Paid Search Keywords: Check Real Queries
A paid search campaign can look tidy in its keyword list and still attract the wrong people. Imagine a shop that sells made-to-measure curtains adding the keyword curtains. A search for curtain repair, a search for a ready-made pair, and a search for a custom installation express different needs. The keyword describes the shop’s category, but it does not settle which of those requests the shop can profitably answer. This is an illustration, not a report of an actual campaign.
That distinction is the starting point for choosing paid search keywords. In Google Ads, a keyword is an advertiser’s instruction to the matching system; a search term is what a person actually entered. Depending on the match type, the two can differ. A keyword list therefore describes the searches you hope to reach, while the search terms report lets you see a portion of the searches that caused ads to appear. Google explains the difference between keywords and search terms here.
The practical goal is not to collect every plausible phrase containing your product name. It is to find requests for which your offer, ad, and destination page provide a credible next step, then pay attention to the queries that actually arrive. A narrow choice can miss useful demand; a broad choice can spend attention on people whose needs you cannot meet. Neither problem is solved by counting keywords alone.
Begin with the offer, then name the searcher’s job
Before entering a keyword, write down what the customer can obtain after the click. For the curtain shop, “made-to-measure curtains with installation” is a more useful description than “home décor.” It identifies both the product and the service. If the shop does not repair curtains or sell ready-made pairs, those distinctions matter before any match setting is chosen. Otherwise, a perfectly fluent ad can still lead a person to the wrong offer.
Next, think in complete requests rather than isolated nouns. Someone searching for “custom curtains for bay window” may need a different page from someone looking for “curtain fabric ideas.” The latter might be interested in the category but may not yet be seeking the service the shop sells. This is an editorial and commercial judgment, not a guarantee about what any one searcher will do. The useful question is whether the advertiser has a specific answer for the request visible in the query.
This approach also guards against a common mistake: treating keyword volume as a substitute for customer fit. A short, familiar word can describe many jobs. A longer request may reveal the material, location, service, or constraint that determines whether the business is suitable. More detail does not automatically make a keyword better, but it gives the advertiser a clearer promise to make in the ad and on the page. When those details matter to the purchase, the keyword plan should preserve them.
I would begin with a small set of distinct offer themes: the main product or service, its important variations, and the customer problems the offer genuinely resolves. The themes should reflect differences in the answer a visitor needs. If “custom curtains for bay windows” and “custom curtains for sliding doors” lead to different product information, separate them in the planning even if both eventually use similar match settings. If two phrases express the same need and lead to the same answer, adding both merely to make the list longer is unlikely to improve the decision.
Brand terms deserve the same discipline. A person searching for your business name may already know whom they want; a person searching for a category is choosing among possible providers. The ad can speak differently to those requests. That is a reason to inspect them separately when deciding what message and destination to use, not a claim that one group will always cost less or convert better. The business’s own results must settle that question.
Match type sets the reach, not a literal spelling rule
Google Ads offers broad, phrase, and exact match for search keywords. These settings describe different relationships between an advertiser’s keyword and a searcher’s query. Broad match can reach related searches; phrase match can reach searches that include the keyword’s meaning; exact match focuses on searches with the same meaning or intent. Google’s descriptions make clear that the system considers meaning, so an exact keyword is not a promise that every matching search will have identical characters. Google’s guide to keyword matching options explains those relationships.
The choice is a decision about how much discovery to permit. Exact match gives the most control of the three, but the smallest range of eligible searches. Phrase match opens a wider range while keeping the product or service meaning in view. Broad match can reach related requests beyond the direct meaning of the keyword. If the offer is specialized and the cost of an unrelated visit is high, I would usually start with more tightly defined keywords and examine the queries they bring in. If the business can serve a wider set of needs and can review the resulting traffic, a broader setting may be worth trying. These are starting judgments, not universal performance rankings.
Consider the illustrative keyword “custom curtains.” An exact version could still connect to a query that expresses the same intent in different words; phrase match can admit more specific requests that carry the same meaning; broad match may look farther into related searches. The important question is not whether a query repeats “custom curtains.” It is whether the person wants something this shop can deliver. That is why the matching setting cannot be evaluated in isolation from the offer.
