Skip to content

Problem

Performance Max not converting: how to fix it

A Performance Max campaign is only as good as the three things you hand it: accurate conversion data, a healthy product feed and enough budget to learn on. When it spends without producing orders, one of those three is usually missing rather than the campaign being wrong for the store. Check the inputs before you rebuild, because a rebuilt campaign fed the same signals will reach the same place.

By CartKernel · Last reviewed 2026-09-07

Does this look familiar?

  • The campaign spends its full budget daily and reports very few conversions
  • Most of the spend sits in display and video placements rather than shopping or search
  • The asset group shows a low ad strength and only a handful of assets
  • Conversions appeared during the first fortnight and then stopped
  • Reported conversions look reasonable but the store cannot see the orders
  • One asset group carries almost all the spend while the others receive almost none

Causes, ranked

Why it happens, most common first

Check them in this order. The first two account for most cases we open.

  • most common

    The conversion action being optimised toward is wrong or noisy

    If the campaign is bidding on page views, add to cart or a broad conversion set rather than purchases with values, it will find people who do the cheap action. Duplicate purchase actions, a test conversion left enabled, or every action marked as primary all send the same confusing instruction.

  • most common

    The campaign has too little budget or too little history to learn

    Value based bidding needs a steady flow of conversions before it can predict anything. A budget that produces only a few orders a week keeps the campaign permanently in a learning state, and every edit restarts it. This is the most common reason a small account sees nothing but spend.

  • common

    Only a small part of the catalogue is eligible

    If most products are disapproved, missing images, or excluded by the listing group, the campaign has very little shopping inventory to work with. It then shifts spend into display and video, where cost per click is low and purchase intent is lower still.

  • common

    The asset group is thin, so the campaign leans on cheap inventory

    Missing headlines, a single image size, no video and no logo variants limit which placements the campaign can enter. When only display sizes are fillable, that is where the money goes. Ad strength is a rough proxy for how many placements you can reach.

  • common

    The target return is set where the campaign cannot operate

    An ambitious target throttles delivery to a trickle of high intent traffic, which is fine, or, when paired with a small budget, produces erratic spend with no conversions at all. Targets changed frequently are worse than a target that is slightly wrong and left alone.

  • occasional

    Audience signals describe the wrong person

    A signal built from a broad interest segment or an outdated customer file points the campaign at people who will not buy. Signals are directional rather than binding, but a poor one wastes the early learning period when the campaign is deciding where to look.

  • occasional

    The landing experience does not match the promise

    Traffic sent to a homepage rather than a relevant collection, or to a product page that is slow on mobile, converts poorly regardless of targeting. The campaign learns from that outcome and reduces bids where it can, which shows up as spend drifting to cheaper placements.

The fix

In this order

Each step is something you can do today. Do them in sequence; skipping ahead is how a review fails twice.

Prevent it next time

  • Keep exactly one primary purchase conversion action with a value attached to every order
  • Fund a campaign only when the budget can produce enough weekly orders for the system to learn from
  • Review asset group coverage and product eligibility on the same schedule as budget reviews
  • Change one setting at a time and record the date, so learning periods can be read cleanly
  1. Audit the conversion actions before anything else

    List every conversion action in the account, check which are marked primary, and confirm purchase is counted once per order with a transaction value. Deactivate duplicates and demote soft actions to secondary so they are recorded but not bid on.

  2. Reconcile reported orders against the store

    Compare Google Ads conversions with orders in the store for a fixed week, allowing for the attribution window. If the platform reports far more, the tag is firing more than once or on the wrong page. Fix that before judging performance.

  3. Check how much of the catalogue is actually eligible

    Filter Merchant Center to products approved for Shopping ads and compare with the listing group in each asset group. Products excluded, disapproved or missing images cannot carry the campaign, and returning them is the highest value change available.

  4. Fill the asset group properly

    Add the full range of headlines and descriptions, images in landscape, square and portrait crops, at least one video even if it is a simple product sequence, and both logo shapes. More fillable formats means more placements where intent is higher.

