Why did my conversion rate drop? How to find the change behind it

Why did my conversion rate drop?

Your conversion rate dropped because one side of a fraction moved: conversions went down, sessions or clicks went up with people who convert less, or both. Before you look for a culprit, work out which side moved, because each side points to a different set of causes.

Each tool defines the rate differently. Google Ads calls conversion rate the average number of conversions per ad interaction, so it only covers ad clicks and other trackable interactions, and it can pass 100% if you count every conversion. GA4 renamed its conversion rate to key event rate: session key event rate is the share of sessions that included at least one key event, and user key event rate is the share of users who triggered one. If the drop shows in one tool and not the other, that difference is your first clue.

This article focuses on the rate. If the number of conversions fell, read why conversions drop and how to find the cause alongside it.

Is the conversion rate drop real or a tracking problem?

Rule out measurement before you change anything on the site. A rate can fall because the numerator stopped counting while the denominator kept going, and no landing page test will fix that.

1. Check the key event is still marked. In GA4 any event can become a key event once you mark it as a key event. A renamed event, a new form plugin or an edited trigger in Google Tag Manager can leave the old key event silent.
2. Check the tag fires on new pages. A redesigned template, a new checkout domain or a thank-you page that moved often ships without the tag.
3. Check consent changes. If you changed the cookie banner or consent mode, fewer users may be measured. GA4 only models users who decline analytics cookies when the property is eligible, which requires at least 1,000 events per day with analytics_storage denied for at least 7 days and at least 1,000 daily users with consent granted on at least 7 of the previous 28 days. Below those thresholds, events from users who decline consent are not reported.
4. Check reporting identity. Modeled data only appears with the Blended reporting identity, so a settings change can move the rate without any change in behaviour.
5. Compare with the source of truth. Put orders from your shop or leads from your CRM next to GA4 and Google Ads for the same days. If real orders held steady, the problem is measurement.

For a deeper checklist on GA4 counting, see how to tell a GA4 tracking problem from a real loss of visitors.

Did your traffic mix change?

Often the site did not get worse; the visitors changed. If you added traffic that is colder, cheaper or less qualified, the average rate falls even when every existing visitor converts as before.

Typical traffic-mix changes that lower conversion rate:

• A new upper-funnel campaign, display or video placements, or a broader audience on Meta Ads.
• Broader match types or new keywords in Google Ads that bring in research queries.
• A newsletter, social post or press mention that sends many casual visitors in one day.
• A shift toward mobile, where forms and checkouts are often harder to complete.
• An organic ranking for an informational query that adds visits but few buyers.

To test it, segment the rate by channel, campaign, device, landing page and new versus returning users. If the rate for each segment is stable but the share of a low-converting segment grew, you have a mix shift, not a site problem. Example: a shop where paid search converts at 3% and a new display campaign at 0.4% will see its blended rate fall the week display launches, while the paid search rate does not move.

A mix shift is not automatically bad. The question is whether the extra traffic brings enough conversions at an acceptable cost, which is a decision about the campaign, not about the website.

Which website changes lower conversion rate?

If the rate fell inside a stable segment, look at what changed on the path to conversion. Site changes hit every channel at once, so a drop across all segments on the same day points here.

For speed, use a public reference instead of a gut feeling. Google's guidance on Core Web Vitals says a good experience means LCP within 2.5 seconds, INP of 200 milliseconds or less and CLS of 0.1 or less, measured at the 75th percentile of page loads on mobile and desktop. If a release pushed a key landing page past those thresholds, it is a strong candidate.

Also check changes outside the website that alter the offer on the page: a promotion that ended, a free shipping threshold that went up, or a competitor launching a cheaper offer in the same week.

How do I match the drop date to the change that caused it?

Find the exact day the rate moved, put every change from the days before it on one timeline, and test the candidates one by one. Guessing from memory is how teams blame the wrong change.

1. Pin the date. Look at the daily rate, not weekly, and compare the same weekdays so normal weekly patterns do not hide the break.
2. Find where it broke. Segment by channel, device, landing page and country. A drop limited to one segment narrows the list of suspects.
3. List every change from the previous one to two weeks. Include your team's edits in Google Ads (Google Ads change history), Meta Ads, site releases, tag and consent changes, CRM or form tools, prices, promotions and external events.
4. Rule out tracking first. Use the checks above before you act on any site hypothesis.
5. Test one hypothesis at a time. Revert or fix one change, compare an affected segment with an unaffected one, and wait for enough data before moving to the next suspect.
6. Record the outcome. Note which change caused the drop and what you did about it, with the date. Add it as a note in GA4 too (GA4 annotations).

Example: a lead-generation site sees its session key event rate fall from 4.1% to 2.8% on a Tuesday. Only mobile is affected. The timeline shows a form plugin update deployed on Monday night. Testing the form on a phone shows the submit button hidden below a new consent checkbox. Fixing the layout brings the mobile rate back within a week, and the team logs both the update and the fix.

How do I explain the drop and stop the next one becoming a mystery?

Show the client or your manager three things: what moved (rate, numerator or denominator), which change caused it, and the evidence that connects them. A dated timeline of changes makes that conversation short. Our guide on how to explain a conversion drop to a client has a ready structure.

The lasting fix is a habit: log every change when you make it, with date, owner and reason, across ads, site, tracking and CRM. Then the next time the rate moves, the list of suspects already exists. A marketing change log is exactly that list.

Changeline keeps these changes on one to-scale timeline per client project, shared with your team, so you can see what happened the day before a metric moved. You can start a free change log while it is in early access. It does not import data from ad platforms or overlay metrics yet; you log the changes and compare with your reports.