The budget that evaporates
A mid-market e-commerce site (€5M to €50M in annual revenue, €30,000+ in monthly paid budget) loses on average €50,000 to €200,000 a year to invisible data leaks. The cause is rarely a spectacular bug or a server outage. It is silent anomalies that eat into the budget day after day without triggering a single alert in standard tools.
These leaks have one thing in common: they sit between tech and business. They are too technical for the CMO to spot and too tied to budget for developers to prioritize, so they fall into an organizational blind spot.
This article reviews the 7 most common leaks, their estimated cost and how to detect them.
Overview: the 7 leaks
| # | Leak | Visible symptom | Estimated annual cost | Korvus alert | Detailed article |
|---|---|---|---|---|---|
| 1 | Dead conversion pixel | Falling ROAS, zero conversions in reports | €15,000 to €60,000 | pixel_dead | What a dead pixel costs |
| 2 | Zero-value checkout | Conversions counted but €0 revenue in ad platforms | €10,000 to €40,000 | checkout_zero | Smart Bidding and bad data |
| 3 | Misconfigured CMP | Conversion data gradually declining | €5,000 to €25,000 | Diagnosis via consent_status | Your CMP is eating your data |
| 4 | Sponsored 404 pages | High bounce rate on some campaigns | €5,000 to €30,000 | sponsored_oos | The 404 tax |
| 5 | Out of stock + paid traffic | Paid clicks on product pages that cannot convert | €10,000 to €35,000 | sponsored_oos | Out of stock + paid traffic |
| 6 | Degraded LCP on paid traffic | Mobile conversion rate falling for no apparent reason | €8,000 to €30,000 | perf_drop | Every 100 ms of LCP has a cost |
| 7 | Wrong GA4 revenue | Gaps between GA4 and accounting, skewed budget decisions | €5,000 to €20,000 | Indirect diagnosis | GA4 and the wrong revenue figure |
Combined total: €50,000 to €200,000 a year for a mid-market site, and most of these leaks run on the same site at the same time.
Leak #1: the dead conversion pixel
What happens. A Meta, GA4 or TikTok pixel stops firing. No visible error. GA4 simply shows fewer conversions. Bidding algorithms lose their signal and start to drift.
Why it's invisible. Sentry doesn't see pixels, because they aren't server errors. GA4 doesn't know data is missing. The traffic manager sees ROAS falling but looks for the cause in targeting or creative.
Estimated cost. For a €50,000 monthly paid budget, a pixel that stays dead for 72 hours (average detection time) costs €5,000 to €10,000. With 2 to 3 incidents a year, the annual cost reaches €15,000 to €60,000.
Korvus detection. The pixel_dead alert fires within minutes when a pixel stops firing while paid traffic is active and consent is granted. The first alert already carries the financial impact in euros.
Leak #2: the zero-value checkout
What happens. The purchase event fires correctly, but the value field is empty, zero or in the wrong currency. Ad platforms count a conversion, but with €0 in revenue.
Why it's invisible. The number of conversions looks normal. It's revenue per conversion that drops. Google Ads default reports show conversion volume, not unit value. The problem goes unnoticed for days.
Estimated cost. Smart Bidding switches to "maximize conversions" instead of "maximize conversion value". CPA rises, ROAS falls. Estimated annual impact: €10,000 to €40,000 in misallocated budget.
Korvus detection. The checkout_zero alert fires as soon as a purchase event reports a zero or null amount. The impact is calculated by multiplying the number of zero-value transactions by the average order value.
Leak #3: the misconfigured CMP
What happens. Your CMP (Cookiebot, Axeptio, Didomi) blocks conversion pixels for most visitors. Not because they refuse cookies, but because the technical implementation is broken: consent signal sent too late, default mode not configured, consent_update missing.
Why it's invisible. The data decline is gradual. It often coincides with a banner change or a CMP update. The team assumes "it's normal, people refuse cookies".
Estimated cost. A CMP that drops the acceptance rate from 55% to 25% hides an extra 30% of conversions. Smart Bidding under-optimizes. Budget cuts land in the wrong place. Indirect impact: €5,000 to €25,000 a year.
Diagnosis. Korvus records consent_status in every session. When a pixel doesn't fire even though consent was granted, the pixel_dead alert fires. When consent is refused, Korvus knows the missing data is normal and raises no false positive.
Leak #4: sponsored 404 pages
What happens. Live ads point to pages that no longer exist: deleted products, URLs changed after a redesign, expired landing pages. Every click is billed. No conversion is possible.
Why it's invisible. Google Ads doesn't recheck URLs after the ad is created. GA4 records the visit but doesn't flag it as a "paid click to a 404". You have to cross-reference the data manually, and nobody does that day to day.
Estimated cost. 50 clicks a day to 404s at a €0.85 CPC = €1,275 a month, or €15,000 a year. For sites with dynamic catalogs, the cost can reach €30,000 a year.
Korvus detection. The sponsored_oos mechanism detects paid traffic landing on a page that cannot convert, whether it's a 404, an out-of-stock product or a broken page.
