E-commerce Google Ads Search Terms That Drain Budget

Retail search terms are messy in a way lead-gen search terms are not. A plumber's account wastes money on "plumber salary" and "how to unblock a sink" — obviously wrong, easy to spot. An online shop selling dishwashers wastes money on "bosch dishwasher door seal", which is a product, has buying intent, mentions a brand you stock, and will never convert because you don't sell seals.

That is the difference. In e-commerce, most wasted spend looks like a sale.

Research terms: real intent, wrong moment

People use Google as a shortlist tool before they use it as a shop. Those searches have high volume, decent click-through rates, and conversion rates that sit far below the account average.

They look like this:

• best wireless headphones 2026
• sony wh-1000xm5 vs bose quietcomfort
• is dyson airwrap worth it
• air fryer reviews
• cheapest washing machine uk (if you ship to one country only, the geography word matters too)

Shopping and Performance Max pick these up constantly, because the feed matches on product titles and a product title contains the brand and the model number. "sony wh-1000xm5 vs bose quietcomfort" matches your Sony listing perfectly on a string level and perfectly badly on an intent level.

The honest part: blocking research terms is not free. Some of these people come back. If you run a long consideration-cycle category — furniture, appliances, bikes — a comparison click today is sometimes an email signup and a purchase in three weeks. Before you block the pattern wholesale, look at assisted conversions and at view-through behaviour, not just last click.

What is almost always safe to block, though, is the pure-information half: review, reviews, vs, versus, comparison, is x worth it, pros and cons, which is better, test, ranking, top 10, buying guide. If a term contains "vs" and has 40 clicks and no conversions across six months, it is not a shortlist — it is a hobby.

Set these as phrase negatives, not broad. Phrase "vs" will not catch tv stand the way a careless broad negative might catch strange things across match types. Broad negatives behave unpredictably across match types, which is exactly what you don't want on a word as short as "test". There is more on that trap in negative keyword match types don't work like positives.

Spare parts, accessories and consumables you don't sell

This is the single biggest category of silent waste in appliance, electronics, tool and automotive retail. A shopper types the brand, the model and a component. Google reads brand plus model, matches your product, and charges you for a click that lands on a page selling the whole machine for $600 when the visitor wanted an $8 filter.

Typical terms:

• bosch dishwasher door seal
• dyson v11 battery replacement
• nespresso vertuo drip tray
• samsung tv remote replacement
• karcher k5 spare hose
• ninja blender lid replacement

The giveaway modifiers are consistent across categories: spare, spares, replacement, part, parts, filter, seal, gasket, belt, hose, battery, charger, remote, lid, blade, bag, refill, cartridge, fuse, bulb.

Two things make this harder than it sounds.

First, some of those words are your business. If you sell vacuum bags, bag is a core term, not a negative. The list has to be built from your catalogue, not copied from a blog post — including this one. Go through the modifier list above and delete every word you actually stock before you add anything.

Second, the same query can be good in one campaign and bad in another. dyson v11 battery is waste for a campaign selling complete vacuums and a core term for a campaign selling parts. That's an argument for campaign-level or ad-group-level negatives rather than one account-wide shared list. On Search and Standard Shopping you can use shared lists; Performance Max does not accept them and takes campaign-level negatives only, up to 10,000 per campaign. That structural difference is covered in negative keywords in Performance Max and the 10,000 cap.

The accessory trap in reverse

The mirror image also costs money: you sell the accessory, you don't sell the machine, and you pay for playstation 5 console because your controller listing mentions PlayStation 5. Shopping feeds leak in both directions. Audit your own titles — a product title stuffed with "compatible with iPhone 15, 14, 13, 12" is an invitation to match on every one of those devices.

After-sales traffic: people who already bought

Someone who owns the product searches support questions, and your product page is the closest thing Google has. You pay to serve a customer you already have.

