What a report should contain, the metrics that hide bad news, and the honest test at ninety days.
Most clients discover an agency is not working somewhere around month nine, which is roughly six months later than the evidence was available. The evidence is not hard to read; it is just not in the part of the report anybody looks at.
What a monthly report has to contain
Five things: organic traffic, positions for the terms you agreed to target, leads or enquiries generated, cost per lead, and a specific list of the work done that month. Anything else is supplementary.
The list of work done matters more than it sounds. It is the only line that reveals whether you are buying the hours you think you are buying, and it is the first thing to become vague when an account is being under-serviced.
The metrics that hide bad news
Impressions, reach, engagement rate, follower growth and "brand awareness" all move independently of revenue and can be improved without improving the business. They are not fraudulent, but a report leading on them is a report choosing what to emphasise.
Watch for the switch. An agency that reported leads in months one to four and reports engagement in month five has told you something important about months five onward.
The ninety-day test
At ninety days, revenue is usually too early — but the leading indicators are not. Has the baseline been established? Have the technical fixes shipped? Is content live against verified demand rather than against opinion? Is anything being tested, and has any test concluded?
Ask the question directly: what would make you tell us to stop? An agency with a real answer has a threshold at which it would advise against its own services. An agency without one will keep invoicing regardless of what the numbers do.
What to expect by channel
Paid media should produce signal within weeks — impressions, clicks and cost per acquisition are visible almost immediately, even if the account needs a month or two to stabilise. If a paid campaign has run for six weeks with no read on cost per qualified lead, something is wrong with the measurement rather than the market.
SEO and content take four to twelve months for revenue, but produce leading indicators much earlier: index coverage, impressions for target terms, pages published against verified demand. Conversion work sits between the two, gated by how much traffic there is to test on.
Judging every channel on the same timeline is the most common error in this assessment, and it usually results in cutting the slowest-compounding channel just before it starts to pay.
The conversation to have at month three
Put it in writing, ask for a written answer, and keep both. What did we agree the baseline was? What has moved? What has not moved that you expected to? What would you do differently if you were starting again now?
The fourth question is the informative one. An agency that has been paying attention has a real answer, usually involving something they misjudged. An agency that says it would change nothing after three months has either been extraordinarily lucky or has not been looking.
When flat results are not the agency’s fault
Sometimes the agency is doing good work and the business is the constraint: enquiries not answered for days, a sales process that loses qualified leads, a product priced above what the market will pay, or an approvals queue that has held the site changes since March.
This is worth establishing honestly before switching, because changing agency resets the clock by six months and fixes none of it. A good agency will have told you already — and if they told you and you did not act, that is the finding.
Before you leave
Confirm you own the ad accounts, the analytics property, the domain, the site and the content, and that you can still access all of it after the final invoice. Do this before you give notice, not after.
Then ask for a handover document: what was tried, what worked, what did not, and what the next team should not repeat. Most agencies will provide it if asked at the right moment, and almost nobody asks.
Common questions
- How long should I give a marketing agency before judging results?
- Ninety days for leading indicators — baseline set, technical work shipped, content live, tests running — and six to twelve months for revenue, depending on channel. Paid media should show signal within weeks; SEO and content take four to twelve months. Judging SEO at ninety days on revenue is judging it too early; judging it at ninety days on whether anything has shipped is entirely fair.
- What are the warning signs that a marketing agency is underperforming?
- Reports that shift from leads to engagement metrics, a vague list of work done, no bad news in any month, senior people who disappeared after the pitch, no clear baseline, and no test that has ever concluded. Any one can be explained; three together is a pattern.
- Who owns the Google Ads account and analytics data when you leave an agency?
- You should, and it should say so in the contract. If the agency created the accounts under its own manager account or holds the analytics property, insist on transfer of ownership before you give notice. Losing historical ad and analytics data means the next team starts blind and the cost of switching roughly doubles.