Sudden spike in conversions? How to tell if it's real and find the cause

Why did my conversions suddenly spike?

A sudden spike in conversions has one of two causes: something changed in how conversions are counted, or something changed in the traffic or offer that produces them. The first kind is far more common when the jump is large, happens overnight and has no matching rise in sales or qualified leads.

Treat a spike with the same discipline as a drop. A spike that is really double counting inflates CPA and ROAS reports, misleads clients and, if the action is used for bidding, teaches Smart Bidding to chase the wrong signal. A spike that is real deserves the same attention, because you want to repeat it.

Group the possible causes before you open any report:

• Measurement changes: a tag firing twice, a new conversion action set to primary, a changed counting setting, a server-side setup sending the same event as the browser, an imported GA4 key event added next to an existing tag.
• Your own campaign changes: budget increases, new campaigns, bid strategy or target changes, new audiences, broader match types.
• Site and offer changes: a new landing page, a shorter form, a promotion, a price cut, a new payment method.
• External events: seasonality, a competitor leaving the auction, press coverage, a public holiday.

This article shows how to tell those groups apart and how to prove which change caused the spike. If your problem runs the other way, see why conversions drop and how to find the cause.

Is the conversion spike real or a tracking problem?

Check whether the spike exists outside the ad platform before you report it. Real conversions leave a trace in your back end: orders in the shop, leads in the CRM, calls in the call log. Counting errors only show up in the reporting tool.

1. Compare with the source of truth. Put platform conversions next to orders or CRM leads for the same days. If Google Ads shows 120 purchases and the shop shows 60 orders, you have a counting problem, not a demand spike.
2. Split by conversion action. In Google Ads, segment the conversions column by conversion action. A spike that sits in one action, or in an action that did not exist last week, points to tracking.
3. Check conversions per click. If conversions rise but clicks and sessions stay flat, ask why each visitor now converts twice as often. Real improvements are rarely that abrupt.
4. Read the dates correctly. Google Ads reports conversions on the date of the ad impression, while most other tools use the date of the conversion. Add the "Conversions (by conv. time)" column before you compare with GA4 or your CRM, and keep in mind that conversion data can take up to 7 days to fully appear.
5. Check the value, not just the count. A purchase spike where the average order value halves, or many conversions with identical values, often means the same order is recorded more than once or a test value is being sent.

If the ad platform and the back end rise together, the spike is probably real and you move on to the business cause. If only the platform rises, read the next section. For reconciling the two tools, see why GA4 and Google Ads conversions don't match.

Which tracking changes create a fake spike in conversions?

Most fake spikes come from a short list of measurement changes, and each has a recognisable symptom. Match what you see to the table before you touch any campaign.

Two details matter when you fix these. A change to the counting setting applies only to future conversions, so the inflated days stay inflated in past reports. And conversions created from Google Analytics events are secondary by default and can only be made primary inside Google Ads, so a doubled count often means someone made that switch on purpose and did not remove the old tag.

The article on duplicate conversions in Google Ads goes step by step through finding and removing duplicates.

How do I isolate which change caused the conversion spike?

Find the first day of the spike, list every change made in the days before it, and test those changes one at a time, tracking first. The cause is usually the change closest to the start date that fits the symptom.

1. Date the first day of the spike
2. Rule out tracking changes
3. Line up every change before it
4. Test one hypothesis at a time

Date the start precisely. Use daily data and, in Google Ads, the by-conversion-time columns, so you know the day the count changed rather than the week. A spike that starts at a single point in time is almost always a change. One that builds over several days is more likely demand, or a ramp-up after a budget increase.

Collect changes from every source. Google Ads "Change history" shows your team's edits and automated changes, Meta's activity history shows ad account edits, Google Tag Manager keeps a version for every publish, and developers or the CMS log site releases. Then add the changes nobody records: a new form plugin, a promotion e-mail, a price change. The Google Ads change history guide explains what each filter shows.

Match the symptom to the change. A spike in one conversion action matches a change to that action or its tag. A spike across all actions with more clicks matches a budget or bidding change. A spike that also shows in the CRM matches an offer, site or demand change.

With every change on one timeline, the answer is often visible at a glance: you would see the GTM publish or the new primary action on the same day the count jumped, before you open a single report. In a shared Changeline timeline, Google Ads edits are imported automatically and the team logs the rest (tags, site, CRM, promotions), so the list is ready when the spike appears.

Example: On Monday a client asks why form leads jumped from about 20 a day to 41 over the weekend while sales calls stayed flat. Lining up the changes from the days before showed that on Friday a developer added a GA4 "generate_lead" key event, which was imported into Google Ads and set to primary, while the original Google Ads form tag kept firing. Every submission counted twice. Setting the imported action to "Secondary (observe only)" brought daily leads back to about 21.

What if the conversion spike is real?

If orders or qualified leads rose along with the platform numbers, the spike is real and the job becomes finding which of your changes, or which outside event, produced it. A real spike you can explain is one you can repeat.

Check quality before you report the win. More leads that the sales team rejects, or more orders that are refunded, are not a better result. Look at qualified leads in the CRM and at net revenue after refunds for the same days.

To separate a seasonal rise from your own work, compare with the same weeks last year, as described in how to tell seasonality apart from your own changes.

What should I do after I find the cause of the spike?

Fix the measurement if tracking was the cause, record the change and its date, and protect your bidding and reports from the inflated data.

• Remove the duplicate, don't just hide it. Set the extra action to "Secondary (observe only)", remove the duplicate tag, or send a shared event_id from the Meta Pixel and the Conversions API. Then confirm the next days match your back end.
• Watch Smart Bidding. If the inflated action was used for bidding, the algorithm saw better results than it got. Google warns that frequent changes to budgets, targets or conversion goals can reset the learning phase, so make one clean fix instead of several. See what triggers the Smart Bidding learning period.
• Mark the affected dates. Add a note in GA4 and in your client reports for the days the count was inflated, so nobody compares next month with a doubled baseline. The GA4 annotations guide shows how.
• Tell the client early. Explain what was double counted, from which date, and what the real numbers are. A correction from you lands better than one the client finds alone.
• Log the fix. Record the tracking change and its date in your marketing change log, so the next spike or drop starts from a full history.