What you'll learn in this article
- Why an SEO list and an Ads list answer two different questions, and what that costs when you confuse them
- The three metrics that invert completely when you move from ranking to bidding
- The four mistakes I find in almost every inherited account whose list came from an SEO export
- How I read commercial intent from the query itself, and the inference that decides whether a term gets budget
- The one place SEO research genuinely helps an Ads account, and how to use it without importing its logic
Most of the broken keyword lists I inherit were not built carelessly. They were built well, for the wrong job. Someone did competent research, exported it, and handed it to the ads side, and nobody noticed that keyword research for Google AdWords and keyword research for organic search are answering two different questions that happen to share a vocabulary and a toolset.
The question in SEO is: which terms can I realistically rank for, and will that traffic be worth the months it takes to earn? The question in Ads is: which terms describe a person whose next action is worth more to me than the click costs right now? Those sound adjacent. In practice they select almost disjoint sets of keywords, and the overlap is smaller than anyone expects the first time they measure it.
What follows is the distinction as I've come to understand it after cleaning up enough accounts where the two got merged, plus the inferences I now use to keep them separate.
Two disciplines, one vocabulary: what keyword research for Google AdWords is actually for
The structural difference is the cost model, and everything else follows from it. An organic page has a fixed cost paid once and an unbounded number of visits afterwards. That economics rewards breadth: a term that brings a thousand curious readers is worth targeting even if nine hundred and ninety of them never buy, because the marginal cost of serving them is zero and some fraction will return later.
Ads inverts that. Every visit has a price, paid again each time. The nine hundred and ninety readers are no longer free, they're a line item. A keyword that would be an obvious win organically can be an obvious loss in an auction, and nothing about the term itself changed. Only who pays for the traffic did.
That's why I stopped thinking of Ads research as a filtered version of SEO research. It isn't a subset with a stricter threshold, it's a different selection criterion applied to the same universe of queries. SEO asks whether a term is winnable and worth winning over time. Ads asks whether the average person behind that term, today, will do something worth more than the click cost. A term can pass one test decisively and fail the other decisively.
The timing gap reinforces it. SEO decisions are bets on months, so ambiguity is tolerable and often desirable, because a page ranking for a research term is building an audience that matures. Ads decisions are settled within weeks, and ambiguity is the thing you're paying to resolve. Every uncertain term is an experiment funded by the budget, which is fine when you've chosen to run it and expensive when you've inherited it without noticing.
The metrics invert: volume, difficulty and cost read backwards
The clearest sign that the two disciplines have been confused is when someone justifies a keyword using a metric that means something different in each context. Three do most of the damage.
Search volume stops being the prize
In SEO, volume is roughly the ceiling on what a ranking page can earn, so it's a reasonable first sort. In Ads, volume is the ceiling on how fast you can spend, which is not the same thing at all. A term with two hundred searches a month and unmistakable purchase intent will outperform one with twenty thousand ambiguous searches every time, because you pay per click on both and only one of them brings buyers. I use volume to size a theme and to decide whether a campaign can reach statistical significance in reasonable time. Beyond that it decides nothing.
Difficulty becomes a signal, not a warning
An SEO difficulty score is a barrier: high difficulty means months of work you might never recoup, so walking away is often correct. The bidding equivalent, a high suggested bid range, points the other way. It means multiple advertisers have run that term, measured what it returned, and chose to keep paying. That's the closest thing to third-party profitability validation available before you spend anything. I've learned to treat an expensive term as worth investigating and a suspiciously cheap one in a competitive vertical as worth suspecting, which is the exact reverse of the organic instinct.
Cost per click has no organic equivalent at all
This is the metric with no counterpart, and it's the one that decides everything. Once you know what a conversion is worth and roughly what rate to expect, a click price tells you immediately whether a keyword is affordable. That single number turns a subjective argument about relevance into arithmetic, and it's why I anchor the whole exercise on conversion value before looking at any list. The mechanics of how those cost estimates are produced, and how they map onto what you'll actually pay, are worth understanding properly, which I go into in my breakdown of how Google Ads keyword prices work.
Where most advertisers get it wrong
Four patterns account for nearly every misapplied list I've audited, and they're all versions of the same import error.
Reusing the SEO list wholesale. The most common and the most expensive. The list is topically excellent, which is exactly what makes it convincing, and it's full of informational queries that were selected because they were rankable. Those terms are content briefs. Running them as ad targets funds an audience of readers at auction prices.
Chasing head terms because the volume looks impressive. Broad category terms are strategic in SEO, where a ranking page catches the whole spread of intent behind them for free. In Ads that spread is precisely the problem: you're paying identically for the buyer and the student, and the student is the majority.
Treating relevance as a proxy for intent. This is the subtle one and the hardest to unlearn. A keyword can be perfectly on topic for the business and still describe someone with no intention of buying anything today. SEO trains you to value that person, correctly, because they might come back. Ads charges you the same for them as for a buyer and they return nothing.
