Is your agency actually working? Here is how to tell.
Most founders cannot answer that honestly, because the agency marks its own homework. Here is the founder-side way to check.
Ask a founder whether their agency is doing a good job and you usually get a pause, then a hopeful “I think so?”. The money goes out every month, the report has green arrows on it, and yet they could not tell you, hand on heart, whether any of it is working. I have sat across from a lot of those founders. The uncertainty is not a personal failing. It is built into the way the whole thing is reported.
The platforms grade their own homework
Here is the uncomfortable mechanic. Meta and Google decide which sales to take credit for, and they are generous with themselves. A customer who already knew your brand, searched your name, and would have bought anyway gets counted as a conversion the ad “drove”. Your agency pulls that number into a slide, and it looks wonderful. A platform ROAS of 6 sounds like a licence to print money. The trouble is it measures the platform’s opinion of itself, not your bank balance.
The number that does not lie
There is a simple alternative the platforms cannot inflate: blended performance. Take all your revenue for the month and divide it by all your marketing spend. That is your blended ROAS, and it is the honest one, because it does not care which channel claims the sale. A close cousin, MER (marketing efficiency ratio), is the same idea: total revenue over total marketing. If your platform ROAS is 6 but your blended ROAS is 2, you have just found the gap between the story and the truth.
Then count the fee
Most reviews stop at ad performance and forget the agency is also a cost. So do the whole sum. Say they charge you £2,000 a month, manage £10,000 of spend, and that channel brought in £40,000. The true return, including the fee, is £40,000 divided by £12,000, about 3.3 times. Now compare that to the return you actually need. If your contribution margin is 40%, you break even at 2.5 times (one divided by 0.4). So at 3.3 you are ahead, they are earning their keep, and you should push them to do more. Drop the revenue to £28,000 and the same sum lands at 2.3 times, below break-even, and the polite green arrows are hiding a channel that is costing you money.
The tells that are not numbers
The maths catches the big stuff. The rest is judgement, and a few questions sort the good from the comfortable. Can you log into your own ad accounts and see the raw data, or do you only ever see their slides? When you ask about profit, do they talk margin and contribution, or retreat to ROAS? Are they bringing you tested ideas, or waiting to be told what to do? And read your contract: a long notice period, heavy lock-in, or accounts they own rather than you, are all quiet ways of making you harder to leave. A good agency makes itself easy to check and easy to leave, then earns the stay.
What to do if it is not adding up
Do not fire anyone on a bad afternoon. The move is calmer than that. Ask for blended numbers and your true cost per customer in plain English; a good partner will have them ready. Set a clear 60-day target you both agree on. And if you genuinely cannot tell, get an independent read from someone with no skin in your ad accounts. The point is not to go to war with your agency. It is to stop paying on faith.
That is the whole reason I run founder-side reviews. I am independent, I have twenty years in ecommerce and nothing to sell you on the media side, and I will plug your own data in and tell you the truth: blended numbers, true cost per customer, and whether every pound of spend plus the fee is paying its way. If you have had that nagging feeling, it is usually worth a look. Start with a Growth Audit, or just email info@meandecisions.com.