The pattern:

• ikea malm instructions pdf
• how to reset bosch washing machine
• nike return label
• track my order + brand
• warranty claim + brand
• [brand] customer service number
• delete my account

If you have a help centre, some of this traffic is useful in a soft way — the customer gets served, you don't lose them. But you are paying search ad prices for a support ticket. In the accounts we audit, this bucket is usually small in percentage terms and surprisingly steady: it never stops, and nobody ever notices it, because each individual term has three clicks a month and $4 of spend. Fifty such terms is $200 a month.

Block the clear support vocabulary: manual, instructions, pdf, how to reset, how to clean, troubleshooting, error code, fault, repair, fix, return label, returns policy, track order, tracking, warranty claim, customer service, complaint, phone number.

One caveat worth stating: repair and fix can be commercial if you sell replacement kits. And if your brand term pulls [brand] customer service, blocking it means an existing customer may land on a competitor's ad. Most retailers accept that trade; a few, with high repeat-purchase rates, do not. Decide on purpose rather than by default.

Free, DIY and second-hand intent

Three modifier families that almost never convert in retail, and almost always get clicks.

Free. free shipping code, discount code, promo code, voucher, coupon, free samples, giveaway. Coupon searches are a genuine debate — a coupon searcher is sometimes a buyer in your checkout who opened a new tab. But if you are paying for [competitor] discount code or generic free shipping code, you are buying affiliate traffic at auction prices.

DIY and second hand. diy, homemade, make your own, how to make, used, second hand, refurbished, ex display, pre owned, rental, hire, wholesale, bulk, dropshipping, supplier, distributor. If you sell new goods retail, every one of those is a different business model.

Marketplaces and retailers. amazon, ebay, walmart, temu, aliexpress, argos, plus the local equivalents in your market. Someone searching air fryer amazon has already chosen where they are buying. Competitor names generally need a decision rather than a rule — that is a longer argument, made in competitor searches in Google Ads, block or keep. Marketplace names are easier: the shopper named a destination, and it isn't you.

Where these terms actually come from

Four sources, and knowing which one you're dealing with changes the fix.

Broad match keywords. Broad match is the most common source of research terms in Search campaigns. It reads intent generously and in e-commerce "generously" means comparison articles and category browsing. Whether that trade is worth it in current accounts is argued through in are broad match keywords worth it in 2026.

Shopping feed matching. Standard Shopping has no keywords at all. Your product titles, descriptions and attributes are the match surface. That means you cannot prevent a match by editing keywords — negatives are the only lever, and they behave differently than they do on Search. The mechanics are in Shopping campaign negative keywords work differently.

Performance Max. PMax blends feed matching, audience signals and search themes. It will find search terms you would never have bid on. It also reports them in a thinner form than Search does — see what the PMax search terms report actually shows for exactly how thin.

Your own catalogue. Long product titles crammed with compatibility lists, model numbers and synonyms widen your match surface every time you add a product. A feed problem is not fixable with negatives alone.

The practical consequence: in a typical retail account running Search plus Shopping plus PMax, the same wasted term can arrive through three doors. Blocking it in one campaign moves the spend to another. Negatives have to be applied across all three campaign types on the same night, or you spend a month playing whack-a-mole with your own account. Where Performance Max wasted spend hides covers the PMax side of that shuffle.

Reading the report without drowning in it

A retail account with 5,000 SKUs produces thousands of unique search terms a month, most of them seen once or twice. You cannot review that list term by term. Review it by pattern.

Sort by cost, not by impressions

Open Search terms for the last 30 to 90 days, filter to conversions = 0, sort by cost descending. The top 30 rows are where decisions live. Every one of them is either (a) a term worth keeping that just hasn't converted yet, (b) a pattern you should block, or (c) a landing page problem. Decide which, row by row, and you will have spent twenty minutes well.

Then search for modifiers, not terms

Use the search box inside the report. Type review. Type spare. Type manual. Type free. Each search gives you a cluster, and the cluster tells you whether the pattern is worth a phrase negative. One term with $6 of spend is noise. Forty terms containing "review" with $340 of spend between them is a negative keyword.

