The costliest Performance Max mistakes for small businesses are counting clicks as conversions, paying for searches of your own name, a budget too small to learn, final URL expansion sending buyers to the wrong pages, no negative keywords or exclusions, letting Google make your video, tinkering within six weeks, one asset group for everything, and treating every sale as a new customer.
Each is expensive for the same reason. Performance Max spends one budget across Search, YouTube, Display, Discover, Gmail and Maps, choosing where the money goes itself. When an input is wrong, the campaign doesn't stall; it spends confidently in the wrong direction, and the blended reporting can make that look like success. Google has added far more visibility since 2025, including a search terms report, a channel performance report and up to 10,000 negative keywords per campaign, so most of these mistakes are now fixable if you know where to look.
What Performance Max does with your money
You give a Performance Max campaign a goal (conversions or conversion value), a budget, a set of assets (headlines, descriptions, images, videos, logos and optionally a product feed) and audience signals, which are hints about the people likely to buy. Google's AI then assembles ads for each channel, bids in each auction using Smart Bidding, and moves budget towards whatever produces the conversions you've defined. You don't choose placements, and you only partly choose searches.
For a small business that has one clear consequence: your inputs are almost the only levers you have. Every mistake below is an input problem.
Mistake 1: Counting clicks and page views as conversions
What goes wrong. Performance Max optimises for your primary conversion actions. If those include cheap, frequent actions such as page views, "add to basket", clicks on your phone number or requests for directions, the campaign will find people who do those things, and there are far more of them than buyers.
How it shows up. A very low cost per conversion that doesn't match your takings. In the Goals section, several actions marked as primary. Conversions reported on days when the till was quiet.
An illustration. An optician launched Performance Max with "click to call", "directions" and "booking form start" as primary conversions. After a month it reported 210 conversions at $2.40 each. The diary showed 11 bookings that came from ads. Most "conversions" were people on mobile tapping the map pin or opening the form and leaving.
The fix. Make only real outcomes primary: completed purchases, completed bookings, qualified enquiry forms. Keep the smaller actions as secondary so you can still see them. For phone bookings, track calls of a meaningful length, or import bookings from your system as offline conversions. Where orders differ in value, send the value.
Check it worked. After the change, reported conversions fall and cost per conversion rises, often sharply. That's the point. Compare reported conversions with your own records monthly; the gap should narrow. The full setup is in conversion tracking before AI ads.
Mistake 2: Paying Performance Max for searches of your own name
What goes wrong. People who search your business name were probably going to find you anyway. Performance Max is very good at winning those searches, because they convert well, and then counting them as campaign success.
How it shows up. A return on ad spend that looks excellent from day one. In the Performance Max search terms report, your brand name and its misspellings near the top. Organic clicks on brand searches falling in Search Console as paid ones rise.
An illustration. A pet shop's Performance Max campaign reported a return of eight times ad spend in its first month. The search terms report showed that 40% of its Search spend went on the shop's own name, variations such as "[shop name] opening times", and a misspelling. Excluding brand terms dropped the reported return to about three times, which was the true figure for customers the ads actually brought in.
The fix. Add a brand list and apply brand exclusions to the Performance Max campaign. If competitors bid on your name, run a small, separate Search campaign for brand terms with a low budget, so you control what that costs. Keep an eye on the search terms report for new variations.
Check it worked. Brand terms disappear from the Performance Max search terms report within a few days, and the brand Search campaign costs a small fraction of what Performance Max was spending on those searches.
Mistake 3: A budget too small for the bid strategy to learn
What goes wrong. Smart Bidding learns from conversions. If the budget only buys a handful a month, the system has almost nothing to learn from, so it keeps exploring expensive placements and results swing wildly.
How it shows up. A campaign that stays "Limited" or bounces between great and terrible weeks. Very few conversions in the conversions column even after a month.
A quick sum. A dry cleaner wants commercial accounts and sets $10 a day. If a qualified enquiry costs around $45, that's $300 a month buying six or seven enquiries. Spread across six channels, the campaign sees perhaps one or two useful signals a week. That's not enough to learn from.
