How do you check your agency's reported numbers are real?

The advice a UK managing partner will find on this question is to compare the agency's report against the platform it came from. That check does not do what it appears to do. The figure in the report and the figure in the dashboard are one figure read twice, because both are produced by the same configuration, and that configuration belongs to whoever holds the account.

What can be verified is the configuration itself, and both platforms publish what their settings do and keep a dated record of who changed them. Google Ads and Google Analytics each hold two years of change history naming the user responsible for each change. Neither record is produced by the agency, which is why it is worth more than the report. One figure on the page cannot be checked in any account at all, and that is the one the partners actually care about.

Comparing the report with the dashboard is not a check

A screenshot and a live view of the same account agree with each other by construction. Both are readings of one set of definitions. Agreement between them shows the figure was copied accurately. It says nothing about what the figure is, and nothing about whether it means the same thing as the figure beside it from three months ago.

What the headline column was set to count

Google Ads counts either one conversion per ad click or every conversion that follows it, and the choice is made on each conversion action. Google documents both in its conversion counting options, with the default differing by the type of conversion action. A second setting decides whether the action reaches the headline figure at all: Google's documentation on primary and secondary conversion actions puts primary actions in the Conversions column and leaves secondary ones in All conversions, where a partner reading the report never meets them.

A worked example shows the size of it. Take 100 ad clicks, 20 of which lead to an enquiry, with 6 of those 20 visitors enquiring twice, once by phone and once through the form. Counted as one per click, the report says 20. Counted as every conversion, the same month says 26. Neither number is wrong and neither is a lie. They answer different questions, and a report that does not say which one it answers cannot be compared with last quarter or with another supplier.

One month, two counts

The record the agency does not write

Both platforms keep two years of change history, and both name the user behind each change. Google Ads lets a firm review the account history and filter it by date and by user. Google Analytics records the same for an account or property: the date and time, the item, the change and the user, with the Editor role needed to read it.

Change history answers the question a review meeting cannot: whether the definition behind a rising number held still while the number rose. It is the only independent record in the engagement.

Here is why the date matters. Suppose the counting setting is switched from one per click to every conversion in month four. Nothing about the firm's marketing has changed, yet the reported line jumps, and the next three months look like growth. The practice's own enquiry record, which has no setting to switch, keeps its gentle slope. The gap between the two lines is the definition, and change history is where it shows.

What a change of definition looks like

If change history shows a change the report never mentioned, the useful response is a question and not an accusation. Ask what was changed and why, who approved it, whether earlier months can be restated on the same definition, and whether the target the agency is held to was reset at the same time. A good supplier answers all four in writing within a day. A defensive or vague answer tells a partner something too, and it is cheaper to learn it at month three than at renewal.

The evidence standard your own practice already applies

A UK audit does not accept a figure because somebody reported it. ISA (UK) 500, the Financial Reporting Council's standard on audit evidence, measures evidence by its appropriateness, meaning its relevance and its reliability, and among its reliability considerations is that evidence obtained directly is more reliable than evidence obtained indirectly or by inference.

The standard governs auditors conducting an audit. It is no duty on a marketing supplier, and no agency is in breach of it. It is useful here because a practice that would refuse a client's screenshot of a bank balance will accept a screenshot of a conversion count.

The column no platform can give you

No advertising or analytics account knows whether an enquiry became a client. That sits in the practice management system, which makes the outcome column the firm's to supply. A dated export of enquiry dates and what each one became closes the gap without naming a client, and it settles the attribution argument before anybody has it.

It also decides which checks matter, because the two channels a practice usually chooses between produce different evidence: paid search has a conversion column to interrogate, and search visibility has query and position data that no conversion setting touches.

Put the three together and the practice ends up holding a short file: a note of what each headline figure counts, the dated change history, and its own enquiry export. Few suppliers are ever asked for it, which is the reason for asking early. Written down at month three, it turns a vague unease at renewal into a comparison with something to compare against.

When to check less, and when not to hire at all

Running this every month is the wrong frequency.

Done properly once, at around month three, it establishes what each figure counts and who has been changing it. After that it is a renewal exercise, plus a look at change history whenever a number moves sharply.

There is also a case for appointing nobody yet. Where no one at a UK firm holds the accounts, or no one can say what an enquiry is worth to the practice, no amount of checking will settle whether the money worked. That is worth the partner time an engagement needs anyway before a supplier is appointed rather than after.

Saying so 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.