  5. Give the campaign one clear job

    Split asset groups by product category or margin band so each has a coherent message and landing experience, rather than one group covering the whole catalogue. Use separate campaigns only where the budget can support each of them independently.

  6. Set a target the campaign can reach and leave it

    Start near the return the account has actually produced on similar traffic, or begin on maximise conversion value without a target for a short period. Then hold the setting for at least two full learning cycles before adjusting.

  7. Replace weak audience signals with first party data

    Use a recent customer list and high intent site segments as the signal, and add search themes written in the language a buyer would use. Remove broad interest segments that describe a demographic rather than a buying moment.

  8. Fix the landing page the traffic actually reaches

    Send each asset group to the most relevant collection or product page, check it on a mid range phone over a slow connection, and make sure price, delivery and returns information is visible without scrolling far.

When to get help

Get help when conversion tracking is verified, the feed is healthy and the campaign still spends without producing orders, because the next step is a structural decision rather than a settings change. That usually means choosing between fewer, better funded campaigns, moving budget into standard Shopping and search where control is greater, or accepting that the catalogue and margin cannot support the channel at its current price. Those calls benefit from someone who can model contribution per order before the budget is committed again.

Get a Growth Analysis

Free. We reply within one business day.

Questions

Asked alongside this problem

By CartKernel · Last reviewed

How long should a Performance Max campaign run before I judge it?

Allow the learning period to complete and then a comparable period of stable delivery, which for most stores means several weeks rather than several days. Judging earlier usually leads to an edit that restarts learning, and a campaign that is restarted repeatedly never produces a readable result.

Should I exclude display and video placements to force shopping traffic?

There is no placement level switch that does this cleanly. The practical lever is to make shopping inventory attractive and available by returning products to eligibility, and to fill the asset group so the campaign has better options than the cheapest impressions.

Is a low ad strength score the reason a campaign is failing?

It is a symptom more than a cause. Low ad strength usually means missing asset formats, which limits which placements can be filled. Adding the missing sizes and a video widens where the campaign can compete, which tends to help, but it will not fix a tracking or feed problem underneath.

Would a standard Shopping campaign work better for a small catalogue?

Often, yes. A store with few products, a modest budget and clear intent to capture can get more predictable results from a Shopping campaign plus a search campaign, where negative keywords and bids are directly controllable. Performance Max tends to reward larger catalogues and steadier conversion volume.

Related problems, answers and terms

All problems
ProblemGoogle Ads conversion tracking not working: how to fix itGoogle Ads conversion tracking not working shows as zero, double or unverified conversions. Find which of the three you have, then fix the tag.OpenProblemPerformance Max spending on brand searches: how to fix itPerformance Max spending on brand searches inflates reported return. How to measure the brand share, exclude it and keep the campaign honest.OpenProblemShopping ads not showing: how to fix itShopping ads not showing usually traces to eligibility, not bidding. Work through account, product, campaign and targeting checks in that order.OpenProblemGoogle Ads ROAS dropped: how to fix itWhen Google Ads ROAS dropped, decide first whether revenue fell or measurement changed. The order of these checks saves weeks of wrong fixes.OpenAnswerIs Performance Max worth it for small ecommerce stores?Performance Max can work for a small store once brand is excluded and the campaign is kept simple. When it fits, when Standard Shopping fits better.OpenAnswerHow long does the Google Ads learning phase take?The learning label usually clears within a week or two, while a campaign takes longer to settle. What lengthens it and what quietly restarts it.OpenGlossaryTarget ROAS biddingTarget ROAS bidding lets Google set bids to hit a return you specify. How the target works, a worked setup, and the conditions it needs to perform.OpenGlossaryReturn on ad spend (ROAS)Return on ad spend (ROAS) is attributed revenue divided by ad spend on one platform. Formula, worked example, why it matters and where it misleads.Open

Stop the leak.

A free Growth Analysis finds what is broken and ranks it by the revenue it costs you.