Marketing Audit: A 12-Point Check for Where Your Spend Leaks

A marketing audit is a structured review of how a business attracts, captures and converts demand, built to find where money or leads are lost. The most useful version follows one lead from first click to closed revenue and tests each handoff, rather than grading channels in isolation.

The reason is where leaks hide. Each platform reports its own results from the inside, and inside the platform things look fine. The ad account shows conversions. The CRM shows records. The inbox shows messages. The loss happens between them: a form that posts but never creates a contact, a lead assigned to someone on leave, a campaign tag that falls off before the CRM sees it.

This is the layer above the single-channel reviews already in this library. A Google Ads Audit Checklist covers the ad account, How to Run a Social Media Audit covers the profiles, and What Is a Content Audit? covers the pages. The twelve checks here test how those parts connect. The checklist is our own assembly of the vendor documentation cited in each check. It is not an industry standard, and the severity scale later in the article is ours.

Work in order. The first four decide whether you can trust any number the later checks produce. Each test is meant to be a short check rather than a project, which is deliberate: an audit that needs a quarter gets postponed.

Start with definitions, because they are free and they explain a surprising number of disputes. If marketing counts a form fill as a lead and sales counts only a booked appointment, both teams can be right and the funnel report still be useless. Write one definition for each stage and publish it.

Then test tracking with real actions. Google describes a key event in Analytics as an event that measures an action particularly important to the success of the business, and notes that any event you collect can become one. That puts the burden on you to mark the right ones. Google Ads says a conversion is whatever you choose to identify as valuable, so a campaign may be optimizing for an action that stopped meaning anything after a site redesign. Check what each campaign is actually told to chase.

Compare counts across analytics, the ad platform and the CRM for the same thirty days. They will not match, because each system counts differently and applies privacy rules differently. The test is whether you can explain the gaps. A difference you can attribute to consent settings or deduplication is fine. A difference nobody can explain is a finding.

Finish the group with source retention and the form-to-CRM handoff. Click a tagged ad on a clean browser, submit the form and read the CRM record. The Thank-You Message Is an Operational Promise and Lead Routing That Survives Lunch, Leave, and Broken Syncs go deeper on both. Tag every test submission clearly so it can be deleted and never counted as a lead.

In our view response time is among the cheapest leaks to fix, because it is a rule and a routing change more than a project. Harvard Business Review's 2011 analysis of online sales leads reported that most companies were not responding nearly fast enough. Rather than quote someone else's benchmark, compute your own: for the last hundred inbound leads, take the gap between submission and first human reply, and look at the distribution, not the average. A median of fifteen minutes with a tail of two-day waits is a different problem from a uniform four hours. Speed to Lead Without the Five-Minute Theater covers how to set a service level that people can keep.

Then read what happens after the reply. The lead quality loop is the audit's most neglected point, because it needs sales to cooperate. Look at the last twenty lost deals. If loss reasons are blank or all say “no response”, marketing is flying without feedback and the ad platforms are being optimized on form fills alone. Send Lead Quality Back to Meta Without Teaching the Wrong Lesson shows how to close that loop carefully.

Pages and forms come next. Google's published thresholds for a good page experience are a largest contentful paint within 2.5 seconds, an interaction to next paint of 200 milliseconds or less, and a layout shift of 0.1 or less, measured at the 75th percentile of page loads and split by mobile and desktop. Use them as a screen, not a growth plan. Core Web Vitals Are a Starting Line, Not a Conversion Guarantee explains why. A page that passes and hides its phone number still leaks.

For search, the page indexing report in Search Console lists why pages were not indexed. Confirm that the pages earning leads are in the index and that business profile details match reality. A Local SEO Audit Checklist and An SEO Audit Template You Can Reuse give the long form of this check.

Content and reputation share a test: look at what a prospect sees. Sort pages by leads in the last year, not by traffic, and ask which pages get visits and no inquiries. Read the last twenty reviews and the replies. Review Management Is a Workflow, Not a Reply Button covers the routines behind that second check.

For paid media, the key question is what the account has been told to optimize and whether that action predicts revenue. Then read the search terms or placement reports for waste. The full ad-account method sits in the Google Ads audit post; this check only confirms the account is pointed at the right target.

AI search visibility needs plain checks before clever ones. Google's guidance says a page must be indexed and eligible to appear in Search with a snippet to be a supporting link in AI Overviews or AI Mode, and that there are no additional technical requirements. It also lists allowing crawling in robots.txt and by any CDN or hosting layer among the fundamentals that stay relevant. OpenAI documents OAI-SearchBot as the crawler used to surface sites in ChatGPT search, so accidentally blocking it removes you from those answers. Bing Webmaster Tools now reports how often your content is cited in Copilot and Bing's AI summaries, with the caveat that counts do not show ranking or authority. AI Search Optimization: What Moves Citations and What Is Folklore covers what to do after these checks.

The last point joins the audit to the money. Work out fully loaded customer acquisition cost and its payback, as shown in Customer Acquisition Cost: How to Calculate It Fully Loaded Without Fooling Yourself. Then check consent. If the business texts or calls leads, find where each consent was captured and where the record lives. TCPA Compliance: What the Rules Mean for Calls, Texts and Lead Follow-Up sets out what that record should show.

Four of the twelve checks depend on pushing a real submission through the system, so it pays to run one disciplined procedure that serves all of them. Agree it with a manager first so nobody mistakes the test for a customer or loses an afternoon on it.

Run all twelve eventually. When time is short, the order below is our judgment about where leaks usually cost the most for each model. It is our judgment, not a research finding, so override it when your own data points elsewhere.

Give each finding three ratings from one to three: money exposed, confidence that the leak is real, and effort to fix. Then sort by exposure and confidence, highest first, and break ties by lowest effort. The scale is our own working method. It exists to force a conversation about size, not to produce a precise number.

To size exposure, use arithmetic you can defend. Take the leads affected, multiply by the close rate they would have had if handled well, then multiply by gross profit per customer. Here is an illustrative case, not a real one: if 15 leads a month get no first reply within a day, a handled lead would close at 20 percent, and a customer brings $1,500 of gross profit, the exposure is 15 × 0.20 × $1,500, or $4,500 a month. The estimate is rough. It is still more useful than the sentence “we think response time is a problem.”

Fix measurement first unless something is cheaper and obviously broken. A form that does not create a CRM record is both. A campaign optimizing on the wrong action poisons every later metric, so it moves up. Redesigning a landing page waits until you trust the conversion data it will be judged on.

Ten working days is realistic for a business with a handful of channels, provided access to analytics, the CRM, the ad accounts and Search Console is arranged on day one. The schedule below is a guide, and the days compress when a business has fewer systems.

Whoever runs it should not own the channels being audited. The incentive is simple: people rarely find the leak in their own pipe. An internal analyst from outside marketing, a finance partner or an outside reviewer all work. What matters is access and independence, not a title.

Keep it short: the five biggest leaks, each with the money exposed, the evidence (a timestamp, a screenshot, a test record), the owner and the fix. Add a list of what you could not check and why, so nobody reads silence as a clean bill of health. A report that lists thirty findings of equal weight gets filed.

Be clear about the limits. An audit is a snapshot, and a new campaign or site change can reopen a leak the following week. Test submissions pollute the CRM unless tagged and removed. Never place test calls to real prospects' numbers, and use your own phone. Re-run points 2 to 5 after any site rebuild, CRM migration or change of ad platform. They are the checks most likely to quietly break.