Article · Template

The Startup Marketing Audit I Run in Week One

This is the actual audit that opens my paid engagements, published in full. Run it yourself in an afternoon; most founders find their constraint before the coffee goes cold.

Dev Sharma·Fractional CMO·10 min read·Updated July 16, 2026

On this pageStep 1: Run the funnel math (60 minutes)+
Key takeaways
  • Five checks in strict order: funnel math, positioning, channel economics, measurement honesty, ownership. Each layer can fake health while the one under it bleeds.
  • The whole audit takes about 4.5 hours; the readout should rank findings by one criterion: which single fix makes every other problem easier.
  • Fix one constraint per quarter, then re-audit. Fixing three things at once fixes nothing.

The short answer: a startup marketing audit checks five things in a strict order: funnel math, positioning, channel economics, measurement honesty, and team ownership. The order matters because each layer can fake health while the one under it bleeds. The full checklist is below; the 3-minute self-scored version names your weakest layer instantly.

I've run some version of this audit at every stage of my career: as a first marketing hire, as a growth lead at a marketplace with roughly 2.3 million users, across 120+ client engagements at the company I founded, and now in week one of every advisory engagement. The version below is the one I actually use, minus only the client-specific spreadsheets.

The five audit layers, in strict order Stacked layers with time budgets: funnel math 60 minutes, positioning 90 minutes highlighted because it hides the real constraint in about half of engagements, channel economics 60 minutes, measurement 45 minutes, ownership 30 minutes. Total about four and a half hours. FIG. 01 / THE AUDIT, IN ORDER FIVE LAYERS, ONE afternoon 001 FUNNEL MATH 60 MIN 002 POSITIONING the real constraint in about half of my engagements 90 MIN 003 CHANNEL ECONOMICS 60 MIN 004 MEASUREMENT 45 MIN 005 TEAM & OWNERSHIP 30 MIN Run in this order. Each layer can fake health while the one under it bleeds. DEV SHARMA
Fig. 01: The five layers with their time budgets, about 4.5 hours total. Positioning is accented because it's where the real constraint hides most often, dressed up as a demand problem.

Step 1: Run the funnel math (60 minutes)

Pull one quarter of numbers, stage by stage: visitors, signups or leads, activated or qualified, paying, retained at 90 days. Compute stage-to-stage conversion, then answer three questions: Which stage converts worst against any reasonable benchmark? When did someone last look at it? What would revenue be if only that stage doubled?

The point isn't precision, it's location. Most startups optimize the stage they can see (traffic, usually) while the actual leak sits two stages downstream where nobody owns the number. Write the leak's name down; the rest of the audit tests whether it's a symptom or the disease. If the math itself is impossible to pull, skip ahead: measurement IS your finding.

Step 2: Test your positioning (90 minutes)

Three checks, all external, because internal opinions about positioning are worthless:

  • The repeat test. Ask three recent customers to describe what you are, as if to a colleague. If you get three different categories, buyers can't refer you, search for you, or defend the purchase internally.
  • The one-liner swap. Could your homepage headline sit on a competitor's site unchanged? If yes, it isn't positioning yet.
  • The sales-call drift. Listen to two recorded calls. Count how the rep describes the product versus how the site does. Drift here predicts message failure everywhere downstream.
From the field In roughly half of my engagements, this step finds the real constraint, dressed up as a demand problem. The fix path is a repositioning sprint, and it's cheaper than the ad budget being wasted on an unclear message. (See how this played out when clarity became the acquisition engine at my own company.)

Step 3: Interrogate channel economics (60 minutes)

Per active channel, four numbers: fully loaded cost (including people time, not just spend), pipeline or revenue attributed, trend over two quarters, and what happens if you turned it off for two weeks. Then sort every channel into three buckets:

The three channel buckets Three columns. Compounding channels build assets: content, SEO, community, lifecycle. Rented channels stop when spend stops: most paid media. Unknown, highlighted, is the bucket where nobody can say, and it is usually the biggest. The unknown bucket is the scandal. FIG. 02 / SORT EVERY CHANNEL THREE BUCKETS, ONE scandal COMPOUNDING builds assets that persist: content, SEO, community, lifecycle KEEP AND FEED RENTED stops when spend stops: most paid media RENT ONLY WHERE MATH HOLDS UNKNOWN nobody can say what it returns. Usually the biggest bucket. AUDIT THIS FIRST The unknown bucket is the scandal. Budgets hide in it because nobody has to defend it.
Fig. 02: The three channel buckets. My Channel Autopsy tool runs this sorting as a six-question verdict per channel if you want it structured.

