Marketing
How to tell whether your agency is doing anything
Written by an agency, about checking an agency. Every document listed is one a client could ask us for tomorrow.
The quickest way to establish whether an agency is working is to ask for five documents that already exist automatically: the search terms report, the change history, Search Console data, the content published, and the links acquired. The Nexclick treats reluctance to share any of them as the answer.
· 6 min read · The Nexclick
In short
If you read nothing else
- Ask for platform-generated documents rather than agency-written reports — they cannot be presented selectively.
- Four of the five are in accounts you should own. Not being able to access them is a bigger finding.
- Judge activity rather than outcomes for the first six months; outcomes lag and activity does not.
- Reporting that only ever contains improvements is a reporting problem, not a performance one.
- Approval delays, slow enquiry handling and unwinnable targets all look exactly like agency failure and are not.
The problem with monthly reports
A monthly report is written by the person being assessed, which is not a criticism so much as a structural fact. It selects which numbers appear, which period they are compared against, and which context is provided. None of that requires dishonesty to produce a misleading impression — choosing a favourable comparison period is enough.
The documents below are different because they are generated by the platforms rather than by the agency. They cannot be presented selectively without the omission being visible, and they show activity rather than outcomes — which matters, because outcomes lag by months and activity does not.
The five documents that cannot be dressed up
| Ask for | Where it comes from | What it proves |
|---|---|---|
| The search terms report, last 90 days | Google Ads, exported unedited | Exactly what your money bought, including waste |
| The change history, last 90 days | Google Ads, automatic | Every edit, who made it, and when |
| Search Console performance, 6 months | Google, not the agency | Impressions and queries, the earliest SEO signal |
| A list of pages published or rewritten | Your own site | Whether content work actually shipped |
| A list of links acquired, with URLs | The agency, but verifiable | Whether link work happened, and its quality |
How to read each one
None of these needs expertise. Each has one thing to look for, and it takes a few minutes.
- Search terms report: sort by cost, descending. Read the top thirty. If you see queries you would obviously never want to pay for, negative keywords are not being maintained — which is the single most common sign of an unmanaged account.
- Change history: look at the frequency and the author. A healthy account shows regular, deliberate changes by a person. Long gaps mean nobody is touching it; a flood of automated entries means Google is managing it rather than the agency.
- Search Console: set the range to six months and look at impressions. Rising impressions with flat clicks means the work is landing and the titles need attention. Flat impressions mean nothing is happening yet, which after six months is a finding.
- Published pages: compare the list against the strategy you were shown. Content that exists but was not in the plan, or a plan with nothing published against it, are both worth asking about.
- Links: open five at random. Check whether the page is real, whether the link is visible on it, and whether anybody would plausibly read that page. Bought links are identifiable within about a minute.
What good looks like, and what stagnation looks like
The distinction is not results — those legitimately take months. It is whether there is evidence of continuous, deliberate work, and the two states look quite different once you know what to look at.
| Signal | Healthy | Stagnant |
|---|---|---|
| Negative keywords | Added most weeks | None added in months |
| Account changes | Regular, by a named person | Nothing, or only auto-applied |
| Reporting content | Includes what did not work | Only ever improvements |
| Comparison periods | Consistent month to month | Changing to whichever flatters |
| Questions asked of you | About customers, close rates, capacity | None since onboarding |
| Response to a challenge | Specifics and evidence | Reassurance and industry context |
The conversation to have before you leave
Changing agency costs three to six months of momentum regardless of who you move to, so it is worth having one direct conversation before deciding. These four questions produce a real answer.
- What has been done in the last quarter, specifically? Not a summary — a list of things that happened.
- What is the single biggest constraint on results right now, and is it something you control? A good agency names something honestly, sometimes on your side.
- What would you do differently if we doubled the budget, and what if we halved it? Both answers are revealing, and the second more than the first.
- What should we stop paying for? An agency willing to name something is worth keeping. One that cannot think of anything has not looked.
When the agency is fine and the problem is elsewhere
This is the outcome nobody expects and it is common enough to deserve its own section. Several of the situations that feel like agency failure are not.
Approval delays are the most frequent. Content sitting in review for three weeks, technical fixes queued behind a product roadmap, or a sign-off process requiring three people turn a twelve-month programme into an eighteen-month one, and the agency cannot fix any of it from outside. If the change history shows work stopping and starting, look at your own side of the relationship before concluding anything.
The second is a downstream leak. Enquiries arriving and being answered slowly, or a close rate that has quietly fallen, produce a picture that looks exactly like poor marketing and is not. The agency can only be judged up to the point where the enquiry arrives.
The third is that the targets were never achievable. If the phrases chosen at the start were not winnable at the budget available, that is a failure at the point of sale rather than in delivery — and it is worth separating, because moving to another agency with the same targets reproduces it exactly.
Questions
Related questions
Is it unreasonable to ask an agency for raw data?
No, and the reaction to the request tells you a great deal. These are exports from accounts you pay for, generated automatically, requiring almost no work to provide. A reluctance framed as "we report on what matters" is managing your impression of the work rather than the work.
How long should we give a new agency before judging them?
Six months before judging outcomes, one month before judging activity. If nothing has shipped in the first month, waiting five more will not help. The distinction between slow results and no work is exactly what the documents in this article establish.
What if we do not have access to our own accounts?
Resolve that first, before any performance discussion. Analytics, Search Console and ad accounts should be owned by the business and access granted to the agency, never the reverse. Rectifying it is usually straightforward, and resistance to doing so is the finding.
Should we tell the agency we are checking?
Yes. Asking openly is a normal part of a working relationship and the response is itself informative. Going around them creates a problem where there may not have been one, and a good agency will be relieved to be asked something specific rather than judged on a feeling.
More articles
Published by The Nexclick. 10 articles so far — we publish when there is something worth saying rather than to a schedule.
Tell us what you are trying to fix
A 20-minute call, no pitch deck. The Nexclick will tell you what we would do, roughly what it costs, and whether we are the right people for it.