Frame the audit
The business goal, the questions to settle, the access you need and who to talk to.

Frame the audit around the decision the company has to make, and write that decision down before anyone opens a dashboard. The data you need and the people you interview both follow from it. Skip this step and you get a thorough tour of the business with recommendations nobody has to act on.
The output of this chapter is a one-page framing note, approved by whoever owns the decision. As an illustration, say a B2B SaaS at 2M ARR, with a free trial, a small sales team for larger accounts, and net new ARR flat for three quarters. The founder’s request is “find out why growth has stalled”. You can’t answer that as written.
Turn the request into a goal with a number and a date
People arrive with a symptom or a favorite solution. Your first job is to trace it back to a business goal, then to state what the audit has to establish about that goal.
| What you hear | What usually sits behind it | Framed goal (illustrative) |
|---|---|---|
| “Growth has stalled” | Net new ARR is flat | Go from 2M to 3M ARR in 12 months and identify the constraint that blocks it |
| “We need more leads” | Sales has capacity left, or believes it does | Establish whether pipeline volume or pipeline quality limits new ARR |
| “Our SEO is underperforming” | Someone compared traffic with a competitor | Decide how much of next year’s new ARR organic can carry |
| “Churn feels high” | A few painful cancellations | Measure retention by cohort and segment, then decide whether it caps growth |
Check that the number is one leadership already tracks and that a named person will decide on the result. If the founder and the head of sales give you two different goals, stop and get them in a room. That disagreement is the first finding of your audit.
Write the questions the audit has to settle
A good audit question has a decision attached. If no possible answer would change what the company does next, cut the question. I keep the list to about six, because each extra question thins the analysis of the others.
| Question | Decision it drives | Evidence needed |
|---|---|---|
| Which stage of the funnel limits new ARR? | Where the next two quarters of effort go | Stage rates by segment and source |
| Who buys, who stays and who expands? | Which segment to target, which to stop serving | CRM and billing joined at account level |
| Which channels pay back, and how fast? | How the budget is split | CAC and payback per channel |
| Can the team carry out the plan? | Hire, outsource or reduce scope | Interviews, shipping history |
| Is the data good enough to answer the above? | Whether tracking work comes first | Review of tracking and attribution |
The last row belongs in every audit, and the data reliability chapter covers how to answer it. The first one needs the business written as a formula, which is the job of the growth equation.
Give the audit a provisional verdict
Spend the first half day on a quick pass: the revenue curve, rough funnel rates, signups by source, and some time on the site as a stranger. Then write three sentences. What is already solid, where the real lever seems to be, and therefore what the audit will focus on.
On an audit I ran for a B2B SaaS, that first pass showed traffic with good foundations and a conversion side nobody had really worked on. So the analysis went to conversion: what each page was for, whether the funnel labels matched reality, where the calls to action led. A standard SEO checklist would have spent its time on the part that was already fine.
The verdict can turn out wrong, and you change it when the data says so. Its value is that it forces you to choose how deep to go on each area, and to write the choice down.
| Area (illustrative SaaS) | Depth | Reason |
|---|---|---|
| Trial to paid, onboarding | Deep | The provisional verdict points here |
| Segments and ICP | Deep | Larger accounts seem to behave differently |
| Organic and content | Medium | Healthy traffic, review the conversion side only |
| Paid ads | Light | Small budget, no decision pending |
| Infrastructure, legal | Out of scope | Other people are better placed |
Tell the client what you did not dig into, and leave recommendations to someone else where you are not the best placed. An auditor with an opinion on everything loses credibility on the points that matter.
Secure the access on day one
Send the list with the framing note and ask for read access only. Where account-level data is enough, ask for exports with pseudonymized identifiers, and make sure your contract covers the personal data you will see.
| Source | What you use it for | Who usually grants it | Level to request |
|---|---|---|---|
| Web analytics (GA4 or equivalent) | Traffic, paths, on-site conversions | Marketing | Viewer |
| Search Console | Queries, indexing, click-through rate | Marketing or tech | Restricted user |
| Ad accounts | Spend and results by campaign | Marketing or agency | Read-only |
| CRM | Pipeline, stages, close rates, sources | Sales ops | Read-only or export |
| Billing | MRR, churn, expansion | Finance | Read-only or export |
| Product analytics or database | Activation and usage | Product or data | Read-only, or export |
| Email and lifecycle tool | Sequences and engagement | Marketing | Read-only |
| Support tool | Contact and cancellation reasons | Customer success | Export |
| CMS or site repository | Templates and tags | Tech | Viewer |
Log the date you asked and the date you received each one. When an access takes weeks to arrive or nobody knows who owns a tool, note it, because the audit of the stack and organization will need that fact.
Choose who to talk to
Book 30 to 45 minutes with each person and ask about the last time something happened. General opinions give you the company’s official story, and the last lost deal gives you what actually occurred.
| Person | What you are after | A question that works |
|---|---|---|
| Founder or CEO | The goal, the constraints, past attempts | What have you already decided not to do? |
| Head of sales and two reps | Real objections, lead quality | Walk me through the last deal you lost. |
| Marketing | What is measured and how | Which number do you trust least? |
| Product | Activation and roadmap | Which action tells you a user will stay? |
| Customer success | Reasons for leaving | Who should never have bought? |
| Finance | Margin and cash constraints | How exactly do you compute the ARR you report? |
| Customers: recent, lost, churned | The real buying story | What were you using before, and what made you look elsewhere? |
Keep a running list of contradictions. When sales says leads are poor and marketing reports a record quarter, you have found a definition problem or a real targeting problem, and either one shapes the rest of the audit. The customer conversations feed directly into the ICP and positioning work.
What to deliver: the framing note
One page, in plain words, sent before the analysis starts.
| Field | What to write |
|---|---|
| Decision owner | One name |
| Business goal | Metric, from what value to what value, by what date |
| Decision the audit informs | One sentence |
| Questions to settle | About six, each with its decision |
| Provisional verdict | Already solid, real lever, focus of the audit |
| Scope | Depth per area, what is left out, who is better placed |
| Access | Source, owner, date requested, date received |
| Interviews | Name, role, date |
| Deliverable | Prioritized action sheet and six-month roadmap |
| Presentation date | Booked now, with the people who decide |
Your next step: write the business goal as one sentence with a metric, a value and a date, then send it to the founder and the head of sales separately and compare what comes back.