The notation is useful, but it is easy to overread. In Google Ads, an unadorned keyword uses broad match, quotation marks indicate phrase match, and square brackets indicate exact match. Those symbols set the matching relationship; they do not force a searcher to use the exact words in the keyword. Google’s match type examples show the syntax and the meaning based behavior.
A practical plan can contain different match types for different reasons. A clearly defined service with a well suited landing page may justify broader reach because the advertiser can assess which related queries are valuable. A highly specific service with many adjacent products the business does not sell may justify narrower reach. The cost of the narrower choice is that some relevant queries may never be tested. The cost of the broader choice is that the advertiser must spend more care reviewing actual searches and excluding poor fits.
Avoid the false precision of filling an account with slight rewrites of the same phrase. A list of “red car” and “car red,” for example, does not necessarily create two meaningfully different opportunities. Google’s matching guidance says very similar keywords can be recognized as duplicates, with only one of the eligible keywords entering a given auction. Its keyword matching guidance gives that example. Build around distinct customer requests and usable page answers instead of spelling variations for their own sake.
Read the queries that triggered ads before changing the list
The search terms report is where a keyword plan meets actual demand. Google says the report shows searches that triggered ads and how those searches performed; its “Keyword” column identifies the advertiser keyword that matched a search term. Google’s search terms report guide describes both fields. When a keyword appears to be doing well or poorly, inspect the queries under it before deciding the keyword itself is the problem.
Start by reading a query as a customer request. If the curtain shop sees “curtain repair near me,” the issue is clear: the shop does not repair curtains. If it sees “made-to-measure bay window curtains,” the request is much closer to its offer. If it sees “curtain design ideas,” the fit is uncertain because the person may be gathering ideas before buying. These are illustrative terms; a real decision would require the account’s own search terms and results. The lesson is that the same advertiser keyword can lead to several commercial situations.
For each substantial query pattern, ask three questions. Can the business supply what is requested? Does the ad tell the truth about that supply? Does the landing page continue the same answer? A failure at the first question calls for less exposure to that request. A failure at the second or third may call for a better message or destination rather than a new keyword. This sequence prevents a keyword edit from being used to solve a page problem.
Do not assume the report is a complete transcript of every search. Google’s help page says that terms with insufficient query activity can be omitted from the search terms report for privacy, while search terms insights may group some of those searches into broader themes. Google explains that reporting limit. This makes the visible queries useful for decisions but incomplete as a census. An apparently clean report cannot establish that every unseen search was relevant.
There is another easy source of confusion: the report’s “Match type” label describes how a reported search term relates to the keyword; it may differ from the match type assigned to the advertiser keyword. Google’s example shows a broad keyword triggering an identical search term that the report calls exact. The report guide distinguishes those labels. When reviewing performance, keep the assigned keyword setting and the search term relationship in separate columns in your mind. Otherwise, you may credit a broad keyword’s result to an exact setting it never had.
Query review should lead to a small number of explicit actions. Keep a query pattern when the offer, message, page, and business result all make sense. Give a valuable recurring request its own keyword or message when the existing one is too general to serve it well. Reduce exposure to requests that the business cannot fulfill. Investigate ambiguous requests before making a sweeping exclusion. The direction of each action comes from the searcher’s need and the advertiser’s actual offer, not from a single word looking unfamiliar.
Exclude mismatches without blocking useful customers
Negative keywords provide a way to stop ads from showing for searches containing excluded terms. Google gives the example of a hat seller that does not sell baseball hats excluding “baseball hats.” Google’s matching options guide describes this use and notes that negative matching behaves differently from positive matching. An exclusion should therefore be chosen from the unwanted request, with attention to how the negative setting works, rather than copied blindly from a positive keyword list.
