A founder told me recently that they were spending five figures a month on ads and could not say which customer segment was converting. Not "would rather not say". Could not. The data had never been collected in a shape that could answer the question.
That is not a marketing problem. That is a deploy to production with no logging, and every engineer reading this just winced.
The audit is your observability layer
You would not ship a service with no instrumentation and then debug it by staring at the traffic graph. But that is precisely what buying ads before auditing your funnel amounts to. You can see spend going out. You can see some signups coming in. The entire causal chain in between is opaque.
A marketing audit is the instrumentation pass. It is unglamorous, it produces no campaigns, and skipping it is the most expensive decision an early-stage business-to-business (B2B) software company makes.
What skipping it costs
One published audit of 43 enterprise B2B software Google Ads accounts, covering $31.3 million US of spend during 2024, put average waste at 36.1 percent (sourced from GrowthSpree). The largest causes were structural rather than creative: broad match keywords with no negative keyword discipline, Performance Max campaigns running without offline conversion data, and default attribution windows that missed most of the revenue.
Two caveats worth stating plainly. Those were enterprise accounts, so the dollar figures do not transfer to a seed-stage company. And the study comes from an agency publishing research about its own service area, which is worth weighing. The failure modes still transfer, and that is the point. Every one of those causes is a measurement problem an audit catches in an afternoon and a campaign manager never catches at all, because from inside the campaign it looks like it is working.
For a sense of what is at stake, one 2026 benchmark set puts marketing spend at 15 to 25 percent of annual recurring revenue for pre-product-market-fit companies under $2M, and 12 to 18 percent for Series A companies between $2M and $10M (sourced from GrowthSpree). At those proportions, a third of the budget going nowhere is not a rounding error. It is a runway question.
What the audit actually covers
Five things, roughly in this order.
Ideal customer profile, stated as a filter and not a mood
Most early decks describe a customer type. An audit asks a harder question: of your last twenty closed deals, which renewed, which expanded, and what did those accounts have in common that the churned ones did not? That answer is your ideal customer profile. It is frequently narrower than the founding team believes, and it is the input to every targeting decision downstream.Positioning against the actual alternative
Not against your named competitor. Against what the buyer does if they do nothing, which is usually a spreadsheet and a person. If your messaging beats a competitor but loses to a spreadsheet, your funnel leaks at the top and no channel fixes that.Instrumentation
Is a signup traceable back to a campaign? Does a closed deal write back to the ad platform? Is your conversion event the form fill, or the sales-qualified lead? Optimise a bidding algorithm toward form fills when half your form fills are junk, and you have taught it to find more junk. Efficiently.The conversion surfaces
Trial signup page, pricing page, documentation. In software, the pricing page is frequently the highest-intent page on the site and the least maintained one. An audit reads those pages as a buyer would, not as the person who wrote them.Channel fit against sales cycle
A four-month enterprise cycle and a self-serve trial want completely different channels. Founders often inherit a channel mix from whoever they last hired rather than from how the product is actually bought.
What a week of this gets you
A short audit should end with three things: the segments actually worth targeting, the measurement gaps that have to close before spend means anything, and a ranked list of fixes with the cheapest and highest-impact first.
It should not end with a campaign proposal. If the deliverable is a proposal, you were sold a sales call.
If you want a cheaper starting point, the free website audit walkthrough covers the site-side half of this at no cost, though it is written for local businesses rather than software companies, so keep that in mind.
One compliance note, since lifecycle email always comes up
If you operate in Canada or email Canadian recipients, Canada's Anti-Spam Legislation applies to your onboarding sequences, your activation nudges and your reactivation campaigns alike. Every commercial electronic message needs express or implied consent, clear sender identification, and a working unsubscribe mechanism. Transactional product notifications and marketing messages are not the same category, and the distinction is much easier to get right in your platform before you have ten thousand contacts than after. This is general marketing guidance rather than legal advice, so run anything consequential past counsel.
The question worth answering first
Before your next campaign, answer this: if it doubles your signups, will you be able to tell which segment they came from, and whether they renewed?
If the answer is no, that is what to fix. Not the campaign.