Leak #5: out of stock + paid traffic
What happens. A product goes out of stock. The product feed isn't updated in real time. Shopping campaigns keep serving the ad. Every click lands on a page with an "Unavailable" button.
Why it's invisible. Merchant Center feeds update 1 to 4 times a day. The lag is structural. During that window, the paid budget runs for nothing. Over a weekend with no one on watch, losses pile up unchecked.
Estimated cost. For a catalog of 200+ products with regular stockouts, the annual cost is between €10,000 and €35,000 in wasted clicks.
Korvus detection. The sponsored_oos alert identifies, in real time, paid clicks landing on out-of-stock pages. The impact is put in euros: number of clicks x average CPC.
Leak #6: degraded LCP on paid traffic
What happens. The Largest Contentful Paint (LCP) of your key pages exceeds 2.5 seconds, especially on mobile. The page loads slowly. The visitor bounces before seeing the product or the buy button.
Why it's invisible. Core Web Vitals are measured globally in Search Console. But the financial impact depends on the traffic: a slow page with free organic traffic costs less than a slow page with paid traffic at €1.20 a click. Standard tools don't make that distinction.
Estimated cost. Every extra 100 ms of LCP beyond 2.5 s costs roughly 0.7% of conversions. For a site with a €30,000 monthly paid budget and a 3.5 s mobile LCP, the annual loss is €8,000 to €30,000.
Korvus detection. The perf_drop alert fires when LCP exceeds the configured threshold specifically on paid traffic. The impact is calculated by crossing the number of degraded sessions with the average CPC and the estimated bounce rate.
Leak #7: wrong GA4 revenue
What happens. GA4 shows revenue that differs from your back office. Sometimes +20%, sometimes -30%. The causes vary: double counting, missing currency, unfiltered test transactions, purchase events on non-transactional pages.
Why it's invisible. Few teams systematically compare GA4 revenue with accounting revenue. The gap grows slowly. Budget decisions have been made on wrong figures for months without anyone noticing.
Estimated cost. If GA4 overstates your ROAS by 20%, you over-invest in campaigns that are less profitable than they look. The reverse is also true: a GA4 that understates by 20% makes you cut profitable campaigns. Indirect impact: €5,000 to €20,000 a year in poor allocation.
Manual diagnosis. Compare GA4 revenue with back-office revenue every week. Identify gaps > 10%. Check the causes: duplicates, currencies, test transactions. Korvus detects the checkout_zero events that contribute to these gaps, but a full audit of GA4 revenue remains a manual diagnosis.
Why these leaks coexist
These 7 leaks are common, and they coexist on most mid-market e-commerce sites, all the time. A single site can have, at once:
- A TikTok pixel that has been dead for 2 weeks (nobody checks TikTok Ads every day)
- 30 out-of-stock products still receiving Shopping traffic
- A 3.8 s mobile LCP on product pages
- A cookie banner blocking 60% of tracking
Each leak on its own seems manageable. Combined, they add up to €50,000 to €200,000 in wasted budget a year. And none of them triggers an alert in standard tools: without a dedicated tool, an incident runs until someone notices, and all that time the paid budget keeps running.
What Korvus covers natively
Korvus monitors 4 of the 7 identified leaks in real time:
| Leak | Korvus alert | Detection time |
|---|---|---|
| Dead pixel | pixel_dead | < 15 minutes |
| Zero-value checkout | checkout_zero | < 15 minutes |
| Degraded LCP (paid traffic) | perf_drop | < 30 minutes |
| Out of stock + paid traffic | sponsored_oos | < 30 minutes |
For the other 3 leaks (misconfigured CMP, 404 pages, wrong GA4 revenue), Korvus provides the context data (consent_status, page state, reported amounts) for a fast diagnosis. Detection is automated for what costs the most, and the data is there to diagnose the rest.
The cost of doing nothing
A mid-market site with a €50,000 monthly paid budget that suffers 3 to 4 major incidents a year (dead pixel, massive stockout, checkout bug) loses on average €80,000 a year between wasted budget and degraded Smart Bidding.
Detection in minutes instead of days cuts the impact of each incident by 90%. On €80,000 of annual losses, going from 72 hours to 15 minutes of detection time saves €70,000 to €75,000 a year.
Dedicated monitoring pays for itself with the first incident avoided.
Priority action plan
Week 1. Audit the 3 easiest leaks to check: conversion pixels (are they firing?), purchase amounts (are they correct?), campaign URLs (do they exist?).
Week 2. Measure your CMP's real acceptance rate. Compare GA4 revenue with back-office revenue. Identify the gaps.
Week 3. Set up continuous monitoring on the 4 highest-impact anomaly types. Configure alert thresholds in euros so every notification triggers an action.
Ongoing. Audit the GTM container every 90 days. Optimize the cookie banner. Check product feeds. A fixed leak comes back with the next deployment, catalog change or tool update.
Key points
- 7 data leaks cost mid-market e-commerce sites between €50,000 and €200,000 a year
- These leaks are invisible in standard tools (GA4, Sentry, Google Ads)
- Korvus natively detects 4 of the 7 leaks in real time, with the impact in euros
- The ROI of monitoring kicks in with the first incident detected in minutes instead of days