Look at a Google Ads account's monthly report and the click count can look perfectly healthy while conversions stay thin. The gap is rarely down to bidding strategy or ad copy; more often it is that the campaign has been showing up on searches it should never have reached. Budget on the search network drains not because the wrong keyword was bid on, but because the right keyword was matched to a use of it that never carried buying intent. The tool that stops this is the negative keyword list, and it only works once it stops being a box ticked at setup and forgotten afterwards.
Why match types leave the door open
Broad match targets a word but also grants itself licence to enter any search it judges related to that word. This is where the system can confuse “related” with “same intent”: someone searching for a service and someone searching for general information about that service often fall into the same word cluster as far as the platform is concerned. Phrase and exact match are assumed to narrow that risk, but the platform's own “close variant” interpretation steps in regardless — synonyms, singular-plural shifts, reordered words and sometimes additional words of debatable relevance can all still be grouped under the exact match label. No match type on its own guarantees an unwanted search will never trigger the ad; the negative list is what fills the gap the match type leaves behind.
What only the search terms report can show
The keyword list shows what the account is targeting, not what the person actually typed into the box. The gap between the two lives in one specific report: the search terms report. The raw query that actually triggered the campaign can only be seen there, never in the keyword list itself. An account manager glancing only at the keywords added can assume things are fine, when the real question is which actual searches those keywords were matched against. In an account where this report is not opened regularly, where the budget has gone only becomes clear months later, once conversion numbers dip — and by then the money already spent cannot be recovered.
The classic budget-draining groups
Certain search patterns repeat across almost every sector and almost never carry purchase intent:
- “Free” — brings in people looking for a no-cost equivalent, not a paying customer.
- “How to” — points to someone trying to solve the problem themselves, not yet considering a paid service.
- “Jobs” / “hiring” / “vacancy” — brings people wanting to work at the business, not buy from it.
- “Complaint” / “complaints” — usually searched alongside the brand name, carrying an intent to air a grievance rather than to buy.
- “Second-hand” / “used” — a demand type flatly mismatched to a business selling new, full-price services or products.
- Student and homework-related searches — bring people researching a topic to finish an assignment, not to buy anything about it.
- Competitor brand names — this is a separate discussion; bidding on a competitor's name is not against platform policy, but whether to add it to the negative list is a deliberate strategic choice, not something that should happen by accident.
What these terms have in common is that none of them are malicious or meaningless in themselves. The problem is not the word — it is the need of the person searching it at that moment.
Why jobseeker and student traffic is particularly damaging
Even a search that ends with nothing bought, after a few seconds on the site, still leaves some non-zero chance that the same visitor becomes a customer later. Jobseekers and students researching an assignment sit in a different category: the moment they find the vacancy page or the background information they came for, their purpose is complete, and the chance of that same person ever buying the service is structurally zero. Worse, this kind of search tends to repeat — the same person can run near-identical queries within a short span, spending the same budget on the same unproductive click again and again. Surface metrics like click-through rate or time on site can look perfectly fine on this traffic; the real damage shows up on the conversion side, in demand that was never there to begin with.
The right level for the negative list, and the logic of a shared list
A negative keyword can be added at campaign level, at ad group level, or across the whole account. An exception specific to a single campaign should stay at campaign level; but a word like “jobs” or “free”, relevant to almost every campaign in the account, is better collected once in a shared list and attached to the relevant campaigns, rather than typed into each one by hand. This looks like a minor technical detail, but it decides the outcome: without a shared list, every newly launched campaign has to accumulate the same well-known problem words from zero, and the days in between count as lost budget.
The risk of overcorrecting
A negative list feels safer the longer it grows, and that feeling itself carries a trap. Adding a broad word to the negative list without thinking it through can push out real customers along with the noise. “Price” is a good example: someone searching that word is often the closest to buying, not someone still gathering information but someone at the decision stage; negating that word wholesale removes exactly the audience the campaign should be targeting. The right approach is to look at what searches actually bring a word in before negating it, and base the decision on that evidence — a word simply sounding “risky” is not reason enough on its own.
Ongoing maintenance, not a one-off task
A negative list is not finished once it is built. Seasonal search behaviour shifts, adding a new product or service can each time spawn a fresh cluster of irrelevant searches, and even changes competitors make to their own campaigns can affect which searches end up pointing at which business. Reviewing the search terms report at regular intervals is the only way to keep the list current; skip that step and the list stays frozen as it was at setup, drifting further from the account's real search behaviour over time.
Why geographic negatives are a separate need
A word-based negative list still needs completing with place names. For a business serving only a defined area, searches carrying the name of a city or country it does not serve drain budget just like an irrelevant word does. This is a separate matter from the campaign's location targeting setting; when targeting is left broad, or the platform's interest-based interpretation is in play, searches naming places outside the service area can still get through. Adding those place names to the negative list should be treated as the counterpart to word-based work, not a separate item that gets forgotten.
Negative keywords, in the end, are the least visible but most decisive layer of a search campaign. Just as a properly built measurement setup shows which search turned into real demand, the negative list shows which search should never have been allowed to show up at all. Working together, the two send budget where actual demand sits rather than where the search happened to occur — and that is the ground on which later steps, such as remarketing, can be built with any confidence.