Account audit template for B2B SaaS

The account audit with the checks a subscription business needs: what proportion of leads were accepted by sales, how much of each named list has been reached, whether routing is still intact, and how spend is distributed across the funnel.

Everything else is the general audit, unchanged. One row per finding, evidence attached, honest severity, one action, one owner, one date.

Run it quarterly, and always before a budget is set for the next one. The findings here mostly concern the relationship between advertising and pipeline, and that relationship is invisible in an ad account and awkward to reconstruct once a quarter has closed.

Run it also whenever sales says the leads are poor, because that sentence is usually about a routing or a definition problem rather than a targeting one.

Almost every finding worth having in this category comes from one join: leads by campaign on one side, what sales did with them on the other.

Done once, it usually produces the same result. Acceptance rates vary widely between campaigns, and the correlation with cost per lead is frequently negative — the cheapest campaign attracts the least qualified people. An account that has been optimised on cost per lead has therefore been optimising towards the wrong thing, quietly, for as long as nobody has done this join.

So the first row in this audit is that comparison, with the export attached and the window named. It is the finding that changes the most and it takes an afternoon.

The second check is how much of each named account list has actually been reached.

Once a campaign has covered most of a list, frequency climbs and cost per lead rises, because there is nobody new left to show it to. From inside the platform that looks identical to creative fatigue, and the usual response — new creative — does nothing, because the problem is the size of the audience rather than the quality of the ad.

Measure reach against list size per campaign and record it. Where a list is substantially covered, the recommendation is a larger list or a different segment, and the effort column should say that acquiring one is a project rather than an edit.

Test a lead through the whole route, as a stranger, and follow the record.

This belongs in an audit rather than only in a launch checklist because it breaks between launches. Forms are edited by somebody optimising a page. Owners leave and their records become unassigned. An integration is updated and drops a field. None of those produce an error anybody sees, and the first symptom is a conversion rate that everyone blames on targeting.

Ten minutes, every quarter, with the screenshot as evidence.

Sum spend by funnel stage across the last two quarters.

The common finding is that the account has drifted towards demand capture, because capture always wins a cost-per-lead comparison and every individual reallocation towards it was reasonable. Nobody decided to stop investing in future demand; it happened one sensible decision at a time.

The finding is the distribution, with the trend. The recommendation is usually a floor for upper-funnel spend in the next allocation rather than an immediate change, because reversing the drift in the middle of a quarter is disruptive and the point is to stop it recurring.

A check worth running alongside the lead quality one: confirm that marketing and sales are using the same definition of an accepted lead, and that the definition in use matches the one in the campaign briefs.

Definitions drift. A sales team under pressure tightens what it accepts; a marketing team reports on a threshold agreed two years ago. Both then have accurate numbers that disagree, and the conversation becomes about whose data is right rather than about what to do.

The finding is the discrepancy, in words. The recommendation is a single written definition with a date, agreed by both, and it is usually the highest-value row in the audit even though it involves no spend at all.

In a long-cycle business, the reporting window silently decides which campaigns look good.

Check which windows the reports in circulation actually use, and compare a campaign's performance across a short and a long one. Where the difference is large — and it usually is for upper-funnel campaigns — that gap is the reason two people in the business disagree about the same campaign.

The finding is the comparison. The recommendation is a single agreed reporting view, named, that everybody uses for budget conversations.

The dull check that eventually matters: who can administer each ad account, page, property and integration, and whether any of them depends on one person or on a supplier who is no longer engaged.

In this category the integrations are the risk, because the path from an ad to a lead record usually passes through a tool somebody set up years ago. Record what it is, who owns it, and who can change it.

The last check is the one that sounds like housekeeping and is not: how many live campaigns follow the naming convention, and how many predate it.

In this category the naming carries the segment and the funnel stage, which are the two dimensions every useful report groups by. A quarter of the account named inconsistently means a quarter of the spend cannot be attributed to a segment without somebody reading campaign names by hand, and that hand-reading is the reason the reports everybody wants are produced twice and then abandoned.

The finding is the proportion, with the list. The recommendation is almost never to rename live campaigns, because that breaks history; it is that every new campaign follows the convention and the old ones are retired on their own schedule.

Seven checks. Every finding separates what was found from what should be done, with evidence and a window. Lead acceptance has been compared across campaigns. List depletion has been measured. The lead route has been tested end to end with a real submission. Spend has been summed by funnel stage and the trend recorded. The definition of an accepted lead has been confirmed with sales in writing. And every row has an owner and a date agreed with the person named.