
As of November 2026, the answer to "what did last quarter's spend deliver" has to be one number that marketing and finance both sign: signed-deal value divided by ad spend, traced through CRM stages to signature, and reported every period on the same basis. Platform dashboards cannot carry that weight, because 93% of brands rely on platforms that grade their own homework. If you are hiring a sem agency, ask what the first report will contain before you ask what the monthly fee is, and whether that report would survive a review with the person who controls the budget.
The research behind that number comes from Lifesight's The State of Unified Marketing Measurement Report, built on research by Winterberry Group and discussed in a Demand Gen Report interview with Winterberry's managing partner Jonathan Margulies on 9 November 2026. The findings are less about tools than about where the evidence sits, who owns the metric, and what happens when a dashboard and a finance system disagree.
Your ad platforms want credit. Your CFO wants proof. Those are different assignments, and the same dashboard cannot serve both.
Of the brands Winterberry Group surveyed, 93% rely on platforms that grade their own homework, which means the LinkedIn, Google and programmatic dashboards most teams report from are operated by the companies selling them more media. When finance asks what last quarter's paid social spend produced, a platform-reported ROAS figure answers a question nobody asked: how well did the platform's attribution model credit the platform.
That gap widens in B2B, where deals take months to close and the touchpoints that move pipeline, such as events and sales conversations, sit outside digital tracking. Brands run nearly five different measurement approaches with no unifying layer between them, so the numbers multiply while the clean signals thin out. Every extra metric makes the budget conversation harder to finish.
The deciding document is not the media plan. It is the first reporting period's output, and it can be specified before the contract is signed.
| Item to demand | What it must show | Why finance cares |
|---|---|---|
| One ruler, written down | The ROAS formula in words, agreed with finance before launch | A definition that changes mid-year cannot be audited |
| Funnel stages, quantified | Every stage from impression to signed deal with a count | Locates the leak instead of blaming the channel |
| Spend tied to revenue | Channel spend joined to CRM opportunities and closed deals | Answers what the money produced, not what a platform claimed |
| Mixed CAC each period | Blended cost to acquire one signed customer | The only number a CFO compares across channels |
| Numbered action list with status | Each action from the last period, done or not | Shows whether recommendations turn into work |
The preparation is less heroic than it sounds. 82% of the brands surveyed already use a data warehouse such as Snowflake or BigQuery, so most teams likely already have the central database an independent measurement layer needs. The blocker sits elsewhere: 69% of brands need help translating measurement results into budget decisions, and 74% need help interpreting results, forecasting outcomes, recommending reallocations and pushing approved changes back to ad platforms.
Only 18% of brands say they have all the people and expertise they need to meet their measurement goals. That single figure explains why a reporting standard belongs in the agency brief. You are buying the translation, not a dashboard login.

A ROAS figure is a multiplier: every unit of ad spend is matched by that many units of revenue, as the platform counted it. Double the spend and double the booking, and the multiplier holds at the same number.
That sentence holds only while three things stay fixed: the denominator is media spend alone, the numerator is revenue the platform attributed to itself, and the attribution window is short enough to catch most conversions. Move any one of them and the same headline number means something different.
Take the denominator. If the agency fee, the landing-page build and the CRM seats are excluded, ROAS flatters the programme by design. Those costs are real, they are paid by you, and a CFO will add them back before comparing vendors.
Take the numerator. A platform attributes a conversion when its own pixel or modelled conversion fires, which happens inside a window the platform controls. Signed revenue arrives later, in the CRM, sometimes months later in a consultation-led business. A campaign can post a respectable multiplier while the deals it influenced close in a different quarter under a different attribution model.
Take the window. A window long enough to capture a six-month B2B cycle is long enough to collect credit for deals that would have closed anyway.
The finance-grade version replaces all three variables. ROAS becomes signed-deal value divided by ad spend, recomputed every period on the same ruler, with attribution starting from your own data layer rather than each platform's back end. At that point the multiplier stops being a platform score. It becomes a statement of signed contract value produced per dollar spent, and it can be reconciled against the sales ledger line by line.
This is the definition worth writing into the agreement. A finance reader should be able to take any period and rebuild the figure from the CRM export.
Once the ruler is fixed, the allocation conversation changes shape. Without a shared way to compare channels, an event is judged on meetings, paid search on conversions, ABM on account engagement and content syndication on cost per lead. Those units do not convert into one another, so nobody can say whether moving a dollar from one channel to another will pay off.
Margulies points to structure as the faster fix: agree on a shared outcome such as incremental pipeline or revenue, hold a regular budget review where reallocation is on the agenda, and give one person authority over the cross-channel budget. Budgets split across teams make every move a negotiation between colleagues, which is why allocation stalls even when the evidence exists.
The same logic applies inside paid search. Search, Performance Max, Display and YouTube draw on one budget, and a platform's own conversion reporting will rarely tell you which of them earned the last deal. For a reader comparing providers on google ads singapore terms, the useful question is whether the shortlisted sem agency can rank those campaign types on signed revenue, and on what basis it splits spend between them.
Here is the test to run in the first meeting. Ask the agency to take one past client, strip the name, and show a single period where every campaign type is ranked on signed revenue rather than platform conversions. An agency that answers with a dashboard screenshot has not yet built the join between ad spend and CRM. One that answers with a funnel, a count per stage and a short list of what was changed as a result is speaking the language your CFO uses.