The curtain shop could consider excluding repair requests if it truly offers no repair service. Yet excluding a broad word such as “fabric” would be dangerous if customers looking for custom curtains routinely mention fabric. The difference is whether the term reliably identifies a service the shop does not offer. Start with unmistakable mismatches, then examine the effect of broader exclusions on relevant requests. This is a judgment about coverage: an exclusion can save wasted visits, but it can also make a suitable customer invisible to the campaign.
A visible search term can also suggest a better positive keyword. Suppose the shop repeatedly sees requests for made-to-measure bay window curtains and has a page that explains that service. A more specific keyword may give the advertiser a clearer way to write an ad and choose the destination. The reason to add it is the distinct promise and page, not a belief that repeating the query verbatim creates a secret ranking advantage. Google’s search terms guidance explicitly presents the report as a way to refine a keyword list and spot content ideas. Its report guide explains that use.
Do not turn every one-off query into a new keyword. A small list organized around real offer differences is easier to interpret than a large list of near synonyms. The stronger signal is a recurring request, a distinct customer need, or a page that can answer a specific variation better. Even then, the advertiser should look at the subsequent results. A query may sound ideal and still produce little value for reasons the words alone cannot reveal.
Make the ad and landing page answer the same request
A keyword is only the beginning of the customer’s path. An ad can be eligible for a relevant query while making a vague promise; a page can use the right product name yet fail to answer the question that prompted the click. For the curtain shop, an ad about custom fitting should lead to a page that actually explains the made-to-measure offer. Sending that visitor to a generic home décor page leaves the most important question unresolved. This is an illustrative editorial recommendation, not a measured claim about a particular shop.
Google’s Quality Score groups three components: expected click-through rate, ad relevance, and landing page experience. Google describes those components in its Quality Score guidance. The components are useful because they direct attention to different weaknesses. If the ad is poorly aligned with the search intent, changing the landing page alone may not fix the message. If the ad accurately promises a service but the page is thin or confusing, adding more keyword variations will not supply the missing answer.
The landing page should reflect the details that made the query worth paying for. If people search for custom curtains for a bay window, show the relevant service, what choices the visitor can make, and a clear way to proceed. If the service varies by location or product type, state the applicable conditions. This is how a keyword plan becomes useful to a real person: the click lands somewhere that resolves the request rather than merely repeating its vocabulary.
Google also says that ad quality is assessed in the context of each search and that the anticipated experience after the click is one factor in its ad system. Its explanation of ad quality supports treating the destination as part of the keyword decision. A term that attracts the right people may underperform when the ad or page fails them. Before discarding the term, read the query, ad, and page together.
There is a cost to tighter alignment. Distinct needs may require distinct copy and pages, and maintaining those pages takes work. The answer is not to create a page for every spelling variant. Create a separate answer when the customer asks a materially different question or when the business offers a different solution. Keep related phrases together when one truthful ad and one useful page can serve them. That makes the structure follow the offer instead of a mechanical rule about word similarity.
Use Quality Score to locate a weakness, not to set the goal
Quality Score is a diagnostic metric, not the score entered into each ad auction. Google describes a scale based on expected click-through rate, ad relevance, and landing page experience; it also says the displayed score itself is not an auction input. Google’s Quality Score guidance details the components, while its ad quality guidance states the auction distinction.
That distinction changes how to act. If ad relevance is weak, compare the search terms with the ad’s actual language and offer. If landing page experience is weak, inspect the destination the visitor reaches. If the expected click-through component is weak, look at whether the ad makes the offer clear and compelling for those queries. Those are plausible directions for work, not a guarantee that any one edit will increase the displayed score or produce profitable customers.
I would not choose a keyword simply because it might raise an average Quality Score. A high score does not tell you that the business can fulfill every associated query, and a low score does not specify a profitable fix. The business question remains: are people with relevant needs reaching a credible offer, and are the resulting customer actions worth their cost? Use the score to decide where to look; use the actual search terms and outcomes to decide what to keep.
Auction-time ad quality also has more context than a single static label can capture. Google’s ad quality guidance notes that quality varies with each search and can affect whether and where an ad shows. Google explains these effects. This is another reason to avoid treating the keyword as an isolated lever. The words, the ad, the destination, and the particular search together shape the opportunity.
