Google Ads Match Types: Design a Query-Control Plan Around Your Goals

Google Ads keyword match types shape the searches your campaign can reach. The useful choice depends on what you sell, which customers you can serve, what counts as a valuable outcome and how much uncertainty your budget can absorb. A small business offering one specialist service faces a different decision from a retailer trying to discover demand across a well-stocked product range.

Start with a query-control plan: a written explanation of the demand you want to reach, the matching approach you intend to use and the evidence that would justify changing it. The plan should make the campaign understandable to someone who did not build it. It should also acknowledge where Google interprets meaning and where your own knowledge of the business is needed.

This guide provides a relevance rubric, fictional query cases and a compact brief for evaluating broader matching. The examples are teaching aids, not account observations or promises about which searches Google will match. Use our high-intent search query guide for the wider relationship between search intent and commercial outcomes.

The starting decision: define the outcome first, choose matching and bidding together, and judge expansion against customer quality and acquisition cost. Exact, phrase and broad match each need that context.

Write the business boundary before the keyword list

A keyword can describe your industry without describing a job you want. Before choosing syntax, finish this sentence: “We want enquiries from these customers, for this service, within this area and operating capacity, at an acceptable acquisition cost.” Give each part an answer the business owner would recognise.

For a fictional company called Cedar Heating, the answer might be residential gas-furnace diagnosis and repairs at approved Toronto service addresses, during staffed daytime hours. Commercial boiler work, equipment sales without installation and overnight emergency attendance are different offers. This fictional boundary will be used throughout the guide; it is not a description of a Canada Create client.

Define the offer at the level a customer buys

“Heating services” is too broad to explain whether a repair enquiry belongs with a replacement enquiry. Write the actual service and its limits. Include the equipment, customer type, availability and important exclusions from the offer. A useful boundary might say that an assessment is available before a repair quote, while an installation price requires a separate consultation.

This exercise can reveal a landing-page problem before it becomes a keyword problem. If one page combines residential repairs, commercial installations and a recruitment message, the intended customer may be unclear. Choose a page that answers the service theme accurately. Do not expand the claim in the ad just to make more searches appear relevant.

Define what makes the outcome valuable

Specify whether the campaign exists to generate purchases, qualified enquiries, booked appointments or visibility around a defined subject. A person submitting a form is an observable action. Whether that person needs your service, is within reach and can buy it is a separate business judgement.

Agree on a cost ceiling using the business’s own economics. Include sales capacity and fulfilment constraints. If the team can handle only a few additional appointments, expansion has a different purpose than it would for a business with unused capacity. Record the ceiling as an operating decision, with an owner and review date, rather than a permanent industry benchmark.

Decide which uncertainty you can afford

Some uncertainty is about language: a customer may use an unfamiliar name for your service. Some is about fit: you do not yet know whether people using a certain phrase need the work you provide. Some is about measurement: qualified outcomes take time to become visible. These require different responses.

Language uncertainty can justify considering alternative wording. An unresolved service boundary calls for an answer from the offer owner. Measurement uncertainty calls for better evidence or a longer observation window. Changing the match type cannot, on its own, answer all three questions.

Understand the current meanings of broad, phrase and exact

Positive keywords make searches eligible for consideration; match types describe the relationship between a keyword and a search. Eligibility does not guarantee delivery. Google’s current definitions are based on meaning as well as wording.

These are semantic boundaries, not literal lists of every permitted query. Google describes Smart Bidding as critical when using broad match, so bids can reflect each search’s auction context. Smart Bidding can also work with phrase and exact; choosing those types does not require manual bidding. Google’s keyword matching documentation

Matching coverage overlaps

For equivalent keywords, phrase covers the matching reach of exact, while broad covers phrase and exact plus additional related searches. Broad matching can use context from landing pages, other ad-group keywords and previous searches. A search associated with an exact keyword is therefore not necessarily exclusive to that match type. Google’s explanation of keyword matching

That overlap changes how to read a keyword plan. Adding the same service in three syntaxes does not create three separate customer markets. It may support a specific management purpose, but write that purpose down. If the reason is simply “we always add every type”, the structure deserves another look.

Close variants remain part of exact and phrase matching

All positive match types are eligible for close variants, with no opt-out. Depending on the language and type, these can include spelling differences, plurals, function words, reordered wording, implied words, synonyms and the same search intent. Square brackets therefore do not create an identical-text-only setting. Google’s close-variant definition

Phrase match also goes beyond preserving a fixed sequence with words added at the ends. Word order matters when changing it changes the meaning. The former broad match modifier is no longer a separate matching behaviour; old instructions built around plus signs should not determine a new plan. Google’s explanation of phrase matching changes

For Cedar Heating, the practical implication is straightforward: describe the service clearly and expect interpretation. The business still needs to decide whether the demand reached is useful. No choice of punctuation replaces that judgement.