The fix. Work backwards from what you can afford. If your budget divided by a realistic cost per conversion gives fewer than about 15 to 20 conversions a month, either raise the budget for a defined test, switch to a bid strategy based on a more frequent action that genuinely predicts sales, or start with a tightly targeted Search campaign instead and add Performance Max later. Don't set an aggressive target cost per acquisition on top of a tiny budget; it restricts the campaign further.
Check it worked. The campaign spends its budget most days and conversion numbers become steadier week to week. What to spend before judging is covered in budgets for AI-run ads.
Mistake 4: Letting final URL expansion send buyers to the wrong page
What goes wrong. Final URL expansion is on by default in Performance Max. It lets Google replace your chosen landing page with another page on your site that it predicts is more relevant, and generate headlines to match. That can help. It can also send paying clicks to your careers page, a tutorial from three years ago, or a product you've stopped selling.
How it shows up. In landing page reports, pages you never chose receiving ad traffic. Spend on pages that don't convert. Ad text you didn't write describing things you don't offer.
An illustration. A garden centre found that a spring campaign had sent about a fifth of its clicks to a page for a Christmas grotto event from the previous year, because the page still ranked well and mentioned "family days out".
The fix. Decide deliberately. Either switch final URL expansion off and send traffic only to the pages you've chosen, or leave it on and add URL exclusions for pages that shouldn't receive ads: jobs, about, policy pages, old events, out-of-stock categories. You can only add URL exclusions while expansion is on, and your campaign's own final URL can't be excluded that way.
Check it worked. Review the landing page report two weeks later. Only commercial pages should be receiving ad traffic.
Mistake 5: Skipping negative keywords and placement exclusions
What goes wrong. Performance Max will match your ads to searches it judges related. Without negatives, a pet shop can pay for "pet shop jobs", "free kittens" and "how to become a dog groomer". Without placement exclusions, your display and video ads can appear in apps and on sites that bring nothing but accidental clicks.
How it shows up. Irrelevant queries in the search terms report. In the channel performance report, Display spending a large share of budget with few conversions.
An illustration. In its first month, an illustrative pet shop's search terms included "pet shop near the hospital job vacancy", "pet insurance quotes" and "dog breeds that don't shed". About $140 of a $900 month went on searches from people who were never going to buy food or toys.
The fix. Since 2025, Performance Max supports campaign-level negative keywords and negative keyword lists, up to 10,000 per campaign. They apply only to Search and Shopping inventory, not to YouTube, Display, Gmail or Discover. Start with a standard list (jobs, careers, vacancy, free, salary, course, training, how to become, DIY where relevant) and add to it monthly from the search terms report. For display and video, use account-level placement exclusions for apps and sites you don't want.
Check it worked. Irrelevant terms stop appearing, and the share of spend on low-converting channels falls in the channel performance report.
Mistake 6: Letting Google make your video for you
What goes wrong. If an asset group has no video, Performance Max may generate videos automatically from your images, text and product listings, in horizontal and vertical formats, and run them on YouTube and Shorts. Separately, "video enhancements" can create altered versions of videos you did upload.
How it shows up. In the asset list, videos marked as automatically created. A slideshow of your product photos with stock music running as your brand's face on YouTube.
An illustration. A nail salon's auto-generated video stitched together its price list graphic, a logo and two close-up photos, then cut the price list off mid-line. It ran for three weeks before anyone watched it.
The fix. Upload your own video. Google's optimisation tips recommend at least one video asset per asset group; a 15 to 30 second phone video filmed in good light is enough for a small business. To stop video enhancements, go to campaign settings, then Asset optimisation, then Video, and untick Enhancement. Uploading your own videos is what switches off auto-generated ones.
What a usable phone video looks like. For the nail salon, a 20-second vertical clip was enough: two seconds of the shop front with the name visible, eight seconds of a technician finishing a gel manicure in close-up, five seconds of the finished nails in daylight, and a closing frame with the price from $38 and "Book online". Film a horizontal version of the same shots at the same time, so Google has both formats and doesn't generate its own to fill the gap.
Check it worked. The asset list shows only your uploaded videos, and no newly auto-created entries appear.
Mistake 7: Judging and tinkering inside the first six weeks
What goes wrong. Google recommends running a new Performance Max campaign for at least six weeks and avoiding frequent changes to budget, bid strategy or status in that period, because changes can reset learning. Owners who check daily see a bad Tuesday, cut the budget, add a target, change the images and pause it for the weekend. Each change restarts the learning.
