The short answer
How soon results show depends on how many enquiries the firm already takes. A UK practice recording eight enquiries a month would need roughly sixteen months of them before a twenty five per cent improvement could be told apart from an ordinary run of months. The channel may work long before that. The proof arrives when the volume is there, so set the first evidence review on the volume and review inputs until then.
Key takeaways
- Two clocks run: how long the channel takes to work, and how long your own numbers take to show it. Only the second sits in your own records.
- Enquiries are counted events, so a month moves about on chance alone. At low volume that movement is bigger than the improvement being promised.
- One quarter at four enquiries a month can only display a rise of around eighty per cent. At sixty a month the floor falls to about twenty one.
- Google Ads counts a conversion up to ninety days after the click, depending on a window somebody chose, so a month closes long after it ends.
- Google's own guidance says to read sixteen months of history before concluding a movement is not simply the time of year.
Government guidance on how to report numbers honestly makes an observation that belongs in every agency review meeting. The Government Analysis Function warns that very frequent measures tend to vary a great deal, and that telling random movement apart from real change can be difficult. It binds the producers of official statistics rather than any supplier, and it describes a UK professional services firm reading its enquiry log exactly.
Put a figure on it and the question changes shape. A practice recording eight enquiries a month has twenty four in a quarter, and twenty four can reach thirty without anything having improved. An agency promising a quarter more enquiries inside a quarter is promising something the firm's own records could not confirm or deny. That is not a claim about the agency. It is a claim about the size of the number.
Every published answer measures the wrong thing
Search this question and every result is a range of months published by someone who sells the service. Three to six months for rankings and six to twelve for traffic, with commercial return put further out again. Each answers how long the work takes to have an effect, which is a fair question and not this one. A managing partner asking when they will know is asking when the firm's own evidence could answer.
4 enquiries a month12 in the quarter
12
82%
8 enquiries a month24 in the quarter
24
58%
15 enquiries a month45 in the quarter
45
42%
30 enquiries a month90 in the quarter
90
30%
60 enquiries a month180 in the quarter
180
21%
Work out your own first defensible review date
Two numbers produce it: last year's enquiries and the uplift the proposal implies. Both belong to the firm and neither needs the agency. The result is a month rather than a range.
| Line | Where the number comes from | Figure | Running result |
|---|---|---|---|
| Enquiries recorded over the last twelve months | The practice management system or the enquiry log, not an advertising account | 96 | 8 a month |
| The uplift the proposal implies | The proposal, read as a proportion rather than as a promise | 25% | 0.25 |
| Enquiries needed in the window on each side | 8 divided by 0.25 squared | 128 | 128 |
| Months of enquiries that takes | 128 divided by 8 a month | 16 | Review at month 16 |
A month does not close when the month ends
Google Ads counts a conversion against the click that led to it, inside a window somebody chose. That window is settable from one day to ninety, with thirty days the default for a click through conversion. A figure read on the first of the month therefore keeps rising for weeks, and the review meeting is reading a number that has not finished. Google also publishes conversion lag reporting, which shows how long the firm's own enquirers take, so a practice can read its real delay rather than accept a round number.
The same promise lands very differently at sixty enquiries a month
Volume moves the review date far more than the agency does. Run the worksheet again for a firm recording sixty enquiries a month, seven hundred and twenty a year, against the same twenty five per cent promise. It still needs 128 enquiries in the window on each side, which at sixty a month is a little over two months. The first defensible review sits around month three, against month sixteen at eight a month. Eight times the volume cuts the wait by eight, because the window scales with the monthly count.
Shrinking the promise has the opposite effect, and sharply. Halve it to ten per cent and the formula asks for 800 enquiries in each window, which is a hundred months at eight a month and a little over thirteen at sixty. So the modest, credible proposal is the one the firm's own records are least able to confirm, and a supplier who quotes a small gain with a short review date is describing a movement the enquiry log could not separate from chance.
The scale above can be read as a quick test of any proposal. Find the row nearest the firm's monthly count and compare its percentage with what the agency is promising for a quarter. A promise larger than the row can be judged at the first quarterly meeting. A promise smaller than the row cannot, however carefully the agency reports, and the sensible response is to move the review to the month the worksheet gives rather than to argue about the quarter. Nothing in this is a criticism of the supplier. It is the same arithmetic applied to everyone who sells by the quarter.
What to get in writing before the contract starts
Four lines settle most disputes before they begin. Ask for the review date as a month, the measure the agency will be judged on at that date, and the comparison period it will be set against, with the same length of history on both sides. Then agreewhat counts as an enquiry, since a call, a form and a referral can each be counted or ignored, and a count that changes definition mid-contract cannot be compared with itself. Finally, record what happens if the honest review date falls after the contract ends: either the measure widens to impressions and queries, or the term shortens to a bounded pilot.