Step 4: Check measurement honesty (45 minutes)

Take last month's trials or leads. Count them three ways: what Meta claims, what Google claims, what your CRM says. If the platforms' sum exceeds the CRM's total by more than half, your budget decisions are being made by whichever dashboard lies most confidently. This exact reconciliation is the first thing I built at Futurestay, and it changed where every subsequent dollar went. Bonus check: can anyone on the team explain, in one sentence, which touch gets credit for a conversion? If not, "attribution" is a word, not a system.

Caveat Don't turn this step into an attribution-tool procurement. The audit's job is naming the gap, not buying software for it. Most startups fix 80% of measurement dishonesty with one spreadsheet and one agreed counting rule.

Step 5: Map ownership (30 minutes)

For each of these, write one name (a name, not a team): the pipeline number, positioning, each active channel, the funnel's worst stage, measurement integrity. Every blank or committee answer is a finding. Then the harder question: who on the team can kill an underperforming program without founder sign-off? If the answer is nobody, your marketing has no immune system, and everything else in this audit will decay after it's fixed.

Want your five layers scored right now?

The free diagnostic runs this audit's logic as a 3-minute questionnaire and names your weakest layer instantly. No email required for the result.

Run the free diagnostic

How to read your results

Sort findings by one criterion: which single fix makes every other problem easier? That's the constraint; everything else waits. The classic mistakes are fixing the loudest problem (usually a channel) instead of the load-bearing one (usually positioning or measurement), and fixing three things at once, which fixes nothing. One constraint, one quarter, then re-audit. This sequencing discipline is most of what people pay senior marketers for, which is why the fractional CMO guide spends so much time on it.

Budget context makes the discipline non-optional: marketing budgets are running at 7.7% of company revenue per Gartner's 2025 CMO Spend Survey, with The CMO Survey's Spring 2025 read at 9.4%, still historically tight. Nobody has slack budget to fix the wrong thing first.

Do this Write the readout as one page: five layers, one finding each, one constraint circled, one quarter's plan for it. If the readout needs ten pages, you're documenting instead of deciding.

Frequently asked questions

How long does a startup marketing audit take?

The self-run version above takes about 4.5 hours of focused work, assuming your numbers are pullable. My paid version runs two to four weeks because it adds customer interviews, sales-call review, and a findings readout the exec team signs. If pulling the numbers alone takes days, that delay is itself your first finding.

How often should we re-audit?

Quarterly, and always after fixing the named constraint. The audit is cheap and the constraint moves: fix positioning and the bottleneck usually surfaces next in measurement or channel economics. Annual audits are theater; the business changes faster than that at startup stage.

Should the audit be run internally or externally?

Run it yourself first; the template exists so you can. Bring in an outsider when the findings implicate things insiders can't say out loud: the founder's pet channel, a mis-scoped hire, an agency relationship past its usefulness. External auditors are paid partly to be unpopular.

What tools do I need?

A spreadsheet, your analytics, your CRM export, and two recorded sales calls. Deliberately no more than that. Audits that start with a tool purchase usually end as tool configuration projects, and the constraint survives another quarter.

Want it run for you?

Three options, in ascending order of my involvement: the free 3-minute diagnostic scores all five layers from your answers; the Ghost Diagnostic reads your public site and scores what's visible in about a minute; and the paid version of this audit runs two to four weeks with your real data, customer calls, and a findings readout your exec team signs. Structures and pricing are in the cost breakdown.

Ran it and found something ugly?

Good. That's the audit working. Bring the finding to a 20-minute call and I'll tell you if I'd fix it the same way. No deck, no pitch.

Book a working call

Dev Sharma

Fractional CMO for seed to Series B startups

Bootstrapped Blockwiz to roughly $5M ARR as a solo founder, led growth and lifecycle at Paxful to about 2.3 million verified traders, and was the first marketing hire at MyPoolin. 120+ brands across SaaS, fintech, crypto, ecommerce, and edtech.

I write the way I run engagements: numbers first, and the caveats stay in.