How it shows up. A change history full of small edits. A campaign that never settles.
The fix. Decide before launch what you'll judge and when: for example, cost per real conversion over weeks three to six, compared with your break-even figure. Allow only two kinds of change in the first six weeks: adding negative keywords and exclusions, and fixing genuine errors such as a broken landing page. Log every change with a date.
A change log can be as simple as a note on your phone. An illustrative one from week two: "Day 9: added 12 negatives from search terms (jobs, free, grooming course). Day 11: fixed broken link on the cat food landing page. Nothing else." When results shift in week four, the log tells you whether anything you did could explain it.
A quick sum for break-even. If a pet shop's average order is $48 at a 35% gross margin, each order earns about $16.80 before ad costs. A cost per order above that loses money on the first sale, though repeat customers can justify a higher figure. Knowing the number stops panic decisions in week one.
Mistake 8: One asset group for every product and service
What goes wrong. An asset group is a set of headlines, images, videos and signals built around one theme. Put everything in one group and Google mixes pieces that don't belong together: a headline about children's eye tests with a photo of designer sunglasses; a puppy food image with a headline about aquarium filters.
How it shows up. Ad previews that combine unrelated offers. Poor ad strength. Conversions concentrated on one product while the others get nothing.
An illustration. An optician split one asset group into three: eye tests (landing on the booking page), frames and sunglasses (landing on the frames range), and contact lenses (landing on the lens subscription page). Each got its own headlines, images and search themes. Search themes, which are phrases you give Google as hints about relevant searches, can now run to 50 per asset group.
The fix. One asset group per distinct offer or customer need, each with matching landing pages. Google's optimisation tips suggest at least 15 headlines, 5 descriptions, 7 images (3 landscape, 3 square, 1 portrait) and at least one video per group, aiming for "Good" or "Excellent" ad strength. For a small business, three well-built groups beat eight thin ones.
Check it worked. Asset-level reporting, which now shows impressions, clicks, cost and conversions per asset, lets you see which pieces earn their place. Replace the weakest every month or two.
Mistake 9: Treating every conversion as a new customer
What goes wrong. Without guidance, Performance Max values a sale from a regular customer the same as one from a first-time buyer. Regulars convert easily, so the campaign drifts towards retargeting people who already buy from you.
How it shows up. Strong reported results but no growth in new customers in your own records. Heavy spend on remarketing-style placements.
The fix. Use the customer acquisition goal, which can bid higher for new customers or bid only for new customers, and give Google a customer list so it can tell them apart. Google also reports new customer acquisition cost in campaign reporting when this goal is used. If winning back lapsed customers matters more, retention goals let you bid more for customers who haven't bought for a while.
Check it worked. Compare new customers per month in your own shop data before and after. That's the number that matters, not the platform's conversion count.
A pet shop's first three months, before and after the fixes
An illustrative independent pet shop with an online store, a $900 monthly budget and a product feed of about 600 items.
Month 1, default setup. One asset group, no video, final URL expansion on with no exclusions, "add to basket" as a primary conversion, no brand exclusions, no negatives. Reported: 380 conversions, return on ad spend of 9 times. Shop records: 41 orders attributed to ads, 17 of them from existing customers searching the shop's name.
Changes at the start of month 2. Purchases only as primary conversion, with order values. Brand exclusions and a small brand Search campaign at $60 a month. A 40-term negative list. Three asset groups (dog, cat, small animals) with a phone-filmed video each. URL exclusions for the jobs page and the grooming price list. New customer acquisition goal set to bid higher for new customers.
Months 2 and 3. Reported conversions fell to about 70 a month and reported return to about 3.5 times. Shop records: 64 and then 78 ad-attributed orders a month, with new customers rising from 24 in month 1 to 51 in month 3. Average order $46. At a 35% margin, month 3's ad-driven orders earned roughly $1,250 in gross profit against $900 of ad spend, before counting repeat purchases.
The reported figures got "worse" and the business got better. That pattern is the clearest sign the fixes were right.
A pre-launch checklist
- Primary conversions are real outcomes only, with values where values differ.
- Offline or phone bookings are imported or tracked with a minimum call length.
