Present the audit and bring the team along
The format of the deliverable and how to get the conclusions accepted.
Put the verdict on the first page: what is solid, where the real lever sits, and therefore what the team should spend the next six months on. Everything after that is evidence. On an audit I ran for a B2B SaaS, traffic had good bases and conversion had never been worked on, so the analysis centered on conversion and said so from the start. A reader who stopped there still knew where the effort had to go.
The other half of this chapter is about people, because correct findings that nobody adopts produce nothing. I’ll keep the example from the roadmap chapter, a fictional B2B SaaS at 2M ARR.
Build the deliverable in four layers
Different readers need different depths, and one document can’t serve them all.
| Layer | Format | Contents | Main reader |
|---|---|---|---|
| Verdict | One page | What is solid, the lever, the focus, what was left out | Founder, board |
| Analysis | Slide deck | Findings by stage, each with its proof | Head of growth, team leads |
| Resource sheets | Spreadsheet | Technical items as problem, consequence, action, with the URLs | Developer, content |
| Action sheet | Spreadsheet | The prioritized plan with owners, days and measures | Whoever runs the program |
The action sheet closes the audit. If a reader can only keep one file, that’s the one.
Write the one-page verdict
Six lines are enough, and each has to survive being forwarded without context.
| Line | Example for our fictional SaaS |
|---|---|
| What is solid | Organic traffic grows and reaches the right topics |
| The real lever | Conversion: readers arrive and have nowhere useful to go next |
| So the audit focused on | Funnel mapping, CTAs, bottom-of-funnel pages, attribution |
| Not covered | Paid social (no usable history), pricing (outside the mandate) |
| Better handled by | The ads agency on bidding, the product lead on the product roadmap |
| Decision needed | Approve the six-month plan and name one owner this week |
The “not covered” and “better handled by” lines are what make the rest believable. Scope was settled when you framed the audit, so this page reports against the questions agreed at the time.
Make each finding stand on its own
A finding has an observation, a proof, a consequence, an action and a confidence level. Slides without those five turn into opinions the moment someone disagrees.
| Part | Weak | Usable |
|---|---|---|
| Observation | CTAs could be improved | Blog CTAs sit at the bottom of the page and none is tracked |
| Proof | None | Page captures and a tracking export showing zero CTA events |
| Consequence | Lost conversions | We can’t tell which articles produce signups, and few readers reach the CTA |
| Action | Optimize the CTAs | Add a contextual CTA after the first section, pointing to the matching feature page, tracked on every template |
| Confidence | Not stated | Measured for the position and tracking, estimated for the impact |
The confidence label is the part most audits skip. Tag each claim as measured, estimated or judgment. A founder can accept a judgment call when it’s labeled as one, and the label gives you room to be wrong on the estimates without losing the measured findings along with them.
Stay blunt about the work. A useless page goes in the trash, by URL, and a category holding two items should not exist yet. Ask what each page is for in the report itself, and aim every criticism at the page or the process, since the person who built it is probably in the room.
Talk to the people concerned before the meeting
Whoever owns an area that comes out badly should hear it from you first, in private. You catch factual errors, you learn the history you missed, and nobody gets surprised in front of their boss.
| Person | Likely worry | What to bring them |
|---|---|---|
| Founder | Where the money goes, whether the ceiling can move | The verdict, the budget, the one decision |
| Head of growth | Being judged on choices they inherited | The proof, what is already solid, the plan they will own |
| Content lead | Hearing their work labeled useless | Declared versus real funnel stage, page by page, and what to keep |
| Developer | Tickets arriving from nowhere | The resource sheet, with effort and order |
| Sales | Lead quality, again | Who reads compared with who buys |
Change the document when they are right. A correction made before the meeting costs you nothing, and that person now has a stake in the conclusions.
Run the meeting around decisions
Walking through the deck page by page uses up the time and the attention. Send the analysis a day ahead and use the meeting for what needs people together.
| Minutes | Topic | What must come out of it |
|---|---|---|
| 10 | The verdict, including what was left out | Agreement on where the lever is |
| 5 | What is already solid | The team sees what will not change |
| 30 | Three to five findings that drive the plan | Challenges heard, corrections noted |
| 15 | The six-month plan, the parked list | Sequence and capacity accepted |
| 15 | Decisions, one line at a time | An owner and a start date per line |
Three to five findings is a hard limit. The others live in the deck for whoever wants them.
Answer objections with the data, and keep them on record
You will hear the same few objections. Prepare the answer before the meeting.
| Objection | Useful response |
|---|---|
| “We already tried that” | Ask when, how it was measured and on which audience. If the test holds up, drop the action |
| “The numbers are wrong” | Agree on the source, recheck it live or within 48 hours, and record the outcome |
| “This isn’t the priority” | Show the score and ask which line it should replace, given the capacity |
| “That’s outside your remit” | Often true. Move it to the “better handled by” line and name who decides |
An objection you can’t answer goes on the list with a date and an owner. Improvising a defense costs you more credibility than saying you will check.
Close on a decision log
The meeting’s real output is one table, sent within 48 hours.
| Action | Decision | Owner | Start | Note |
|---|---|---|---|---|
| Track blog CTAs and menus | Accepted | Head of growth | Month 1 | |
| Build versus landing pages | Amended: two pages first, then review | Content lead | Month 2 | Volume reduced after pre-read |
| Second language | Refused for now | Founder | Reopen after the month 3 checkpoint |
A refusal with a reason is a good outcome, since the team knows the topic is closed and why. The dangerous line is the one nobody decided on, because it comes back six months later as a surprise.
Your next step: write the six lines of the verdict before you touch the deck. If you can’t fill “the real lever” in one sentence, the analysis isn’t finished. Once the log is sent, the program moves to the tracking and iteration routine.