A marketing audit is a structured review of marketing spend, output, and its numbers. Most published guides start with brand and channels. I start with revenue math. A channel review built on broken measurement is a review of fiction.
I have spent 15 years as a VP of Marketing inside PE-backed B2B SaaS companies. The audits that changed anything were never the ones that produced a 40-slide deck. They found one broken definition, fixed it, and made the forecast trustworthy again. This is the order I use, the evidence I pull, and the checklist I hand over.
What a marketing audit is (and is not)
An audit answers three questions in order. Are the revenue numbers real. Is the measurement that produces them intact. Is the spend behind them efficient.
It is not a brand critique. It is not a competitor teardown. It is not a list of channels you could try. Those are strategy exercises, and they are worthless until you trust the inputs.
Most published guides treat the audit as a sweep across the marketing mix. Pipedrive and SmartBug both frame it that way: inventory the channels, score each one, build a plan. That framing is fine for a company whose CRM is clean. In PE-backed B2B the CRM is rarely clean. The sweep then draws confident conclusions from bad data.
An audit that starts with channels ends with opinions. An audit that starts with revenue math ends with a fix.
The other failure mode is scope. An audit that inspects everything finds nothing worth acting on. I cap the whole exercise at four weeks and force a ranked fix list at the end. If the audit does not name the single biggest constraint, it did not work.
Before you start: the three numbers to pull
Do not open a dashboard until you have these three. Pull them from the systems of record, not a slide.
1. Pipeline coverage against the current-quarter number. Open pipeline divided by the quota or plan for the quarter. Pull it from the CRM opportunity table, filtered to the quarter's close dates. Note the stage mix, not just the total.
2. CAC payback in months. Fully loaded sales and marketing spend divided by new logos. Then divide by monthly gross profit per new customer. If finance and marketing quote different numbers, that gap is your first finding. The full method is in CAC payback.
3. Forecast accuracy over the last four quarters. Forecast at the start of each quarter against actual closed won. Four data points. If the average miss is over 20% either way, the pipeline math is the audit. Everything downstream is decoration.
These three take a day to assemble and they set the entire agenda. A company with 3x coverage, 14-month payback, and 5% forecast error has a channel problem. A company at 1.4x coverage with 30% swings has a pipeline problem, not a marketing one.
Step 1: Revenue math
Start where the money is, because every other finding gets ranked against it.
Pipeline coverage by segment and source. Blended coverage hides the failure. Split it by ICP segment and by lead source. A healthy 3x blend often hides a 5x self-serve segment carrying a 1.2x enterprise segment.
Stage conversion rates, stage by stage. Pull the last four quarters of opportunity history. Compute the conversion rate at each stage transition. Look for the stage where the rate collapsed, not the stage with the lowest rate. A permanently low rate at first meeting is a design choice. A rate that halved last quarter is an incident.
Deal velocity and aging. Median days in stage, and the count of opportunities past two times the median. Aged pipeline inflates coverage without ever closing. In most audits I run, 15% to 25% of open pipeline is dead and nobody has closed it out.
CAC payback by channel. Not blended. If one channel runs 9 months and another 34, the blended 18 tells you nothing useful.
Red flag to look for first: a coverage ratio that never changes month over month. That is a sign the pipeline is being managed to a target rather than measured.
Step 2: Measurement integrity
This is the step every published guide skips and the step that finds the most defects.
Conversion definitions. Write down what counts as a lead, an MQL, an SQL, and an opportunity. Then check whether those written definitions match what the systems actually do. They usually do not. Definition drift is why last year's numbers cannot be compared with this year's.
Form-to-CRM plumbing. Take a real form on the site. Submit it. Follow the record all the way through to the CRM. Time how long it takes to arrive. Check which fields populated. Check whether source attribution survived. I have found forms that wrote to an automation list and never created a CRM record.
Analytics event hygiene. In GA4, key events are the conversion measure. The platform counts every occurrence of one by default. In real audits I keep finding one form submission firing more than one event. Sometimes a thank-you page view is a key event alongside the submit event. The analytics lead count then runs 1.5x to 2x the real CRM record count. The board hears the analytics number. Sales hears the CRM number. Nobody notices they describe the same thing.
Attribution model and its limits. Note which model is in use and which touches it can actually see. Then check the dark spots: direct traffic with no prior touch, offline events, and AI traffic with no referrer. State the limits explicitly in the audit. An attribution number quoted without its limits is a claim, not a measurement.