This is also how you catch what the report hides. Low-volume terms are aggregated away for privacy reasons; the spend still happened. Your clicks column will not add up to your campaign clicks, and the gap is almost always made of hundreds of one-click junk queries. Google Ads search terms report, and what it hides goes through the size of that gap and what you can still infer from it.

Watch for seasonal arrivals

Retail search language changes every quarter. black friday arrives in October. christmas delivery cutoff arrives in December and becomes an after-sales query by the 22nd. back to school arrives in August. Last year's negative list will not contain this year's variants, and a pattern that was dormant for eleven months can spend $800 in three weeks. This is one reason monthly reviews are too slow in retail — the argument for a tighter cadence is in how often should you check search terms in Google Ads.

Match types and the one-word problem

The thing that separates a good retail negative list from a damaging one is that half these terms sit one word away from money.

iphone case converts. iphone case diy does not. iphone 15 case converts. iphone 15 case template pdf does not.

If you add case as a negative, you lose the category. If you add broad negative diy, you block every query containing that token in any order, and broad negatives behave unpredictably across match types — you can lose more than you intended with no visible warning in the interface, because blocked impressions don't show up anywhere.

Three working rules for retail:

1. Phrase negatives for modifier families. "how to make", "replacement part", "return label". The multi-word phrase protects you from accidental collisions.
2. Exact negatives for single ambiguous words. If you want to block the standalone query manual but keep manual coffee grinder, exact is the only safe choice.
3. Broad negatives only for words that can never be commercial in your catalogue. salary, jobs, wikipedia, meaning, pronunciation. And if you also hire, see job seekers clicking Google Ads before you block the recruitment words.

Also remember Google ignores nothing about word order in phrase negatives but does normalise plurals and misspellings inconsistently. review does not reliably block reviews. Add both. Add the obvious misspellings too — in retail, model numbers get typed wrong constantly, and wh1000xm5, wh-1000xm5 and wh 1000 xm5 are three different strings.

If blocking starts to cost you volume, the symptoms are specific and visible before the revenue drop: impression share falling in campaigns you didn't touch, a sudden drop in new search terms. Six signs your negative keywords are hurting performance lists what to watch.

What this is worth, roughly

Be careful with waste numbers. The honest way to size this is to take your own search terms report for the last 90 days, filter to zero conversions, and read the cost column — then subtract the terms you would genuinely keep. What's left is the real figure, and it is always smaller than the scary one and larger than the comfortable one. The method, including the parts people cheat on, is in how to calculate wasted ad spend honestly.

In the retail accounts we audit, the four patterns above — research, parts, after-sales, free intent — typically account for a meaningful minority of non-converting spend rather than the majority. The majority is usually terms that are genuinely on-topic and simply didn't convert, which is a landing page, price or stock problem, not a negative keyword problem. Blocking those would be a mistake. If your campaigns are spending with nothing to show, work through the causes in order rather than reaching for negatives first: Google Ads spending money, no conversions.

Keeping it maintained

A retail negative list is never finished. New SKUs widen the match surface. Seasons bring new vocabulary. Google changes how broad match reads intent, and terms you blocked a year ago stop appearing while new ones take their place.

Three things make maintenance survivable:

Review weekly, not monthly. In retail, a month of a new bad pattern is a month of spend. Fifteen minutes a week on the cost-sorted zero-conversion list beats two hours a quarter.

Document why you blocked. Six months later, nobody remembers whether refurbished was blocked on purpose or by accident. If you manage several shops, this matters more than the list itself — see managing negative keywords across client accounts.

Keep the undo path open. Every negative is a hypothesis. Some of them are wrong. A list you cannot reverse quickly is a list you will be reluctant to edit, and a list nobody edits goes stale.

That is the problem PPCleaner solves mechanically: it reads the terms your account actually paid for across Search, Shopping and Performance Max, adds the ones that were never going to convert as negatives every night, never touches a term that converted in the last 90 days, and keeps every block reversible in one click for 90 days. It doesn't touch budgets, bids, ads or keywords. If you'd rather do the first pass by hand, start with how to find wasted ad spend in Google Ads and work the cost column.