79% of brands already use AI in measurement, and the largest share of that group uses it to generate insights and summarise results; forecasting and scenario planning, and data preparation, are the other two uses the report names. Those uses sit in the safe zone, because every output can be checked before it costs money.
Budget recommendations and strategy execution sit outside it. Margulies' rule is to automate where the evidence is strongest and the cost of an error is lowest, and to keep a person on what gets measured, which metrics matter and what the business does next.
Data quality decides whether any of it works. Poor data quality was cited as the most significant barrier to AI adoption in measurement, at 55%. A model fed by three systems that disagree will forecast confidently and wrongly, and the faster it updates the more quickly it acts on the error.
The report cites an average sales lift of roughly 9.8% for brands using unified measurement. It also cites Seidensticker, which generated 11.8% more revenue while spending less after reallocating budget based on causal evidence.
Margulies warns against anchoring to 9.8%. Realistic gains depend on how effectively current budgets drive revenue, the channel mix and the sales cycle. In B2B, incremental pipeline will likely show impact sooner than revenue, because the deals take longer to close than the reporting cadence.
Set the milestones before the first report, not after. Agree with leadership what success looks like, the timeline to measure it and what the findings will inform. A vendor who promises a specific lift has substituted a forecast for a method.
Topkee builds its paid-search work around the same definition: ROAS measured as signed-deal value divided by ad spend, with attribution starting in the client's own data layer instead of each platform's back end. Topkee Media publishes this in Hong Kong as a service that answers to one number and optimises the whole funnel from impression to signed deal, rather than stopping at the click.
The mechanics are concrete. Topkee's ROAS Truth reporting puts Google Ads, Yahoo Ads, Facebook Ads and GA4 into one report on one ruler, and the Campaign Intelligence Report runs a six-layer funnel with a numbered action list whose status is tracked per item, so a reader can see which recommendations were executed and which were not. TTO CDP holds the data foundation under the client's own name, across an integration catalogue that covers servers, websites, e-commerce, live chat, raw data, CRM and advertising, so the ledger remains auditable and portable.
A sem agency that runs on this footing reports paid search the way finance already reads a sales ledger. Topkee's Google lead-generation plans are quoted as service fees rather than ad budget, and the media spend runs through the client's own ad account. Every plan includes the Weber landing page, TTO CDP, Leadbox CRM and YME live chat at no extra charge. The engagement is staged at 7 days for traffic and funnel foundation, with the deeper conversion work and the ad expansion following on longer cycles; the first stage alone gives finance a checkpoint inside the first month.
One client case worth reading carefully: Mansfield Consulting, a market research and corporate training company, achieved an average ROAS above 400% through Topkee's Google Performance Ads, on a published ad spend of HK$9,848.04. Four hundred percent is a return, not a margin, and the case comes from one industry with one sales process, so treat it as an illustration of the reporting format rather than a benchmark for your category. The method is the transferable part: a named funnel, a fixed formula and a record of what was done about each finding.

Before you sign, ask any shortlisted sem agency for one redacted first-period report and read it against the table above. If the funnel stages are counted, the spend is joined to closed deals and the action list shows status, the vendor can carry a finance review. If the report ends at platform conversions, the buyer is paying for a dashboard.
Topkee Media runs the sem agency service this article describes, and its performance-google-ads page is the place to compare the Google side of that brief; for the data side of the same question, the guide to first-party data marketing growth explains how the customer records behind the ledger get built, and the AI-powered Google Ads conversion article covers where automation fits once the measurement is sound.
It depends on your margin and your sales cycle, and no single figure generalises across categories. Any sem agency quoting one number before it has seen your deal values is guessing. Start from your own baseline: what a signed customer is worth, what share of ad spend you can afford to spend acquiring one, and how long the deal takes to close. Margulies' advice to B2B teams is to benchmark against that baseline rather than the 9.8% average lift cited in the report.
It can, but only as well as the CRM allows. The precondition is clean, connected data: one customer identity across website, sales conversations and CRM, with conversion events synced back to the ad platforms. A data layer such as a CDP supplies that identity. Without it, the report will quietly fall back on platform conversions.
Incremental pipeline usually appears before revenue, because deals take time to close. Agree the milestones in advance: what the first period should show, what the second should confirm, and which decision each result feeds. A quarterly review cycle with reallocation on the agenda matches the cadence of change in most B2B campaigns.
ROAS divides revenue by ad spend and ignores every other cost. ROI subtracts the whole investment, including fees, technology and margin of the product sold, from the return. A multiplier above one can still be a loss once fees and gross margin are applied, which is why finance-grade reporting names its formula and states what each figure excludes.
If the number is going in front of a CFO, yes, or the report should show both versions side by side. Media-only ROAS is useful for optimising bids inside a channel. Fully loaded ROAS is what shows whether the programme earns its place in the budget, and the gap between the two is often the first thing finance asks about.