The instinctive response to a sudden conversion drop is to change something: budgets, bid strategies, creative. If the cause isn’t in the campaign, those changes don’t fix anything and can make things worse by restarting learning.
The goal of diagnosis is to find the cause before touching the account.
Question One: Did Real Sales Drop?
Compare the platform’s numbers against the business’s own records: the order management system for e-commerce, or CRM lead counts and closed deals for lead generation.
Branch 1: Sales Steady, Platform Down
| Check | What to look for |
|---|---|
| Tag status | Google Ads conversion action statuses such as “Tag inactive” or “Needs attention,” per Google’s troubleshooting guide, and last-received times in Meta Events Manager |
| Site releases | A new checkout, a changed thank-you URL, or a theme update that dropped the tag container on the same date as the decline |
| Consent banner changes | A redesign that makes “Reject” more prominent lowers observed conversions overnight, and without Consent Mode that loss shows up directly |
| Browser and OS changes | Operating system or browser privacy updates that limit click identifiers or cookie lifetimes for part of the audience |
| Conversion definitions | Primary actions moved to secondary, imported GA4 key events removed, or double counting fixed |
| Platform definition changes | Meta’s March 2026 narrowing of click-through attribution to link clicks, which moved some conversions into engage-through reporting |
| Change history | Edits to conversion settings, counting method, or conversion windows on the day of the drop |
A drop that comes from fixing double counting or correcting primary conversions isn’t a loss. It’s a return to accurate reporting, and it should be explained to stakeholders as such. Consent mechanics are covered in GDPR and Google Consent Mode. The attribution change is covered in What Is the Attribution Window?.
Branch 2: Sales Down Too
Conversions are the product of impressions, click-through rate, and on-site conversion rate. Finding which factor fell points to the cause.
| Factor that fell | Likely causes | First check |
|---|---|---|
| Impressions (account-specific) | Budget limits, disapprovals, target changes, new competitors | Campaign status, ad approvals, impression share lost to budget or rank |
| Impressions (market-wide) | Demand itself declined | Search Console impressions and the same period last year |
| Click-through rate | Creative fatigue, stronger competitor offers | Frequency and CTR trends by ad |
| On-site conversion rate | Site speed, checkout errors, stock, pricing | GA4 checkout funnel, device and browser breakdowns, release notes |
Impressions falling: account or market?
If impression share holds while total impressions fall, the market shrank. If impression share fell, the problem is inside the account. That split is explained in What Is Impression Share?.
On-site conversion rate dropped
If ad traffic stays flat but sales fall, the problem is usually on the site. In GA4, the checkout funnel steps show which step the loss starts. A sudden drop on a single device or browser almost always points to a broken form or payment step there, not to advertising.
Price changes fit here too. When competitors run seasonal campaigns, your relative competitiveness drops, and conversion rates fall even if your price stayed the same.
Common Diagnostic Mistakes
- Reading the last few days. Conversions report with a lag, especially for longer purchase cycles.
- Changing several things at once. When budget, creative, and bidding all change on the same day, nobody can tell which change worked.
- Treating any platform as ground truth. Google Ads, Meta, and GA4 count the same sale under different attribution rules. Backend data is the anchor.
Summary
Start by comparing platform conversions with real sales. If sales held, check tags, site releases, consent changes, conversion definitions, platform attribution changes, and change history. If sales fell, find whether impressions, click-through rate, or on-site conversion rate declined, and whether the cause is inside the account or in the market. Diagnosing before changing anything avoids resetting learning for a problem the campaign didn’t cause.