How to check a marketing agency's results before you believe them

By Doug Dinwiddie, Founder, Maitland

Published 2026-08-20

Every agency shows big percentages. Here is how to check whether the results are real: the four things a case study must contain and the questions to ask.

Executive summary

Agency results are reported by the people being judged on them, so a percentage on its own tells you very little. A genuine case study contains four things: the baseline, the timeframe, the method and the money. Watch for the denominator problem, where the count underneath a percentage is never stated, and the metric swap, where the number that moved furthest replaces the one the client cared about. Run reference calls on what went wrong rather than whether they are happy, weigh independent sources over awards, and apply all of it to our own numbers too.

Every marketing agency you meet will show you big percentages. How do you work out which of them are real, and which are true numbers arranged to flatter?

You are in the right place. This guide gives you the checks, and it applies them to our own numbers at the end, because an agency that teaches scrutiny and then exempts itself is not worth reading.

Here is what it covers: why agency results are unusually hard to judge, the four things a genuine case study must contain, the denominator problem that makes honest people quote wrong figures, how to run a reference call properly, where third-party validation actually lives, and what to make of the numbers on our own case studies page.

Why this is harder than it looks

Marketing results are reported by the people being judged on them. That is the whole problem in one sentence.

No regulator audits an agency case study. Nobody checks the baseline. The agency chooses which client to feature, which period to measure, which metric to lead with and which to leave out, and every one of those choices is legal, normal and enormously flattering. A number can be entirely true and still tell you almost nothing.

"Traffic up 400 per cent" is the classic. Four hundred per cent of what? If the site had 50 visitors a month, it now has 250, which is a rounding error dressed as a triumph. If it had 40,000, that is a serious piece of work. The percentage is identical in both cases and only one of them is worth your money.

So the job is not to distrust every figure. It is to ask the four questions that turn a percentage back into a fact.

The four things a real case study contains

Hold every agency case study, ours included, against these.

The baseline. Where did the number start? Without a starting point, a percentage is unreadable. A serious agency records the baseline before it begins work, because it expects to be judged, and we have written before about why tracking going in before promises come out is the strongest early signal you get.

The timeframe. Over how long? Revenue up 300 per cent in 12 months is a partnership working. The same figure over four years is a business growing, possibly despite its marketing.

The method. What was actually done? A case study that cannot describe the work is usually describing a coincidence. You are looking for a mechanism you can follow, not a list of activities.

The money. What did it return, and what did it cost to get there? Traffic, rankings and engagement are instruments. Revenue and return on spend are the destination. Any case study that never mentions cost is telling you half a story, and it is the half that flatters.

Miss one of the four and ask for it. How readily it arrives tells you as much as the answer does.

The denominator problem

Here is the mistake that catches careful people, and it is worth understanding because you will meet it constantly.

Percentages are governed by what sits underneath them, and the number underneath is almost never stated. An agency reporting a 90 per cent success rate has chosen what counts as an attempt. Were unsuccessful campaigns included? Were clients who left mid-contract counted? Does the figure cover all work, or only the work the agency selected to measure?

Change the denominator and the same honest data produces four different headline figures, three of which are misleading and none of which are lies. This is why any percentage worth trusting arrives with its count attached. "Nineteen from 22" can be checked. "Eighty six per cent" cannot, because you cannot see what the 22 was.

When an agency gives you a percentage, ask what the total was. It is a short question and it is remarkably clarifying.

The metric swap

There is a quieter version of the same trick, and it is the one we see most often.

A case study leads with the metric that moved furthest, rather than the metric the client cared about. Traffic rose 400 per cent, so traffic is the headline. Revenue may have moved barely at all, and it will not appear anywhere on the page, because nothing obliges anybody to publish the number that did not shift.

The check is simple. Ask which metric the client was actually being judged on when the work started, and whether that is the metric on the page. If a case study leads with rankings, ask about enquiries. If it leads with impressions, ask about revenue. A partnership that genuinely worked will have both, and the agency will be glad to give you the second one.

Watch also for the reverse, where an agency quotes a return on ad spend without the spend attached. A 16 times return on £500 and a 16 times return on £50,000 are the same ratio and completely different businesses. The ratio is the boast. The absolute number is the fact.

