Google Ads Question Searches Wasted? Test Before You Block

Most "how to" and "what is" searches don't convert on the click, but that doesn't make them all waste. A question like "how to fix a leaking tap" is usually a do-it-yourselfer who will never call a plumber. "What is SOC 2 compliance" can be the first touch from a buyer who signs a contract six weeks later. Before you block, work out what each question cluster costs per conversion it helped produce, and compare that with your target CPA.
If the number is far above target, or there is no assist at all, block the cluster with phrase match negatives built from the question words. If it lands near target, keep paying for it and cap its budget. When the data is unclear, run a split test: block the questions in half the traffic for four to six weeks and compare total conversions, not just spend.
Question searches are the cheapest-looking waste in a Google Ads account. "How to unclog a drain", "what is a heat pump", "why does my boiler keep turning off". They rarely convert on the click, so the easy move is to add how to and what is as negatives and move on.
Sometimes that is right. Sometimes it cuts off the first click in your best customers' path. This article treats the question as a pricing decision: what does each question cluster cost per conversion it helped produce, and is that price worth paying?
Are question searches wasted spend in Google Ads?
Some are and some aren't. It depends on whether the person asking can solve the problem themselves. If the answer to the question is the product ("how to fix a dripping tap" → a video), the click is almost always waste. If the answer leads to buying something they can't do alone ("what is a SOC 2 audit" → hiring an auditor), the click can be an early step toward a sale.
The search terms report shows you the first kind clearly: clicks, cost, zero conversions. It shows the second kind badly, because the conversion often happens days or weeks later, from a different search, sometimes on a different device. Last-click reporting gives that conversion to the branded search at the end. The question that started it gets nothing.
So "zero conversions" on a question term means two different things:
- No buying intent. The person wanted an answer, got it (or didn't), and left. Blocking costs you nothing.
- Early intent. The person wanted to understand the problem before paying someone to solve it. Blocking saves the click and may cost you the sale.
How much of your spend sits in question terms varies a lot. In accounts we audit, it ranges from a few percent of search term spend in tight, exact-heavy accounts to well over 10% in accounts running broad match or AI Max on topics people research a lot: health, home repair, software, finance. If you are near the low end, block and move on. If you are near the high end, the decision deserves a test.
Which question searches are buying signals
Read the question word, then ask who solves the problem. "How much", "who", "where" and "which" questions usually point toward a purchase. "How to", "what is", "why" and "can I" questions usually point toward self-service or curiosity. That is a starting rule, not a law, so check it against your own terms.
Here is how the same trade splits by question type, using a plumbing account as the example:
| Search term | Question type | Typical intent |
|---|---|---|
| how much does a plumber cost to replace a toilet | Price | Buying, comparing quotes |
| who fixes burst pipes on weekends | Provider | Buying, urgent |
| where to get a water heater installed | Provider | Buying |
| how to replace a toilet flapper | DIY | Self-service, rarely converts |
| what is a backflow preventer | Definition | Research, sometimes converts later |
| why is my water heater making noise | Symptom | Mixed: DIY first, plumber if it gets worse |
| plumber salary | Not a question, but same pattern | Job seeker, never converts |
Three patterns show up in almost every account:
- Price questions convert. "How much does X cost" is one of the most commercial searches there is. A broad negative on
howwould block it. Don't. - DIY questions rarely convert for service businesses. "How to replace a toilet flapper" is someone holding a wrench. They want a video. The click is waste for a plumber. For a retailer selling toilet flappers, the same query might be a buyer.
- Symptom questions sit in the middle. "Why is my water heater making noise" is DIY today and a service call next week if the noise gets worse. These are the ones worth pricing before you decide.
For more on sorting terms by intent in service accounts, see negative keywords for service businesses by intent.
Put a price on the assist
The useful number is cost per assisted conversion: what the question cluster cost, divided by the conversion credit it earned. Compare it with your target CPA. If it is far above, block. If it is close, keep it. Google Ads doesn't give you this number at search term level directly, so you build it.
Read the fractional conversions
If your conversion action uses data-driven attribution, Google Ads splits credit across the ad clicks in a path. A click that started a path can get part of a conversion, which shows up as a fractional number in the conversions column: 0.4, 0.15, 1.3. Those fractions are the closest thing to an assist signal you will find in the search terms report.
To use them:
- Open the search terms report for the last 90 days. Shorter windows miss the slow paths that make question terms worth keeping.
- Filter search terms containing your question words:
how,what,why,can i,is it,does. - Group them by question type, as in the table above. Sum cost and conversions for each group.
- Divide cost by conversions for each group.
Two examples, with round numbers:
- Plumbing, DIY questions. "How to" terms cost $600 in the period and earned 1.5 conversions of credit. That is $400 per conversion. The account's target CPA is $80. Block.
