Why Your CEO No Longer Trusts Your Metrics, and What Fixes It
%20(11).jpg)
Published:
October 30, 2025
Updated:
November 20, 2026
In October 2026, Forrester reported that 78% of respondents to its Q3 2026 CMO Pulse Survey said executive leadership demands stronger evidence of value than a year earlier. In a Gartner survey released in February 2026, only 32% of executives said their CMO makes compelling business strategy recommendations based on market or customer data. When executives at the US companies in The CMO Survey’s March 2026 edition cut costs, they cut marketing 45.4% of the time, more often than other expense categories.
The dashboard is accurate. It answers the wrong question.
Platform reports measure credit, not cause. By default, Meta credits a purchase up to seven days after a link click or a day after a view (per Jon Loomer’s 2026 guide), and Google Ads counts 30 days after a click. A buyer who clicked a search ad and saw a Meta ad hours before buying counts in both. Neither platform checks the other’s books, so channel totals can exceed the sales finance booked.
In 2014 Bing experiments on branded search, Simonov, Nosko and Rao found a brand’s own ad raised traffic from brand searches by only 1% to 4% when no rivals advertised, but rivals took 18% to 42% of its clicks when they bid on the name and the brand stayed out. When Edmunds paused brand ads in 105 of 210 US markets in 2015, organic links recovered only about half of that traffic. Your dashboard cannot tell which case applies.
In a survey of almost 1,400 marketing and finance executives that Bain ran with Google (an ad seller) and published in July 2026, both groups ranked marketing ROI and revenue impact as their top measures, yet only 41% of marketers felt equipped to tie performance to business outcomes.
Stop sending numbers that cannot change a decision
A sensible test for any executive metric: if it doubled, what would the company do differently? Impressions, followers, engagement rate, open rate and raw lead volume rarely pass. Nor does platform ROAS summed across channels. Keep them on the team’s dashboard, not the CEO’s. Social is especially exposed: it is easy to count and hard to tie to revenue. Every number the CEO learns to skip makes the next one easier to skip.
A one-page report finance will read
A sensible starting point is four lines, each sourced on the page.
- Revenue or qualified pipeline, from the finance system or CRM, never from an ad platform.
- Customer acquisition cost and payback in months, fully loaded with salaries, agency fees and tools.
- Tested contribution: the share of revenue confirmed by a holdout or lift test, shown as a range, next to the share of spend never tested.
- One brand health measure, such as share of search, tracked consistently, so the brand share of your budget has a number of its own.
Printing the untested share looks like an admission, yet the third line earns the most trust: it shows finance you can tell measured results from modeled ones. Then shrink the untested share each quarter, with paid media managed against tested contribution rather than platform ROAS.
How to present a test result to finance
Agree the rules first. Bain found that the strongest marketing-finance partnerships lock the measurement framework before budgets are allocated and say openly what is not easily measurable. With finance, write down the question, holdout design, length and which result moves money which way.
Present it in that order: question, design, result as a range, decision. A made-up example: “Pausing brand search in four regions for six weeks cost less revenue than it saved in spend, even at the top of the range, so we are cutting it where no rival bids on our name.” To read the range, see how to interpret an incrementality result. Finance already demands this of marketing’s AI budget requests, so offer it for media unasked.
If you doubt your numbers would survive that meeting, our work with a client usually starts with an audit of its current marketing, to see what we can do for it.
Frequently Asked Questions
Who should own marketing measurement, marketing or finance?
Marketing should run it, and finance should co-sign the design. Let finance supply the revenue figures, agree which tests count before they run, and see every result, including the null ones. The 2026 Bain and Google survey found companies with strong marketing and finance relationships were 2.5 times more likely to have credible data. Shared ownership of the method matters more than who builds the spreadsheet.
Is ROAS a bad metric to show the CEO?
ROAS is useful inside a platform for comparing campaigns. It misleads at the executive level because each platform calculates it from its own attributed conversions and windows, so one sale can appear in several reports. It also counts sales that would have happened anyway. Show it to a CEO only after an incrementality test has corrected it, and label the corrected figure as tested.
How long does it take to win back a CEO’s trust in marketing numbers?
Nobody can promise a timeline. Trust tends to follow consistency: the same four lines, from the same sources, every quarter, including the quarters that look bad. Share the tests that found nothing alongside the ones that worked. The first test that visibly changes a budget decision often does more than any presentation, because it proves the numbers are used, not displayed.
What if our budget is too small for incrementality tests?
If your account is too small for a platform’s own lift study, a simpler holdout still works: switch a channel off in some regions or weeks and compare against those where it ran. Expect noisier results, wider ranges and longer tests. Where no test is possible, say so in the report and label that spend as untested, rather than presenting modeled credit as proof.



