Revenue Leak AuditGuide · Cross-functional15 related articlesBy Manoj Kumar

Paid media revenue leaks: where ad budgets actually go missing

Every paid media account leaks money. The question is how much, and where. After auditing accounts from $16K to $150K a month across B2B and B2C, the leaks show up in the same six places almost every time. They rarely look dramatic on a dashboard. They look like a ROAS that won't move, a CAC that creeps up every quarter, or a sales team complaining about lead quality.

This guide is the framework behind our Revenue Leak Audit. Each section explains one leak, what it looked like in a real account, and how we closed it. If you want the numbers for your own account, the audit is free and takes 30 minutes.

1. Search-term waste

The biggest leak in Google Ads is paying for searches that were never going to buy. Broad match and Performance Max expand into queries you would never choose yourself, and without daily governance the waste compounds.

At FlightSearchDirect, spending $80K–$150K a month, ROAS had fallen to 1.02. We grew the negative keyword list from 200 to more than 1,800 terms and classified every incoming query daily. Within 60 days, spend on irrelevant queries fell 67%, and ROAS recovered to 2.08. For a US travel agency that sells by phone, the same kind of framework found that 40% of spend was going to queries that never produced a call (case study).

How to check it yourself: pull the search terms report for the last 90 days, sort by cost, and tag every term as buyer, researcher or irrelevant. If more than a fifth of spend sits in the last two groups, you have a governance problem, not a bidding problem. Then look at match types. Broad match with Smart Bidding can work, but only with a negative list that grows every week.

Go deeper:

2. Bid integrity: optimising toward the wrong conversion

Smart Bidding does exactly what you tell it. If the conversion you feed it is a form start, a page view or a repeat purchase, it will find more of those, cheaply, and none of them pay the bills.

A Nordic bank's previous agency had been optimising for application starts. Only 18% of those completed. Once we rebuilt tracking around completed applications, the account delivered 1,970 applications at $8.30 each (case study). For a SaaS subscription business, switching to value-based bidding tied to subscription lifetime value lifted value per conversion by 261.9% (case study).

How to check it yourself: open conversion settings and list every action marked as primary. Anything that is not revenue, a qualified lead, a completed application or a real call should be secondary. Then check conversion values. If every lead is worth the same, the algorithm cannot tell a $300 customer from a free trial.

Go deeper:

3. Attribution leaks

When every platform claims the same sale, budget follows whichever claims loudest. Last-click and misconfigured models systematically underfund the channels that start the journey.

For a North American ecommerce brand, correcting attribution showed that 28% of revenue came from channels that had been under-credited and underfunded. Rebalancing spend on that basis was part of how revenue grew from $257K to $610K year on year (case study). Our analytics and attribution work starts here because every other fix depends on it.

How to check it yourself: compare what each platform claims against your store or CRM revenue for the same month. If the platforms add up to more than 120% of real revenue, attribution is double-counting and your budget split is based on fiction. Move reporting to blended numbers (MER, new-customer CAC, pipeline) before you move budget.

4. Tracking gaps

A conversion you can't see is a conversion the algorithm can't learn from. Missing server-side events, uncounted phone calls and lead forms that count spam are all tracking leaks.

For a Canadian immigration law firm, call tracking showed that 45% of previous "leads" were not genuine consultation requests. Removing them from the conversion count and bidding on real consultations cut CPL by 38% in six weeks and more than doubled the qualified consultation rate (case study).

How to check it yourself: submit a test lead and a test purchase, then trace them through to the ad platform. Check that phone calls are tracked, that spam and duplicate leads are filtered out, and that server-side events are deduplicated against the browser pixel. Every gap you find is a signal the algorithm is not learning from.

Go deeper:

5. Creative and landing-page decay

Ads and landing pages wear out. A creative that worked in month one quietly becomes the most expensive thing in the account by month four, and nobody notices because nobody is scoring it.

At FareScraper we scored 40 creatives daily on CTR and CPA and auto-paused any that stayed below threshold for three days. CTR rose from 3.8% to 6.1% and CPA fell 34% in 90 days. For a pet insurer, testing eight landing-page variants over 90 days lifted page conversion rate 44% and cut CPA from $54 to $30.39 (case study).

How to check it yourself: rank active ads by cost over the last 30 days and compare their CPA or CAC with their first 30 days. Anything that has decayed by more than 30% should be paused or replaced. For landing pages, compare conversion rate by page and device. A page that converts at half the rate of your best one is costing you every click that lands on it.

Go deeper:

6. Audience leaks: one campaign for two different buyers

When two audiences with different motivations share one campaign, creative and budget get pulled toward whichever converts cheaper, and both underperform.

A Life Plus, a DTC meal-delivery brand spending $30K–$60K a month, was targeting health-conscious shoppers and government-funded customers with the same ads. Splitting them into two tracks with separate creative and landing pages cut CAC from $102 to $74 while spend scaled. Connecting Shopify first-order data also showed that 34% of previously reported "conversions" were repeat purchases, so the real new-customer CAC had been even worse than it looked.

How to check it yourself: list the distinct buyer types you sell to and check whether each has its own campaign, creative and landing page. If two groups with different motives share everything, split them and give each its own budget and CAC target.

Leaks by business model: B2B pipeline vs DTC CAC

In B2B the leak usually shows up as pipeline that never closes: cheap MQLs, slow lead-to-SQL, sales ignoring marketing leads. For a Salesforce ISV partner, industry-specific landing pages and ABM targeting cut CPL from $98 to $54 and grew demo bookings 3.5× in one quarter (case study).

In DTC and ecommerce it shows up as rising CAC and platform ROAS that doesn't match the bank account. The fix is measuring blended new-customer CAC and MER rather than each platform's self-reported numbers. See our ecommerce and DTC and Meta Ads approach.

In both models, the fix is to move the conversion the platform optimises for as close to revenue as your data allows: SQLs and opportunities in B2B, first orders and contribution margin in DTC.

Checklist: what to fix first

  1. Search terms tagged for the last 90 days, with negatives reviewed weekly
  2. Primary conversions limited to revenue or qualified outcomes
  3. Conversion values reflect real customer value
  4. Platform-claimed revenue reconciled against store or CRM revenue
  5. Calls, forms and purchases traced end to end, with spam filtered
  6. Ads and landing pages ranked by decay, with a pause rule
  7. Each buyer type has its own campaign, creative and landing page

Frequently asked questions

What is a revenue leak in paid media?

A revenue leak is ad spend that produces no real business outcome: irrelevant search terms, bidding toward the wrong conversion, broken attribution, tracking gaps, worn-out creative or mixed audiences. Most accounts have several at once.

How much budget do most accounts leak?

It varies by account, but in the audits behind this guide the leaks were large: 67% of spend on irrelevant queries at one travel platform, 40% of spend on non-calling queries at another, and 45% of reported leads that were not genuine at a law firm.

How long does a Revenue Leak Audit take?

The free audit call is 30 minutes. Most accounts show their first material leak within 20 minutes of looking at search terms, conversion settings and attribution.

More articles in this guide