Reconcile across systems. The same conversion should appear in the ad platform, the analytics tool, and the CRM. Where the counts diverge by more than 10%, that divergence is the finding. Do not average them. Chase the gap to its cause. More on the modeling side in marketing ROI measurement.
MQL as a governing metric. If MQL volume drives budget and headcount decisions, that is itself a finding. I explain why in why we do not measure MQLs.
Step 3: Channel efficiency
Now, and only now, look at channels. With trustworthy definitions, this step is fast.
Spend by channel against pipeline sourced by channel. One table. Twelve months. Cost per opportunity and cost per closed won by channel. The CRM is the denominator, never the ad platform.
Paid search waste. Search term reports for the last 90 days. Spend against terms with zero conversions. Brand versus non-brand split, reported separately, always. Blending brand into paid search is how an account looks fine while non-brand burns.
Conversion tracking on every paid platform. Confirm each platform's conversion action points at the same event the CRM counts. Mismatched conversion actions are why platform ROAS and CRM pipeline disagree.
Outbound efficiency. Meetings booked per rep per week. Then the meeting to qualified opportunity rate. If meetings are healthy and qualification is not, the targeting is wrong.
Channel concentration. What percentage of pipeline comes from the single largest source. Above 60% is a fragility finding, regardless of how well that channel performs today.
Step 4: Content, search and AI visibility
Content is audited as an asset base, not as a publishing calendar.
Traffic and conversion by page, not in aggregate. Rank landing pages by sessions, then by conversion rate. The interesting pages are high-traffic low-conversion, and low-traffic high-conversion. The first is a fix. The second is a promotion opportunity.
Decay. Pages that lost more than 30% of organic sessions year over year. In most B2B libraries a few pages carry most of the organic traffic. Their decay goes unnoticed while aggregate traffic looks flat.
Search coverage against the buying committee. Map the keywords you rank for against the questions each role in the committee asks. Gaps are usually at the top and the bottom, not the middle.
Technical fundamentals. Indexation, canonical tags, structured data on money pages, and Core Web Vitals. Fast to check, and a broken canonical can silently remove a page from search.
AI visibility, tested directly. Ask ChatGPT, Claude, Perplexity, and Google AI Overviews the questions your buyers ask. Record whether you appear, and who does. This is a real audit line now. A growing share of buyers never reach a blue link. Check that AI crawlers can reach your key pages. Check that the content answers a question in the first 40 words.
Step 5: Team, tooling and spend
Tool inventory against actual use. Every tool, its annual cost, its owner, and its last meaningful login. Every audit I have run found a four-figure subscription nobody opened in six months.
Overlap. Two tools doing the same job is common after a merger or a leadership change. Name both and pick one.
Data flow map. One diagram: where records enter, what transforms them, where they land. Most measurement defects are visible on this diagram before you test anything.
Headcount against the motion. Compare the team shape to the go-to-market motion. A product-led motion staffed for field marketing is a mismatch no campaign fixes.
Spend split. Percentage of budget on people, tools, media, and agencies. Track it against the pipeline each bucket produces.
Most of what breaks here is organizational, not tactical. The pattern behind it is in why GTM initiatives fail.
The marketing audit checklist
Twenty-five checks, in the order I run them. Each row names the evidence to pull, so this is executable rather than aspirational.