- B2B software, definition questions. "What is" terms cost $900 and earned 3 conversions of credit. That is $300 per conversion against a target CPA of $350. Keep, and probably give them their own budget.
If your conversion action uses last-click attribution, you won't see fractions. Every question term with no direct conversion shows zero, whatever role it played. In that case the fractional method can't help you, and you need the split test in the next section.
A word on what counts as a conversion. If your primary conversion is a newsletter signup or a PDF download, question terms will look great. They are good at producing exactly that. Run this calculation on the conversion you actually get paid for: a booked job, a demo, a purchase. Otherwise you are measuring how well a question gets a free answer.
Isolate the questions in their own campaign
If the question cluster is big enough to matter, move it. Add the question terms you want to keep as phrase or exact keywords in a separate campaign, and add the same terms as negatives in the original campaign. Now the questions have their own budget, their own bids and their own attribution line.
This does three things:
- You can cap spend. Questions stop competing for budget with "emergency plumber near me".
- Smart Bidding in your main campaign stops learning from clicks that were never going to convert on the spot. We cover why that matters in how junk queries skew your bids.
- The attribution reports in Google Ads, which group conversion paths by campaign and keyword, can now show how often the question campaign appears early in a path. They don't break paths down by search term, so a dedicated campaign is the way to get a clean line.
The cost: one more campaign to manage, and a lower-volume campaign that may sit in learning longer. If the question cluster is small, skip this step and decide from the fractional numbers alone.
How to test blocking question searches
When the assist numbers are unclear, test the block instead of arguing about it. Block the question cluster in part of the traffic, leave it running in the rest, and compare total conversions and cost per conversion after four to six weeks. Total is what matters: blocking always lowers spend, so the question is whether conversions dropped with it.
Option 1: a campaign experiment
Google Ads experiments let you split a Search campaign's traffic between the original and a trial version. Add the question negatives to the trial only. Run it at a 50/50 split for at least four weeks, longer if your sales cycle is long. Then compare:
- Conversions (your real, paid-for conversion)
- Cost per conversion
- Conversion value, if you track it
If the trial spent less and converted about the same, the questions were waste. Apply the negatives. If the trial converted noticeably less, the questions were doing work, even if the search terms report never credited them.
Option 2: a before-and-after block
If experiments aren't practical (low volume, or a campaign type they don't fit), block the questions for four weeks and compare with the four weeks before. This is a weaker test. Seasonality, competitor moves and budget changes all land in the same comparison. Use it only when nothing else is available, and check the same weeks last year if you have them.
What to watch during the test
- Branded search volume. If question clicks were introducing people to you, branded searches may dip a few weeks after the block. This is often the earliest sign.
- Conversion lag. A block in week one may only show up in conversions in week four or later. Don't call the test early.
- Volume. If the question cluster produced a handful of clicks a week, no test will give a clear answer. Decide from intent and the fractional numbers, and stop there.
How to block the question searches that fail
Block with phrase match negatives built from the question phrase, not with broad negatives on single question words. how to as a phrase negative blocks "how to replace a toilet flapper" and leaves "how much does a plumber cost" alone. A broad negative on how would block both.
A starting list for a service business, as phrase match negatives:
"how to""how do i""how can i""what is""what does""diy""do it yourself""tutorial""step by step""meaning""definition"
Notes before you paste it:
- Negatives don't cover close variants.
"how to"won't block "how too" or "howto". Add the spellings that actually appear in your search terms report, not every one you can imagine. - Check for conflicts.
"what is"would block "what is the best plumber near me" if that ever shows up. Run the list against your converting terms first. Negative keyword conflicts Google Ads won't flag walks through how. - Avoid broad negatives here. Broad negatives behave unpredictably across match types. A broad negative
how toblocks any search that contains both words in any order, which catches more than you meant. - Keep the commercial questions. Don't add
how much,who,whereorwhichunless your own data says they fail.
For more lists that are safe to reuse, see phrase match negative keyword examples you can copy safely.
Where to put the negatives
- Search. Use a shared negative keyword list so every campaign gets the same treatment, and add exceptions at campaign level if one campaign is your deliberate question campaign.
- Standard Shopping. Accepts negative keywords and shared lists, like Search. Question queries are less common in Shopping, but "how to clean a cast iron pan" can still trigger a skillet ad.
- Performance Max. Accepts up to 10,000 campaign-level negative keywords and does not accept shared lists. You add the question negatives to each PMax campaign one by one. Also note that PMax search term insights group queries into themes, so a question cluster may be hidden inside a category label; see why search term insights are not queries.
- Display, Video and App. These don't report search terms, so there is nothing to read and nothing to block on this basis.