Before the next review meeting
Promised a result your own numbers could not display?
Free to your firm. The agency pays us, and only if the relationship works, which is why we would rather ask a supplier what its first review will show than place one that has never been asked.
Get a recommendationRule out the year before crediting anybody
Google tells site owners to read sixteen months of history before deciding a movement is not simply the time of year. Its guidance on debugging a drop in Search traffic puts seasonality alongside technical faults as an ordinary cause, and asks the owner to check a query in Google Trends to see whether the change happened to the site or across the web. A UK accountancy practice already knows this in its own calendar, where January does the work no supplier can claim.
The measurement carries errata of its own
Google keeps a page listing periods when its own figures were wrong. The data anomalies page records known issues across the last three to sixteen months, and among them a logging error running from 13 May 2025 until 27 April 2026 that prevented accurate reporting of impressions, with the click through rate and average position derived from them. Google says the issue affected data logging rather than Search performance, which is the important half: the work may have been sound and the instrument was not. So a figure that will not reproduce is no proof of bad faith, and checking whether the reported numbers are real starts with the configuration rather than the supplier.
When not to set a review date yet
Where nobody at a UK firm records enquiries, there is no clock to start. A dated log of enquiries and what each one became is the whole measurement system, and it takes an afternoon to begin. Appointing a supplier first means paying for a quarter that cannot be assessed.
There is also a case for buying leads instead of waiting, and it is a real one where the firm needs matters in the diary this quarter rather than a durable channel. The arithmetic there is different, because a bought enquiry has a price per client won that can be worked out in advance. It costs more per enquiry and it answers sooner, which is a trade a managing partner can price.
Saying any of this costs us the introduction fee, because we are paid by the agency and only when a relationship works. What we examine before recommending anybody is set out for accountancy practices, recruitment agencies, tax advisory and financial advice firms.
Frequently asked questions
What partners ask about agency timelines.
How long before a marketing agency should show results?
Two clocks run. The channel's clock sets when the work can first affect anything, and the firm's enquiry volume sets when a change becomes visible above ordinary monthly movement. A UK practice recording eight enquiries a month cannot separate a twenty five per cent improvement from an ordinary quarter. Set the first evidence review on the second clock.
Why does our own enquiry volume change the answer?
Because enquiries are counted events, and a count moves about even when nothing has changed. The smaller the monthly number, the larger that movement is as a share of it. Twelve enquiries in a quarter can reach fifteen on chance alone, so fifteen proves nothing. At a hundred and eighty a quarter, a genuine twenty per cent rise is thirty six enquiries, nearly twice the ordinary swing.
What should we review at month one, then?
Inputs and early indicators rather than enquiry counts. What was published, what has been indexed, which queries the site now appears for, and the measured delay between a click and an enquiry. Each is readable within weeks inside the firm's own accounts. None of them proves fee income, and none should be presented as though it did.
Our agency says three to six months. Is that wrong?
Not wrong, and not an answer to this question. A range like that describes how long the work takes to have an effect. It says nothing about whether the practice's own numbers are large enough to display it. Ask which reported figure moves first, and at what month the enquiry line could carry the argument alone.
Can we shorten the wait?
Partly. Widening what counts is the honest lever, because impressions and queries accumulate far faster than enquiries. A bounded paid test is the other, since it buys volume quickly at a price set in advance. What cannot be shortened is the quantity a small enquiry count needs before a modest gain is visible.
Sources and useful reading
- Government Analysis Function, Communicating quality, uncertainty and change, for very frequent measures varying a great deal, and for the difficulty of separating random movement from real change.
- Google Search Console Help, data anomalies in Search Console, for known issues across the last three to sixteen months, and for the impressions logging error of 13 May 2025 to 27 April 2026 affecting data logging only.
- Google Search Central, debug Google Search traffic drops, for seasonality as an ordinary cause of a movement, and for the sixteen month range and the Google Trends check it asks for.
- Google Ads Help, about conversion windows, for a window settable from one to ninety days and the thirty day default on click through conversions.
- Google Ads Help, how long it takes customers to convert, for conversion lag reporting and for a conversion being reported up to ninety days after the click.
This article is commercial decision support and is not legal advice. The statistical guidance cited is addressed to producers of official statistics and is no duty on any supplier or firm. The detection figures are arithmetic on the assumptions printed beside them rather than a measured result. No figure describes any real practice. Each platform behaviour is documented by the platform that operates it and is changed without notice, so read the current page. All sources were located on 7 October 2026. Nothing here describes any named agency or engagement.