ChatGPT
Claude
Gemini
Grok
Perplexity
There are five things a monthly ad report leaves out, and each one hides whether your ads made money. The report is only as honest as the conversion tracking underneath it, and most founders have never opened that screen. This guide gives you the five red flags, what each one hides, and the exact thing to ask for instead.
ROAS means return on ad spend: the revenue an ad platform credits to itself, divided by what you spent there. Almost every agency report leads with it. It is a real number and a useful one, and it is not the same thing as profit.
The Five Red Flags, in One Table
Screenshot this and hold your last report against it.
| # | The red flag | What you see | What it hides | What to ask for |
|---|---|---|---|---|
| 1 | ROAS leads, profit is missing | "ROAS 4x, trending great" at the top | The real net after product, shipping, fees, tools, and the agency fee | The net after all costs, including the retainer |
| 2 | No new versus repeat split | One combined revenue number | Whether the ads won new customers or re-sold to repeat buyers | New-customer cost against your break-even |
| 3 | Never matched to your store | The platform's revenue treated as the truth | That nobody compared it to your real store purchases, so double counting goes unnoticed | Platform numbers next to your store's, same dates |
| 4 | Soft metrics up front | Impressions, reach, clicks, click-through rate leading | The profit questions, which go unanswered | The money numbers first, the reach numbers last |
| 5 | The timeframe moves | "Month over month," "year over year," or one hand-picked good month | A weaker full picture, with the window shifted to whichever view supports the story | The same window every month, judged at your profit and loss |
Red flag 3 is the one that decides the others. If nobody ever matched the platform's number against your store's own sales, none of the other four can be answered honestly.
Red Flag 1: ROAS Leads and the Profit Number Is Missing
A reported 4x can be losing money. The platform's ROAS counts the revenue that platform believes it caused, divided by what you spent there. It does not subtract your product cost, your shipping, your returns, your transaction fees, your tools, or the retainer you pay the agency. Put all of that in and the picture can invert.
The problem is when it arrives alone. Ask for the net after every cost, including the retainer, for the same period the report covers. If your agency does not produce that, the break-even and contribution-margin math is worked in full here and you can build it yourself in an afternoon.
Red Flag 2: One Revenue Number, With No Split Between New and Repeat Customers
A single revenue figure cannot tell you whether the ads brought you customers or took credit for sales that already happened through another channel. If most of that revenue came from people who were already buying from you, the ads are not doing the work the number implies.
Ask for your cost to acquire a new customer, and ask what it is measured against. The honest version puts all of your ad spend against new customers only, then compares the result to your break-even. The full method for judging ads on new-customer profit is here.
Red Flag 3: The Platform's Number Was Never Matched to Your Store
This is the one I can prove, so here is the account.
I took over the Google Ads for a pond and water garden supply store. The prior agency ran it for $1,500 a month flat, and its dashboards showed spring at roughly 4x to 5x. The account had two conversion trackers firing at once: the GA4 purchase event, and an old duplicate AdWords conversion tag left in place. The duplicate double counted.
Google treats this as a real hazard, which is why conversions built from Analytics events are set to secondary actions, to prevent counting the same event twice. Nothing was guarding this account.
Here is what that did to a single month.
| Window | Source | The number |
|---|---|---|
| May 2026 | Campaign-reported | $34,817 |
| GA4 purchase event (the real purchases) | $58,286 | |
| Duplicate tracker | about $73,000 | |
| Aug 1 to 10, 2026 | Duplicate tracker | $13,275 |
| GA4 purchase event (the real purchases) | $6,422 |
Three numbers, three answers, for the same business. Note what the May column actually shows, because it is not the simple story: the duplicate tracker ran high, and the campaign-reported figure ran low. Neither matched reality. That is the point. Once two trackers are live and nobody has checked them against the store, no single number in the report can be trusted, in either direction.
It got worse after a restructure. The prior agency rebuilt the account on June 9, which broke tracking badly enough that June reported $12,829 against $31,862 in real purchases. Blended ROAS for the month came out around 1.76x. That was a losing month. In the report it looked like a dip, and nobody flagged it.
The owner's memory did not survive his own data either. He remembered "10 to 15% cost to revenue" under his previous setup. His best month all year was January, at 4.92x, which is 20.3% cost to revenue, and even that figure sat on the inflated tracking. His real history was worse than he believed it was.
Ask for the platform's revenue next to your store's own sales, for the same dates. If they are far apart, the tracking is the reason, and the walk-through for fixing it is here.
Red Flag 4: Soft Metrics Lead and the Money Questions Go Unanswered
Impressions, reach, clicks, and click-through rate are real metrics. They should not define whether a campaign worked, unless that is genuinely what you hired the agency to do.
When those metrics are the goal
Usually a large company with a big budget running awareness, where reach is the product being bought. If that describes your business, this red flag is not one.
When they are a substitute
An ecommerce brand selling products at a margin. Here the reach numbers cannot answer the profit question, so leading with them changes the subject.
Ask for the money numbers first and the reach numbers last. The order of a report tells you what it is optimized to make you feel.