If you run broad match or AI Max, expect question searches to keep appearing in new forms. "How to" becomes "ways to", "guide to", "is it possible to". The controls that work with AI Max search terms apply here too.
Where this gets hard
The method above works when you have enough volume and your conversions are tracked inside Google Ads. Several situations break it, and it is better to know that before you rely on the result.
- Offline sales that aren't imported. If your real conversion is a signed contract in a CRM and you don't import it into Google Ads, every calculation in this article runs on the wrong number. Fractional credit, experiments and PPCleaner's judgement all depend on what Google Ads can see.
- Long sales cycles. If a B2B deal takes four months, a 90-day window and a six-week experiment both miss most of the effect. You may have to decide from intent and accept some uncertainty.
- Hidden search terms. Part of the spend in the search terms report shows no query at all. Question searches are often long and rare, so they are more likely to fall below the threshold where Google shows them. Your question cluster is probably bigger than what you can see. Search terms hidden by privacy covers what you can still do.
- Small numbers. A question term with 3 clicks and 0.2 conversions tells you very little on its own. Group terms before you judge them. One rare question is noise; fifty of them are a pattern, as low-volume terms that add up shows.
- Content strategy, not ads. If you want to answer "what is" questions to build an audience, an organic article usually does it more cheaply than paying per click. Paying for informational clicks is a choice to make on purpose, with its own budget, not a default.
What PPCleaner does with question searches
PPCleaner reads the search terms your account paid for every night, across Search, Shopping and Performance Max, and judges which ones were never going to convert. A DIY question with no conversions, like "how to replace a toilet flapper" in a plumbing account, is the kind of term it blocks. A price question like "how much does a plumber cost" is not.
Three things it does not do:
- It never blocks a term that converted in the last 90 days. If a question term earned conversions in that window, it stays live.
- It doesn't see conversions Google Ads doesn't see. If your assist value lives in a CRM and isn't imported, PPCleaner judges the term on the data it has, like any tool that reads the account.
- It doesn't change budgets, bids, ads or keywords. If you decide to keep question searches in a dedicated, capped campaign, building that campaign is your job.
If you run a split test and find that a blocked question cluster was helping after all, every block can be undone in one click for 90 days. That makes the decision cheaper to get wrong, which is the point of testing it.
For the wider routine these checks fit into, see how to analyze a search terms report in 20 minutes. And if your CPA is climbing and you suspect questions are part of it, start with checking search terms when CPA is too high.
Frequently asked
Should I add "how to" as a negative keyword in Google Ads?
For most service businesses, yes, as a phrase match negative. "How to" searches are usually people who want to do the job themselves. Check first that no converting term in your account contains the phrase, and don't add a broad negative on "how", which would also block high-intent searches like "how much does a plumber cost".
How can I tell if a question search assisted a conversion?
If your conversion action uses data-driven attribution, look for fractional conversions (0.3, 0.5) on question terms in the search terms report over a 90-day window. That is partial credit for being part of a path. With last-click attribution you won't see this, so the reliable method is a split test that blocks the questions in half the traffic and compares total conversions.
Are "how much" searches informational or commercial?
Mostly commercial. Someone asking "how much does it cost to replace a water heater" is usually comparing prices before buying. Treat price questions separately from "how to" and "what is" questions, and don't block them unless your own conversion data says they fail.
Can I block question searches in Performance Max?
Yes, with campaign-level negative keywords. Performance Max accepts up to 10,000 of them per campaign, but it doesn't accept shared negative keyword lists, so you add the question negatives to each PMax campaign separately.
How long should a test of blocking question searches run?
At least four weeks, and longer if your sales cycle is long. Question clicks often influence conversions that happen weeks later, so a short test can make blocking look free when it isn't. Compare total conversions and cost per conversion, not just spend.
Clean your search terms. Save time and money.
Every night, PPCleaner reviews the searches your campaigns actually paid for, groups them by intent, and excludes the ones that never convert. No spend threshold to wait for, nothing touched but negatives — and one click undoes any of it.
Start 7 days freeSearch, Shopping and Performance Max · Cancel in one click
Related articles

Google Ads Reviews Searches Wasted? Measure Before Blocking
The usual advice is to add "reviews" and "complaints" as negatives. Some of these searches come from people one step from buying, so measure what they assist before you block them.

Misspelled Search Terms in Google Ads and Which to Block
Close variants send you typos of the queries you want and typos of the ones you don't. Most misspellings are harmless and convert fine. A few are a different search altogether, and your negative list won't catch them.

Google Ads Search Terms Review Checklist in 15 Minutes
A weekly search terms review only works if every term gets the same decision every time. Here is a 15-minute checklist and a five-question decision tree for block, keep or add as keyword, including the borderline cases.
