Why did my ROAS drop? How to find the change behind a falling ROAS

Why did my ROAS drop?

ROAS is conversion value divided by ad spend, so a ROAS drop always comes from one of two sides: you recorded less value, or you spent more to get it. Before you look for causes, find out which side moved.

Google defines it plainly in its Target ROAS help page: conversion value divided by ad spend, multiplied by 100%. A target of 500% means you want $5 in sales for every $1 spent. That formula gives you four patterns, and each one points to a different set of suspects.

The third row is the one people miss. If conversions held steady but value per conversion fell, no bid change explains it on its own. Something changed in what you sell, how you price it or how the value is sent to the ad platform. For the spend and volume side of the same problem, see why your CPA increased.

Is the ROAS drop real or a reporting effect?

Check this before anything else: a large share of sudden ROAS drops are reporting effects, not lost revenue. The three most common are conversion delay, attribution changes and broken value tracking.

Conversion delay

Google warns in its tips on measuring Smart Bidding performance that some conversions take "days or even weeks" to be reported, so recent performance can look weaker than past performance. Google suggests judging over longer periods, "like a month, or at least 50 conversions", and over at least 2 full conversion cycles. To see your own lag, use the segment Conversions > Days to conversion on the Campaigns or Ad groups report, with a date range that ends at least 30 days ago, as explained on the conversion lag help page.

Attribution changes

If someone changed the attribution model, the numbers can move without a single customer behaving differently. In GA4, the reporting attribution model lives in Admin > Data display > Events > Attribution settings, and Google states that changing it "applies to historical and future data" (GA4 attribution settings). In Meta Ads Manager, compare like with like: if the attribution setting on the ad set or in your report columns differs between the two periods, the comparison is not valid.

Value tracking

Open a few recent orders and compare the value in your shop or CRM with the value the ad platform recorded. Typical breaks: the purchase tag sends a fixed default value instead of the real one, currency changed, tax or shipping is now included or excluded, a duplicate tag was removed, or a new conversion action was set as primary. Any of these changes ROAS overnight and none of them is a performance problem.

Which ad account changes lower ROAS?

Inside the ad platforms, ROAS usually falls after you asked the system for more volume. A lower target, a bigger budget or a broader campaign buys extra conversions at a lower return, and that is often by design.

Google's own guidance on Target ROAS is direct: to get more volume you lower the target, to get more value you raise it. After a target change "the bidder will react immediately but will need some time to hit the new target (give it 1-2 conversion cycles)". So a ROAS dip in the week after a target edit may be the strategy settling, not failing.

Google Ads keeps these edits in Change history (Campaigns menu) for the past 2 years, with who made each change, according to the change history help page. Our Google Ads change history guide and Meta Ads activity history guide show how to filter them.

What changes outside the ad platforms cause a ROAS drop?

Many ROAS drops start on the website or in the business, not in the ad account. The ad platform only sees that the same clicks now produce less revenue.

• Pricing and promotions: a sale ended, a discount code started, free shipping thresholds changed, or prices went up. Discounts lower the value per order even when conversion rate rises.
• Product mix and stock: best sellers went out of stock, a high-value product was removed, or ads now send traffic to cheaper items.
• Site releases: a new checkout, a form change, a slower template or a broken payment method cut conversions on the same traffic.
• Consent and tags: a new cookie banner or tag manager change lowers how many purchases are observed.
• CRM and offline values: for lead generation, a change in how lead values or offline conversions are imported changes reported value without changing real revenue.
• Market: seasonality, a competitor promotion or a shift in demand. These are real, but you should only conclude this after ruling out your own changes.

The common thread: these changes are made by people outside the media team, so they rarely appear in Google Ads or Meta Ads history. That is why the analysis below starts with a single timeline.

How do I isolate the cause of a ROAS drop?

Put every change from the days before the drop on one timeline, rule out tracking first, then test one hypothesis at a time. This turns a vague "ROAS is down" into a short list of dated suspects.

1. Date the drop precisely. Look at daily ROAS, conversion value and spend for the last 30 to 60 days. Note the first day the line broke, not the day someone noticed.
2. Find where it happened. Segment by campaign, device, country and product category. A drop limited to one campaign points to an account change; a drop everywhere points to tracking, the site or the market.
3. Build one timeline. List every change in the 14 days before the drop: Google Ads and Meta edits, bid and budget changes, automated changes, site releases, tag and consent changes, price and promotion changes, CRM imports and external events.
4. Rule out tracking. Compare platform values with real orders, check conversion lag and attribution settings, and confirm no conversion action was added, removed or changed from primary to secondary.
5. Test one hypothesis at a time. Start with the change closest to the drop date that affects the right scope. Revert it or compare against a campaign it did not touch, then wait for at least one conversion cycle before judging.
6. Write down the result. Record what you tested, what happened and what you decided, so the next drop starts with history instead of guesses.

Example: a shop sees ROAS fall from 450% to 320% on 14 September. Conversions are flat, value per order fell by about a quarter. The ad accounts show no edits that week, but the timeline shows a sitewide 25% discount launched on 13 September. The ads did not get worse; each order was worth less.

This is the method behind our marketing root cause analysis post, applied to value instead of volume.

How long should I wait before reacting to a ROAS drop?

Wait long enough to see complete data, but not so long that a real problem keeps costing money. In practice that means at least one full conversion cycle after the drop, and 1 to 2 cycles after any major bidding or budget change.

Google recommends you "allow 1-2 conversion cycles for the system to stabilize after major changes to bidding or budget before you make further adjustments". If you react to a three-day dip by cutting the target again, you stack a second change on the first and lose the ability to tell which one did what.

Two exceptions justify acting fast: a tracking break (value suddenly zero, or orders missing) and a site problem such as a broken checkout. Both are confirmed by checking real orders, not by waiting. Also keep in mind the data requirements: Google lists at least 15 conversions in the past 30 days for Target ROAS on Search and Shopping, and recommends reporting values for 4 weeks or 1–2 conversion cycles before switching to value-based bidding (source). Campaigns below that volume will show noisy ROAS from week to week.

How do I explain a ROAS drop to a client or manager?

Lead with what happened, what caused it and what you are doing, in that order. A client trusts a dated cause far more than a list of possible reasons.

• The fact: "ROAS went from 450% to 320% from 14 September; spend was flat, value per order fell."
• The cause: the specific change on the timeline, with its date and owner.
• The evidence: the segment where the drop is concentrated and the checks you ran on tracking.
• The action: what you changed, or why you are waiting, and when you will review again.

Our guide on how to explain a conversion drop to a client has a ready structure for this conversation. The work is much faster when changes are logged as they happen: a marketing change log gives you the timeline in seconds instead of reconstructing it from five tools. You can start a free change log in Changeline and keep ad, site, tracking and CRM changes for each client on one timeline.