- A brand list is created and brand exclusions applied.
- A starting negative keyword list is attached.
- Account-level placement exclusions are set for apps and sites you don't want.
- Final URL expansion is switched off, or on with URL exclusions for non-commercial pages.
- Each asset group covers one theme, with matching landing pages and search themes.
- Every asset group has your own video; video enhancements are off if you don't want them.
- The customer acquisition goal is set if new customers are the aim, with a customer list uploaded.
- A break-even cost per conversion is written down, and the review date is at least six weeks away.
A 20-minute monthly review
- Search terms report (5 minutes). Add irrelevant terms as negatives; check brand terms aren't creeping back.
- Channel performance report (4 minutes). Which channels spend and which convert? Look for a channel taking a large share with little result.
- Landing pages (3 minutes). Any pages receiving traffic that shouldn't?
- Assets (4 minutes). Replace the lowest-performing images or headlines; check for auto-created assets you didn't approve.
- Reality check (4 minutes). Compare reported conversions and new customers with your own records. Write the gap down each month.
If the reported numbers and your shop's numbers keep diverging, suspect tracking before you blame the campaign. If the landing page itself is the weak link, building a landing page that converts is the better place to spend the next hour.
When Performance Max isn't the right campaign at all
Performance Max suits businesses with a product feed, reliable purchase tracking and enough budget to produce steady conversions. It's a poorer fit when:
- You're a local service business with a budget under a few hundred dollars a month. A Search campaign on a short list of high-intent phrases gives you more control per dollar.
- You can't track the real outcome, for example because most bookings happen by phone and you have no way to import them.
- You have no images or video of your own and no time to make any.
- Your leads need qualifying and cheap form fills would swamp you.
In those cases, start with Search, get tracking right, and revisit Performance Max when you have a few months of reliable conversion data. The wider picture of how Google's automation decides bids and placements is in how Meta and Google use AI to run your ads, and if you'd rather keep the campaigns in-house, managing Google Ads without an agency sets out a routine one person can keep up.
Performance Max questions from small advertisers
What is the minimum budget for Performance Max?
Google doesn't set a practical minimum, but the campaign needs enough conversions to learn from. Work backwards: divide your monthly budget by your likely cost per conversion. If the answer is under about 15-20 conversions a month, learning will be slow and noisy, and a well-built Search campaign may use the money better until volume grows.
Can I see which search terms Performance Max bought?
Yes. Google added a full search terms report to Performance Max in 2025, with the same level of detail as Search and Shopping campaigns, alongside a channel performance report that splits results across Search, YouTube, Display, Discover, Gmail and Maps. Check both monthly and turn what you find into negative keywords and asset changes.
Should Performance Max run alongside a Search campaign?
It can, and often should for brand terms and your most important keywords. Where a Search campaign has an exact-match keyword identical to the search, Google generally prefers the Search campaign. Keep the structure simple: a small Search campaign for brand and core terms, Performance Max for broader reach and shopping, and brand exclusions on Performance Max.
Is Performance Max suitable for a service business without a product feed?
It can work for lead generation if you track real outcomes such as completed bookings or qualified enquiries and import offline results. Without good tracking it tends to fill forms with low-quality leads. Many small service businesses do better starting with Search and adding Performance Max once they have reliable conversion data.
Further reads
- How to Keep Your Brand Safe When AI Places Your Ads — Exclusions and checks that keep automated placements brand-safe.
- 12 AI Marketing Mistakes Small Businesses Make (and the Fixes) — Marketing mistakes beyond ads that undercut the same budget.
- How to Test Ad Variations With AI on a Small Budget — Test headlines and images without wasting a small budget.
- Should a Small Business Use Meta Advantage+ AI Campaigns? — Meta's equivalent automation, and whether it suits you.
- How to Use AI to Find Out Why Website Visitors Don't Buy — Fix the landing page before blaming the campaign.
- How to Measure Whether AI Is Improving Your Marketing Results — Judge ads on your own sales data, not the platform's.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Google Ads Help pages on Performance Max, Final URL expansion and URL exclusions, Performance Max video automation, optimisation tips for Performance Max, the Performance Max search terms and channel performance reports, and Google Ads 2025 highlights (negative keyword limit, search themes, customer acquisition and retention goals).