| Area | Question | Evidence to pull | Red flag |
|---|---|---|---|
| Revenue | Is pipeline coverage sufficient for the quarter? | CRM open pipeline / quarter plan | Below 3x, or unchanged month over month |
| Revenue | Does coverage hold by segment? | Coverage split by ICP segment | One segment below 1.5x hidden by blended total |
| Revenue | Where does stage conversion break? | 4 quarters of stage transition rates | A rate that halved in one quarter |
| Revenue | How much pipeline is actually dead? | Opportunities past 2x median stage age | Over 20% of open pipeline aged out |
| Revenue | What is CAC payback? | S&M spend, new logos, gross margin | Finance and marketing quote different numbers |
| Revenue | Does payback hold by channel? | Payback computed per channel | Range wider than 3x between channels |
| Revenue | Is the forecast trustworthy? | 4 quarters forecast vs actual | Average miss above 20% |
| Measurement | Are conversion stages defined in writing? | The written definitions document | No document, or it contradicts the CRM |
| Measurement | Do the systems match the definitions? | CRM stage entry criteria and automation rules | Stages advanced manually with no criteria |
| Measurement | Does a form submission reach the CRM? | Live end-to-end test submission | Record missing, delayed, or fields empty |
| Measurement | Does source attribution survive the handoff? | Source field on the test record | Source blank or overwritten to Direct |
| Measurement | Do GA4 key events overcount leads? | Key event count vs CRM record count | Analytics count 1.5x or more above CRM |
| Measurement | Is more than one event firing per submission? | GA4 DebugView on a live submission | Submit event plus thank-you page both counted |
| Measurement | Do platform, analytics and CRM agree? | Same conversion counted in all three | Divergence above 10% |
| Measurement | Are attribution limits stated? | The attribution model and its blind spots | Numbers quoted with no stated limits |
| Measurement | Does MQL volume drive budget? | Board deck and budget rationale | Spend justified by MQL count alone |
| Channel | What does each channel cost per opportunity? | Spend by channel / CRM opportunities | Any channel with spend and zero opportunities |
| Channel | Is brand separated from non-brand in paid search? | Search campaign structure and reporting | Blended into one performance number |
| Channel | How much paid search spend converts nothing? | 90-day search term report | Over 20% of spend on zero-conversion terms |
| Channel | Do platform conversion actions match CRM events? | Conversion action config per platform | Platform counting a different event |
| Channel | How concentrated is pipeline by source? | Pipeline share of largest source | Above 60% from one source |
| Content | Which pages lost traffic year over year? | Organic sessions by page, YoY | Top pages down over 30% |
| Content | Which pages get traffic but do not convert? | Sessions and conversion rate by page | High traffic, near-zero conversion |
| Content | Do AI assistants cite you for buyer questions? | Direct prompts to the major assistants | Competitors cited, you are absent |
| Team | Is every tool actually used? | Tool list, cost, owner, last login | A paid tool unopened for six months |
Marketing audit template
Copy this outline and fill it in as you go. Keep the whole thing to five pages.
1. Scope and dates. Period audited. Systems inspected. What was deliberately excluded.
2. The three numbers. Pipeline coverage. CAC payback. Forecast accuracy over four quarters. Each with its source system named.
3. Findings, ranked by revenue impact. One line each. Finding, evidence, estimated impact in dollars or months of payback. Ranked, not grouped by department.
4. The constraint. One sentence naming the single biggest blocker. If you cannot write this sentence, keep auditing.
5. Measurement defects. Every broken definition, plumbing gap, and double-counted event. Name the test that proved it.
6. Fix list. Each fix with an owner, a date, and the metric that proves it landed.
7. What I did not check. The honest boundary. An audit that claims total coverage is not credible.
What to do in the 30 days after
An audit that ends in a deck is a cost. An audit that ends in shipped fixes is an investment. Here is the sequence I use.
Week 1: fix measurement, nothing else. Every downstream decision depends on the numbers being real. Deduplicate the key events. Repair the form-to-CRM path. Write the conversion definitions down and get finance to sign off on the gross margin. This is unglamorous and it is the highest-leverage week of the month.
Week 2: close out dead pipeline and re-baseline coverage. Then recompute the three numbers. The deck built on week one's fixes will differ from last month's. That difference is the first deliverable.
Week 3: cut the obvious waste and reallocate. Zero-conversion search terms. Unused tools. Duplicate subscriptions. Move that budget to the shortest-payback channel, not the best-looking dashboard.
Week 4: ship one structural fix and instrument it. One. Pick the fix that hits the named constraint. Ship it. Attach a metric that moves inside a quarter. Then write down what you will check in 30 days.
That sequence is exactly the growth diagnostic I run as a 30-day engagement. I connect the evidence, find the constraint, rank the fixes, and ship the first one. I work through the marketing leader who already owns the mandate.
Sources
- Pipedrive, marketing audit guide: representative of how most published guides frame the audit as a channel and mix sweep.
- SmartBug Media, marketing audit: the agency framing of audit scope and deliverables.
- Google Analytics 4 key events documentation: how key events are counted, which is the mechanism behind the overcounting pattern described above.
- HubSpot forms documentation: form configuration and record creation behavior, for tracing the form-to-CRM path.
- OpenView SaaS benchmarks: benchmark reference for CAC payback and efficiency ranges used when ranking findings.