What is a growth audit?
The scope, duration and deliverables of an audit that leads to decisions.

A growth audit is a time-boxed diagnosis of the whole path from first visit to renewal, ending in a ranked list of actions and a six-month roadmap. Its first page is a verdict: what is already solid, where the real lever is, and what the rest of the work concentrates on. If you finish reading an audit and still don’t know what to stop doing, you have been given a tour of your own company.
Throughout this chapter, say you run an inbound-led B2B SaaS at 2M ARR, with a marketing and growth team of six.
The scope: look at everything, dig where it matters
The audit crosses the whole AARRR chain quickly, then spends most of its time on one or two stages. Trying to go deep everywhere produces a thick document and no decision.
| Stage | Question to settle | What you collect | Typical output |
|---|---|---|---|
| Acquisition | Which channels bring customers who stay, at what CAC and payback | Ad accounts, Search Console, GA4, CRM sources | Verdict per channel |
| Activation | How many signups reach the moment of value, and what stops the others | Product events, onboarding flows | Friction list, step by step |
| Retention | Which cohorts stay, and why the others leave | Billing data, cancellation reasons, support notes | Cohort curves, causes of churn |
| Monetization | Whether pricing and packaging capture the value delivered | Plans, discounts, expansion history | Packaging recommendations |
| Referral | How many customers arrive through someone else | Referral sources, reviews, community | Loop to build or to drop |
| Foundations | Whether data and organization can carry the plan | Tracking, stack, team, pace of shipping | Quick and advanced fixes |
The discipline is in saying what you left out. On an audit I ran for a B2B SaaS, the scope was inbound only and the analysis centered on conversion, because traffic already had good bases. The document stated what I had not dug into, and I gave no recommendations where I was not the best placed. That honesty is what makes the rest credible.
How long it takes
For a company like our example, count four to six weeks. Here is a realistic six-week calendar.
| Phase | Week | What happens | Output |
|---|---|---|---|
| Framing | 1 | Business goal, questions to settle, access, interview list | One-page scope |
| Data check | 1 to 2 | Tracking review, reconciling sources | What can be trusted, first fixes |
| Analysis | 2 to 4 | Funnel, competitors, site review, interviews | Written findings |
| Prioritization | 5 | Scoring impact, confidence and effort | Ranked action sheet |
| Presentation | 6 | Review with the team, owners named | Six-month roadmap |
Access and tracking are what stretch the calendar. A last-click model flatters brand and hides where conversions start, and missing UTMs on blog CTAs, menus or alerts distort every channel figure. So if your attribution can’t be trusted, the data phase grows, because every later conclusion depends on it. The chapter on data reliability shows how to check it.
What to ask for on day one
A request sent in one go saves a week of back and forth. The framing chapter goes into detail, but the core list is short:
- Read access to GA4, Search Console and each ad account.
- A CRM export with source, stage, amount and close date for each deal.
- A billing export per account per month.
- Product events for signup, activation and key usage. A product analytics tool tells you what users do, and it was never built to tell you where they came from.
- A full crawl of the site.
- Interviews with the founder and the leads of sales, support and product, plus customers recently won and recently lost.
What the deliverable contains
| Section | Content |
|---|---|
| Verdict | One page: what is solid, where the lever is, what the audit focuses on, what it leaves out |
| Data reliability | What can be trusted, with a quick fix and an advanced fix for each problem |
| Market | About three competitors, each picked for a stated reason, a cold read of their sites, a positioning map |
| Funnel | Rates by stage, declared TOFU, MOFU and BOFU labels against reality, the bottleneck |
| Site and resources | Review template by template, and a matrix of resource types scored on awareness, signup and conversion |
| Technical sheet | Each item stated as problem, consequence, action |
| Action sheet | Every action ranked, with owner and effort |
| Roadmap | The actions sequenced over six months |
I open with the verdict because the people who decide will read one page closely and skim the others. Slide or document matters little, as long as the action sheet exists in a format the team can sort and assign.
What a finding looks like
Every finding follows the same chain, and any link missing means it isn’t ready. Here is an illustrative one for our SaaS.
| Field | Example |
|---|---|
| Observation | CTAs in blog posts and links in the menu carry no UTMs and no click tracking |
| Evidence | A test session from the blog to signup comes out as direct traffic in the reports |
| Consequence | The blog looks like a pure cost and brand looks like the main source, and the budget follows |
| Quick fix | Track the blog’s CTAs and the navigation, add an acquisition analytics tool |
| Advanced fix | Multi-touch model, for example 40% first touch, 40% last touch, 20% shared between the middle touches |
| Owner and effort | Head of growth with one developer, two days for the quick fix |
Findings then roll up into the action sheet, scored as impact times confidence divided by effort. The prioritization chapter covers the scoring in depth. The illustrative rows below show the expected format.
| Action | Stage | Impact | Confidence | Effort | Score | Owner | Month |
|---|---|---|---|---|---|---|---|
| Track blog CTAs and navigation | Foundations | 8 | 9 | 2 | 36 | Head of growth | 1 |
| Remove useless templates and thin pages | Acquisition | 4 | 7 | 2 | 14 | SEO lead | 2 |
| Move BOFU content from posts to landing pages | Conversion | 8 | 6 | 5 | 9.6 | Content lead | 2 to 3 |
How to tell a good audit from a bad one
These are the tests I would apply to any audit, mine included:
- The first page, read alone, tells you where the effort goes.
- Every recommendation rests on your data. Strike any benchmark with no source.
- Something is explicitly out of scope.
- The list is ranked and includes things to stop.
- Each action can be assigned to a named person on Monday morning.
Who runs it matters less than those tests. An outsider brings a cold read and no stake in past decisions, while an internal head of growth has the context and needs protected time, plus someone willing to challenge the conclusions.
Your next step: write the scope on one page, with the business goal, the two or three questions the audit must settle, the access required and the people to interview. Then send the day-one request.