Diagnosing a Sudden RPV Collapse on a Single Channel
When one channel's revenue per visitor drops 25% week-over-week, run the suspects in this order: tracking break, creative fatigue, audience expansion, landing-page change, promo cannibalization.
Quick answer
When a single channel's RPV drops 25%+ week-over-week, check the suspects in this order: (1) tracking break, (2) silent landing-page change, (3) creative fatigue, (4) audience expansion, (5) sitewide promo cannibalization. Rule out the cheap ones (tracking, page change) before you touch ad spend — they account for roughly half of false-alarm RPV crashes.
Diagnosing a Sudden RPV Collapse on a Single Channel
A fixed-order diagnostic for finding why one channel's revenue per visitor dropped sharply while the rest of the site held steady.
A sudden RPV collapse on a single channel is the specific failure mode where one acquisition source — say Meta paid social or Google non-brand — sees revenue per visitor fall 25% or more versus the prior week, while sitewide RPV stays within normal variance. The diagnostic exists because the obvious explanation ("the channel is broken") is wrong about half the time. The actual cause is usually a tracking regression, a landing-page change a developer shipped on Friday, or a sitewide promo that quietly redirected revenue to a different source. Running the suspects in a fixed order prevents you from pausing campaigns that were never the problem.
The page assumes you've already confirmed the drop is real and isolated to one channel. If you haven't, start with distinguishing a channel-specific RPV collapse from a sitewide drop — the two have completely different root causes and remediation paths.
Why this happens: the five mechanisms
RPV is sessions in the numerator's denominator and revenue on top. Anything that inflates sessions without proportionally inflating revenue — or deflates revenue without touching sessions — shows up as an RPV crash. There are only five common ways that gap opens on one channel.
Tracking breaks deflate revenue (orders stop firing, or fire to a different source) while sessions keep counting. Creative fatigue and audience expansion inflate sessions with worse-converting traffic. A landing-page change drops conversion rate on the channel that sends traffic there. A sitewide promo can shift attribution credit toward direct or email, starving paid channels of revenue they actually earned.
Don't pause the campaign first
The expensive mistake is pausing a Meta or Google campaign on Monday morning, then discovering Wednesday that a GTM container deploy on Friday silently dropped the purchase event. You've now lost three days of learnings on a campaign that was performing fine. Always rule out tracking before you touch spend.
How to detect: run the suspects in this order
Suspect 1 is tracking. Open GA4 or your warehouse and compare purchase events per session for the affected channel versus the prior week. If the event rate dropped but the channel's click-through to product pages didn't, you're looking at a tracking break, not a performance problem. The full procedure for ruling out a tracking break before blaming the channel is the first thing to run.
Suspect 2 is a silent landing-page change. Pull the top three landing URLs for the channel and diff them against last week's versions — page weight, hero copy, above-the-fold offer, form fields. A developer shipping a "small" hero swap on Friday afternoon is the single most common cause of a Monday RPV drop. Catching a silent landing-page change that tanked channel RPV is suspect 2 specifically because it's cheap to check and high-frequency.
Suspects 3-5: the spend-side causes
Suspect 3 is creative fatigue. Look at frequency and CTR on the top-spending ad sets over the last 14 days. If CTR is down 20%+ and frequency is above 3.0, you're seeing creative fatigue as the RPV culprit on Meta — the algorithm is showing tired ads to a saturated audience, and the people still clicking are lower-intent.
Suspect 4 is audience expansion. Check whether anyone on the team toggled Advantage+ audiences, broadened a lookalike percentage, or removed an exclusion list. Audience expansion diluting your RPV without you noticing is invisible in the campaign UI — the spend graph looks identical, but the audience composition shifted toward colder, lower-AOV buyers.
The promo wildcard
Suspect 5 is the one most teams miss: a sitewide promo running on email and direct can cannibalize paid channel RPV by shifting last-click credit. Paid traffic still sees the product, still adds to cart — then bounces back via an email coupon the next day. Paid gets the session, email gets the revenue. See: when a sitewide promo cannibalized one channel's RPV.
How to fix: per-suspect remediation
Tracking break: roll back the offending GTM container version, or republish the missing purchase tag. Landing-page change: revert the deploy if the conversion rate delta is >15%, or A/B test the change against the prior version if the drop is marginal. Creative fatigue: rotate in three new creative concepts and cap frequency at 2.5. Audience expansion: re-tighten the audience to the prior week's configuration and watch RPV for 72 hours. Promo cannibalization: switch the affected campaigns to data-driven attribution for the duration of the promo.
The full Monday-morning 24-hour RPV triage timeline gives you a clock — suspect 1 by 10am, suspects 2-3 by lunch, suspects 4-5 by end of day. If you've ruled out all five and RPV is still down, see how many days of RPV decline before you act vs wait — three consecutive days is the threshold to escalate, not one.
Frequently asked questions
25% week-over-week on a channel doing more than 10% of total revenue. Smaller drops on smaller channels are usually within normal variance and not worth a full triage. If the channel drives less than 5% of revenue, wait for a 3-day trend before investigating.
Tracking breaks are binary — either purchase events are firing or they aren't — and take five minutes to verify. Creative fatigue takes an hour to diagnose properly. Always run the cheap, high-frequency check first. Roughly one in three Monday RPV alarms is a tracking issue.
That's a different problem. A sitewide RPV drop points to checkout, payment processor, sitewide pricing change, or a broken promo. See distinguishing channel-specific RPV collapse from a sitewide drop before running this diagnostic — the suspects are completely different.
Yes, especially on Google Shopping and price-comparison-heavy verticals. If a competitor undercuts your price during their Black Friday warm-up, click volume holds but conversion rate drops. Check competitor pricing on your top three SKUs before blaming creative.
Broadening a lookalike from 1% to 3%, or turning on Advantage+ audience expansion, lets the algorithm spend the same daily budget on a wider, colder audience. Sessions hold or rise, but the marginal buyer has lower intent and lower AOV — so RPV drops even though the dashboard looks normal.
No. Pausing kills the learning phase and adds a week of recovery once you restart. Run the diagnostic with the campaign live. The only exception is if you've confirmed creative fatigue and have replacement creative ready to ship within 24 hours.
Pull conversion rate by landing URL for the affected channel, week-over-week. If one URL's CR dropped sharply while others held, check the page's deploy history. A 15%+ CR drop on a single URL inside one week is almost always a deploy, not a traffic-quality issue.
Attribution changes the apparent RPV without changing reality. If GA4 quietly shifted from last-click to data-driven, paid channels can lose 10-20% of credited revenue overnight. Always check whether your reporting attribution window or model changed in the period you're diagnosing.
About one in five cases. Creative fatigue plus a promo, or audience expansion plus a tracking partial-break, is common. Run all five suspects to completion even if you find a likely culprit at suspect 2 — confirming the others rules out compounding causes.
Partially. Tracking, landing page, and promo cannibalization apply directly. Creative fatigue translates to subject-line and template fatigue. Audience expansion doesn't apply — replace it with list-quality decay (recent unengaged segments getting included in sends). The five-suspect structure holds; the labels shift.
Get an AI expert review of your site
Paste your URL — Metricuno's AI runs the same heuristic checks a senior CRO consultant would, scoring your page and prioritising the fixes that'll move conversion fastest.