Red Flag 5: The Timeframe Moves to Fit the Story
Watch for "trending in the right direction." Watch for "month over month," "year over year," and "week over week" when the window shifts between reports. Any single one of those views can be the honest one. What is not honest is picking whichever one looks best this month.
Ask for the same window every month, and judge it at your profit and loss. If you want the deeper version of which number to trust when the platform and the business disagree, that comparison is worked here.
Founders Arrive Already Distrusting the Number
The report is usually not what brought someone to the call. The distrust got there first.
A handmade home goods brand told me agency fees were stacking on top of rising costs, and she needed better profitability reporting just to know where she stood. Her TikTok ads had already been cut, after losses the reporting had not made obvious. A natural skin care brand said inventory and listing problems under prior management had left her distrustful of what she had been told. A drinkware brand's owner put it most plainly: he wanted profitable scaling, not a return to his prior manager's numbers.
Three different businesses, one shared position: they no longer trusted the number they were being shown. That is not proof of how common this is, and I am not going to pretend I have a figure for it. It is the reason the five checks above are worth running yourself, rather than waiting for a report to volunteer the answer.
Why This Happens, and Why It Usually Is Not Lying
Agencies report the number they are paid to manage. They are hired to run a channel, so they operate inside that channel, and they are measured on its metric. Reporting platform ROAS is not a trick, it is the job as it was scoped.
The rest is quieter than dishonesty. Many agency staff inherited a clean account or lean on tools, and never had to set tracking up from scratch. So they may not notice something is broken, because nobody ever went looking. That is what happened on the account above: a duplicate tag sat there for months while everyone read the dashboard it was feeding.
Which is also why the honest version cuts both ways. Sometimes you run the net and find out the work was paying after all. What you are looking for is not a villain, it is a number nobody produced.
What to Ask For, and How to Check It Yourself
You do not need permission to check any of this.
- 01
Open your conversion actions and count what is firing.
In Google Ads that is the conversion actions list, in Meta it is Events Manager. If two things are counting purchases, you found your problem before you read another report.
- 02
Pull your store's own purchases for the same dates.
Your ecommerce platform is the source you can reconcile against your bank. Put the two numbers side by side.
- 03
Rebuild on the one measurement nobody has touched.
On that account it was the GA4 purchase event, which had been sitting there accurate and ignored all year. Pick your equivalent and report against it from now on.
- 04
Do the net once, by hand.
Take a period's sales and subtract every cost, including the retainer, and write down what is left. That is the number the report should be built around.
- 05
Ask for the same window, every month.
Consistency is what makes a report readable over time.
Want to pressure-test your last report right now? Paste this into any AI tool:
AI Prompt
Copy this and paste it into ChatGPT or Claude. It will walk you through your own report.
I get a monthly ad report from my agency. Here is what it says: reported revenue [X], ad spend [X], reported ROAS [X]. My own store shows [X] in sales for the same dates. My average order value is [X] and my cost of delivery per order is [X] (product, shipping, returns, fees). I pay a [X] monthly retainer. Walk me through, one step at a time: the gap between the reported and real revenue, my real profit after every cost including the retainer, and the three questions I should send back to my agency.
If the whole account needs checking rather than just the report, the five-check paid media audit is the full run-through. And if you conclude the relationship itself is the issue, the agency versus in-house math is worked out here.
Common Questions
Agency Ad Reporting FAQ
How do I know if my agency's ad report is accurate?
Start with the one check that settles most arguments: open your conversion actions and count what is firing. In Google Ads that is the conversion actions list, in Meta it is Events Manager, and if two things are counting purchases you have found your problem. Then pull your store's own purchases for the same dates and put the two numbers side by side. If they are far apart, the tracking is the reason, not the ads.
What should a good agency ad report include?
Five things, and most reports have one or two of them. The net after every cost, including the retainer, not just platform ROAS. Your cost to acquire a new customer, measured against your break-even. The platform's revenue next to your store's own sales for the same dates. And the same reporting window every month, so the numbers stay comparable.
Why does my agency's revenue number not match my store?
Usually because more than one thing is counting purchases. On one account I took over, a GA4 purchase event and an old duplicate conversion tag were both live, and the duplicate double counted. It can also run the other way: a campaign restructure can break tracking so the platform reports less revenue than actually happened. Either way the platform number is not wrong on purpose, it is unverified. Your store's own purchases are the figure you can reconcile against your bank.
Is it a red flag if my agency only reports ROAS?
It is worth a question, not an accusation. Agencies report platform ROAS because that is the channel they were hired to manage and the metric they are measured on, so leading with it is the job as it was scoped. The problem is when it arrives alone, because a reported 4x can be losing money once product, shipping, fees, tools, and the retainer are in. Ask for the net after all costs for the same period. If the answer is that nobody has built that number, that is the real finding.
The Bottom Line on Agency Ad Reporting
Five red flags: ROAS leading with no profit number, no split between new and repeat customers, a platform figure never matched to your store, soft metrics up front, and a timeframe that moves. The report is only as honest as the conversion tracking underneath it. Open your conversion actions, count what is firing, and put the platform's revenue next to your own sales for the same dates. That one check settles most arguments about a report.