None of this means the agency is dishonest. It means case studies are written by marketers, who are professionally inclined to lead with their best number. Your job is to ask for the rest, and the shortlisting method in our guide to choosing an agency is built around exactly that habit.

How to run a reference call properly

Ask for two reference clients: one from the agency's showcase, and one chosen from their client list rather than offered to you. The second request tells you a great deal on its own.

On the call, avoid the questions that produce polite noise. "Are you happy with them?" gets you a yes from almost anybody. Ask instead: what did they change in the first 90 days? What has gone wrong, and how did they handle it? Can you see your numbers whenever you want, or do you wait for a report? Has the plan changed since you started, and did they explain why? Would you have signed again, knowing what you know now?

The question about what went wrong is the important one. Every real partnership has a difficult month. A reference who cannot recall a single problem has either had a very short relationship or is being kind. How an agency behaves in the bad month is the thing you are actually trying to buy.

Where third-party validation lives

Agency websites are marketing. Look for evidence the agency does not control.

Google reviews and independent directories carry weight because the agency cannot delete an unflattering one. Client-side LinkedIn posts are useful, since people rarely praise a supplier publicly without meaning it. Long relationships are their own review: an agency that has held a client for five years has passed a test no case study can fake. Awards are the weakest signal, because many are paid entries judged on the submission rather than the result, so treat them as a tiebreaker and never as proof.

One further check costs nothing. Search the agency's own name and read what comes back on the second page.

Length of relationship deserves more weight than it usually gets. Anybody can win a client. Keeping one for five years means surviving at least one bad quarter, one budget review and probably one change of contact on the client side. If an agency's showcase is full of six-month engagements, that is worth a question, because it may mean the work stops being worth the money once the easy wins have been taken.

Now check ours

Applying this to ourselves is only fair, so here are our numbers with the context most case studies leave out.

Across all our clients in the last 12 months, growth averaged 52 per cent. That figure is correct as of July 2025, it covers all clients rather than a selected group, and like any average it hides a spread: some clients sat well above it and some below.

Our headline client results carry the same obligation. ONE Sports Warehouse saw revenue rise 1098 per cent at an 11 times return on investment. The Auckland Project saw revenue rise 1208 per cent at a 14 times return. Kirkgate Dental Surgery saw conversion growth of 429 per cent. Those are real, and they are also the strongest results we have, which is exactly why they are the ones on the website. Every agency does this, us included, and you should read all such pages that way.

What we can add is the part that makes them checkable. The businesses are named, so you can find them and ask. Every one of those partnerships began with a diagnostic month that recorded a baseline before any work started. The results are reported against that baseline rather than against a convenient later date. And each figure names its metric, so a revenue number is revenue and a traffic number is not being passed off as one.

Read them all at our case studies , and hold them to the four checks above. If something does not stand up, we would genuinely rather you asked.

Frequently Asked Questions

How can I tell if a marketing agency's case study is genuine?

Check it contains four things: the baseline the number started from, the timeframe it covers, the method used, and what it returned in money against what it cost. A percentage without a baseline cannot be read at all. If any of the four is missing, ask for it, and note how readily it arrives.

What does "traffic up 400 per cent" actually mean?

On its own, nothing. Four hundred per cent of 50 visitors is 250, which is negligible. Four hundred per cent of 40,000 is substantial. The percentage is identical and only the baseline tells you which one you are looking at. Always ask what the starting number was.

What should I ask an agency's reference client?

Ask what changed in the first 90 days, what has gone wrong and how it was handled, whether they can see their numbers on demand, whether the plan has changed and why, and whether they would sign again. The question about what went wrong matters most, because every real partnership has a difficult month.

Are marketing agency awards worth anything?

They are the weakest signal available. Many are paid entries judged on the quality of the submission rather than the result delivered. Treat them as a tiebreaker between two agencies that have already passed the evidence checks, never as evidence in themselves.

Where can I find independent reviews of a marketing agency?

Look for sources the agency cannot edit: Google reviews, independent directories, and posts written by clients on their own profiles. Long client relationships are strong evidence in themselves. Also search the agency name and read the second page of results, which is where the unmanaged material tends to sit.