Importing difficulty scores as bidding advice. A third-party tool telling you a keyword is hard is answering a ranking question. Applied to an auction it filters out the terms most likely to be profitable, which is an almost perfectly inverted heuristic.
The tell in an audit is consistent: a term with high impressions, respectable click-through and zero conversions over a meaningful window. That combination isn't a landing page problem or a bidding problem, whatever the account notes say. It's a term doing exactly what it was selected to do, which was attract readers. The corresponding queries usually deserve a place in the exclusion list rather than another round of optimisation, and I cover how I build that list during research in my guide to negative keywords in Google Ads.
Commercial intent: the inference that changes everything
Everything above reduces to one judgement, made per term: does this query describe someone ready to transact? It can't be looked up, only inferred, and after enough repetitions the inference becomes fairly reliable.
The query's grammar carries most of the signal. Questions indicate research, almost without exception. Modifiers like price, cost, near me, buy, hire and comparison phrasing indicate readiness. Bare product nouns are genuinely ambiguous and need a second source. Brand plus product usually indicates late-stage evaluation, someone who has already narrowed the field.
When grammar is ambiguous, the results page settles it. I search the term and read what Google has decided to serve, because that decision is made from vastly more behavioural data than I'll ever have. Guides, definitions and encyclopedic results mean Google concluded the intent is informational, and it concluded it from evidence. Ads, product listings and comparison pages confirm the commercial read. I've stopped arguing with that verdict, because the arguments I won on paper I lost in the account.
Then the landing page test, which catches what the other two miss. I picture the person behind the query arriving where I'd send them. If the honest prediction is "read a paragraph and leave", the term fails no matter how relevant it looks. If it's "check the price, see whether it fits, get in touch", it passes. That test rejects a lot of topically perfect terms, and rejecting them is the point.
One inference I'd underline because it took me longest to accept: intent belongs to the query, not the product. Two people searching for the same product with different phrasing are at different stages and are worth different amounts to me, and treating them identically because they're interested in the same thing is the error underneath all four mistakes above. That's why intent, not topic, drives how I group terms, a principle I apply account-wide in my Google Ads keyword strategy.
Reading a Google Ads keyword research tool for what it actually tells you
Tool choice matters less than knowing which question each one answers. A Google Ads keyword research tool and an SEO tool can return the same term with the same volume and mean completely different things by it, because they're modelling different outcomes.
Keyword Planner's advantage isn't idea generation, third-party tools do that at least as well. It's that the cost and volume figures come from the auction I'm about to enter rather than from an external model. Banded and rounded, but native. Google's own documentation is explicit that the tool exists to show the searches a keyword receives and the cost to target them, and that framing is the whole distinction in one sentence: it's a bidding instrument, not a ranking one. When I'm weighing a click price against a conversion value, that's the only source I want.
SEO tools stay in my process, but demoted to discovery. They're better at surfacing vocabulary I hadn't considered and at showing what competitors have built content around, which is often where the useful long tail hides. I read their suggestions and ignore their verdicts, because a difficulty score is a well-calculated answer to a question I'm not asking.
The source that beats both, once it exists, is the account's own search terms report. It isn't modelled or estimated. It's what actually matched, what actually cost money, and what actually converted, and within a few weeks of launch it starts correcting the research more reliably than any tool can. Pre-launch research exists to make those first weeks cheap enough that this data gets a chance to accumulate.
Using both together without letting one contaminate the other
None of this makes SEO research useless to an advertiser. It makes it useful for a specific and limited purpose: raw material, not a launch list.
What I take is the commercial subset, plus the vocabulary. SEO research is unusually good at surfacing how people phrase things, including phrasings no ads tool suggested, and that language often becomes the seed set that makes the expansion useful. What I leave behind is the entire informational layer and every difficulty judgement attached to it.
Running both properly also produces a genuine advantage that neither gives alone: the informational terms SEO wants are frequently the exact terms Ads should exclude, so one discipline's target list is the other's negative list. I build both from the same research pass now, which costs nothing extra and stops the same queries being paid for twice.
There's a second effect worth knowing. Where a page ranks organically, running ads on the same term produces reporting that tells you whether the paid click is adding incremental volume or buying traffic you'd have received free. That question is unanswerable from either side alone, and it's the one place where the two disciplines genuinely need each other rather than merely coexisting. It's also the clearest argument for keeping the lists separate but the analysis joined.
So the whole thing, condensed: SEO selects for what you can win and hold, Ads selects for who's ready to act now. Volume is a ceiling on spend rather than a prize, high cost is closer to validation than to warning, and relevance is not intent. Import the vocabulary and the commercial subset from your SEO work, leave the logic behind, and let the search terms report take over from week three. Every account I've fixed was fixed by drawing that line, and every one I've inherited broken had erased it.