Judge keywords by the customer action you intended to buy
Clicks and impressions can tell you whether an ad reached people; they cannot, by themselves, tell you whether those visits served the business. Before comparing keywords, specify the action that counts. For the curtain shop, an enquiry about a fitting might be more meaningful than a visit to a gallery. For a shop that sells directly online, a completed purchase might be the relevant action. These examples illustrate different decisions; the account’s actual conversion setup determines what its reports count.
In Google Ads, the “Conversions” column reports configured primary conversion actions, while “All conversions” can include a wider set of tracked actions. The reporting also depends on choices such as attribution and how conversions are counted. Google explains its conversion columns and settings. If a newsletter signup and a confirmed order are both tracked, a total that combines them does not answer the same business question as an order count. Decide which action you mean before declaring a keyword successful.
The conversion definition matters especially when two keyword groups seem to disagree. A broad keyword might show many tracked actions because it reaches a wide set of queries; a narrow keyword might show fewer actions but more of the action the business values. Without knowing which conversion actions feed the column, the comparison is incomplete. Inspect the definition of the column, then compare like with like. This is a measurement choice, not a claim that one match type systematically wins.
Attribution can also change which keyword receives credit. Google’s conversion guidance explains that attribution settings determine how credit is assigned across clicks, and that some reporting columns place conversions at the time of the click while others use the time of the conversion. Google’s conversion tracking guide describes these differences. If two reports appear to disagree, check their conversion actions and time basis before changing the keyword list. A reporting difference is not necessarily a change in customer demand.
Use ratios only with their denominators in view. Cost per conversion means spend divided by the conversions counted in the selected column; conversion rate compares conversions with eligible interactions. Google defines these related columns. A cost per conversion has little meaning until you know whether the counted action is an enquiry, purchase, or some earlier step. Likewise, a keyword with a few clicks and one reported conversion tells you less about repeatable performance than a larger, stable pattern, even if its displayed rate looks impressive.
An illustrative calculation makes the point. Suppose one keyword spends $200 and records ten brochure downloads, while another spends $200 and records two completed sales. The first shows $20 per counted action and the second $100, but those figures measure different outcomes. They cannot be ranked without knowing what a download and a sale are worth to this business and whether both are included in the same reporting column. The numbers are hypothetical; the decision rule is to define the customer action before comparing the arithmetic.
Refine in a cycle that follows the query
A workable routine begins with the existing offer, a limited set of keyword themes, and an explicit match choice for each. After ads have run, read the visible search terms and group them by the need expressed, not merely by shared words. Compare those needs with the business’s offer. Then inspect the ad and page that each worthwhile group saw. Only after that should you choose among adding a specific keyword, improving the message, improving the destination, or excluding an unwanted request.
Keep the business result in that same loop. A query that looks relevant but produces only a weakly defined tracked action deserves more scrutiny. A query that looks unusual but repeatedly leads to the desired action may reveal a real customer need the initial list missed. Because Google matches by meaning and intent, the account can discover useful phrasing beyond the advertiser’s first vocabulary. Google’s matching guide explains that broader reach; its search terms guide shows where to inspect the resulting queries.
Change one business question at a time when possible. If you replace a broad keyword, rewrite the ad, change the landing page, and alter which actions count as conversions at once, a subsequent change in results becomes hard to interpret. This is a practical recommendation, not a platform rule. A disciplined sequence helps you decide whether the problem was reach, promise, destination, or measurement. Sometimes several parts must change together because the current offer is misleading; in that case, record what changed and judge the new path as a whole.
The best keyword set is therefore neither the longest list nor the tightest possible list. It is a set of invitations to searches the business can answer, with enough room to discover related demand and enough attention to reject requests that do not fit. Match type controls how broadly those invitations can be interpreted. Search terms reveal the actual requests that became visible. Ads and landing pages carry the promise through the click. Conversion definitions tell you whether the result was the customer action you meant to buy. Keep those four decisions connected, and keyword choices become clearer as the campaign runs.