests and lift studies.
Both tools still need someone who can judge whether the output is plausible. An agency’s MMM work should come with the model inputs and assumptions, not only the charts.
What KPIs Should a Performance Marketing Agency Report?
A performance marketing agency should report KPIs in three layers: business outcomes, channel efficiency and operational quality. Business outcomes show whether the engagement is working, and the other two layers explain why.
| Layer | KPIs | What it tells the client |
|---|---|---|
| Business outcome | Revenue, new-customer revenue, blended CAC, contribution margin after ad spend, LTV:CAC ratio, payback period | Whether marketing is growing the business profitably |
| Channel efficiency | CPA, ROAS, CPC, conversion rate, incremental CPA from lift tests, split of spend between new and returning customers | Where budget works hardest, and where it buys conversions that would have happened anyway |
| Operational | Tests launched per month, share of tests that produce a usable result, reporting accuracy against backend data, time from finding to action, transparency of media spend and fees | Whether the agency is improving the account or only maintaining it |
Blended CAC is total marketing spend divided by new customers in the period, with customer counts taken from the store or CRM. It’s harder to flatter than ROAS because it doesn’t depend on any platform’s attribution.
Contribution margin matters because revenue can rise while profit falls. An agency can hit a revenue target by leaning on discounts or low-margin orders, and only a margin-based KPI shows that trade. Our article on agency reporting and campaign performance measurement covers the dashboard side of this.
Judging agencies on business results also matches how the industry’s effectiveness bodies frame the question. According to the IPA (2026), its Effectiveness Awards Search Engine holds 1,500 case studies entered since 1980. Each one is built around showing that communications activity paid back.
How Should Measurement Be Written Into an Agency Contract?
Measurement belongs in the contract as four items: a baseline, a measurement plan, a review cadence and data ownership. Without them, client and agency spend review meetings debating numbers instead of making decisions.
- Baseline: Record CAC, new-customer volume, revenue and conversion rates for the period before the agency starts, using the same definitions the agency will report against.
- Measurement plan: Agree which conversions count, which attribution model governs reporting and which incrementality tests will run in the first two quarters.
- Review cadence: Set weekly channel check-ins, monthly business reviews against the KPI scorecard and a quarterly review of test results and budget allocation.
- Ownership: Register ad accounts, analytics properties, pixels and data exports to the client, with the agency added as a user. This keeps the account history intact if the relationship ends.
Fee transparency belongs in the same section of the contract. Media spend should be shown separately from management fees, and any platform rebates or markups should be disclosed. Typical fee structures at smaller budgets are set out in our guide to PPC agency pricing.
How Do You Compare Agency Performance Across Markets?
Agency performance across markets should be compared on CAC and payback period in a common currency, not on raw ROAS. ROAS varies with local prices, order values and competition, so a market with lower ROAS can still be the more profitable one.
A brand selling in Singapore, Japan and the US will see different click costs and basket sizes in each. Spend and revenue should be converted to one reporting currency at an agreed monthly rate, then CAC compared with each market’s gross margin. Fixing the rate method upfront stops exchange swings from distorting results.
Consent rules change what tracking can see. According to Google Ads Help (2026), basic consent mode sends no data before a user consents, while advanced mode sends cookieless pings and supports more detailed conversion modeling. Markets with different consent setups report different shares of modeled conversions, so geo tests and MMM are a useful check.
At Creative For More, we agree the reporting currency and conversion definitions before a cross-border campaign launches. Our offices in Singapore, Tokyo and New York then report every market on the same basis. The selection side of this is covered in how to choose a performance marketing agency for cross-border expansion.
What Should a Brand Ask an Agency About Measurement Before Signing?
A brand should ask how the agency defines success, how it proves incrementality and who owns the data. The answers show whether the agency is willing to be judged on outcomes rather than on platform dashboards.
- Which numbers will the agency be judged on, and which system do they come from?
- Will an incrementality test run in the first six months, and on which channel?
- How are branded search and retargeting treated in reporting?
- Who owns the ad accounts, analytics property and pixels?
- How are media spend, fees and any rebates shown on invoices?
An agency that answers these clearly is usually easier to evaluate a year later. Brands comparing outside teams can review the scope of our media buying service against the same questions.
If you want cross-market reporting tied to CAC and incrementality rather than platform ROAS, start with the measurement plan. If your company is looking for a strategic marketing partner, the Creative For More team can help. Book a discovery call to explore how we can support your brand’s growth.
Read More
Geraldine Pang
Creative For More
Geraldine Pang is the founder of Creative For More, a cross-border digital marketing agency with offices in Singapore, Tokyo, and New York. She specializes in brand strategy, market entry, and cross-border marketing across Southeast Asia, Japan, and the United States.