Use examples to reason, not to predict delivery

The following cases are illustrative matching possibilities. They are not observed Google Ads results, and the example pairs are not deterministic guarantees. A query shown beside one match type may also be reachable through another. The table separates a possible matching relationship from the question the business must answer.

The final row illustrates an offer conflict if that demand is encountered; it does not assert that broad match will serve that query. Preserve that distinction in your own examples. “Related to our keyword” is a platform question. “Worth pursuing for this campaign” is a commercial question.

A useful case library includes uncertainty and poor-fit demand alongside obvious fit. Otherwise, reviewers may approve a plan because every example was written to make the chosen approach look good. Include the questions a receptionist or salesperson actually needs answered before accepting a booking.

Apply a reusable relevance rubric

The rubric below is an editorial planning tool. It does not reproduce Google’s matching system, predict a Quality Score or calculate a conversion probability. Its purpose is to make the business reasoning visible before a team increases reach.

Assess one proposed demand theme against one named offer. A row about furnace repair should not quietly inherit the capabilities of a separate installation team. Write the evidence behind each judgement and leave unknowns visible.

Keep unknown separate from conflict

Supported means there is a reason to believe the dimension fits the named offer. It does not mean a sale is likely. Conflict means a known requirement falls outside the offer. Unknown means the available wording or evidence cannot resolve the question. Each label needs a short reason.

For “heating system technician near me”, service fit is unknown because the equipment is unspecified. Customer fit is also unknown because “near me” does not establish the address or customer type. In the companion case, the destination’s ability to answer this wider request is unresolved, so the decision is hold for clarification. Bounded exploration becomes defensible only after the owner establishes a service and page boundary that can contain the uncertainty.

Distinguish an unknown about one prospective customer from an unknown about the business’s capabilities. A service theme can remain a core candidate while individual enquiries still need an address and equipment details. Hold a proposal when an unresolved offer-level fact prevents a defensible decision; use bounded exploration only when the business can safely evaluate the uncertainty it has named.

Use the same standards for familiar and unfamiliar vocabulary. A conventional phrase can hide a poor-fit request; unusual wording can describe a valuable job. The rubric makes both possibilities discussable without treating the keyword’s appearance as proof.

Turn the assessment into one of four planning decisions

1. Core candidate: the offer can serve the demand, the page supports it and there is a clear business reason to pursue it. This is a candidate for the campaign’s main service theme, subject to settings and evidence.
2. Bounded exploration: the apparent need fits sufficiently to investigate, but the vocabulary, customer mix or outcome evidence is less established. State the unresolved question and the permitted evaluation budget.
3. Hold for clarification: an unresolved fact could materially change the decision. Assign an owner to resolve that fact before changing reach.
4. Outside this offer: a known conflict makes the demand unsuitable for this campaign. Record the conflict and whether another real offer could serve it.

Do not total these labels into a single score. Four favourable dimensions cannot cancel a service the business does not provide. Equally, one unknown does not automatically make a query worthless. The decision depends on which fact is missing and how consequential it is.

Work a price enquiry through the rubric

Consider the relevance-only case “furnace repair cost”. Cedar Heating offers diagnosis and repairs, so service fit is supported by the fictional offer. The words do not establish customer type, service address or appointment needs; those dimensions remain unknown. A page explaining assessment fees and the quoting process could support landing-page fit without promising a repair price before diagnosis.

Outcome evidence is also unknown until there is meaningful business evidence. The planning decision could be bounded exploration, with qualification based on the actual job request. If the business already has comparable, trustworthy qualified outcomes for that demand, it might instead become a core candidate. The same wording can justify a different decision when the evidence changes.

This assessment is not a matching claim. It shows how to evaluate a demand theme if you are considering it or encounter it. It also prevents a blanket rule that treats every price question as ready to buy or every price question as unhelpful.

Make disagreements useful

Have the offer owner and campaign manager assess a few cases independently, then compare their reasons. If one assumes commercial work is available while the other assumes residential only, the disagreement is about the offer. Resolve it there. If both agree on the offer but disagree on whether the query reveals enough intent, preserve the uncertainty and define the evidence needed.

Record the decision date and the offer version. A company that adds evening appointments may legitimately reassess demand previously outside its capacity. A prior decision should be traceable, not permanent simply because it was entered in a spreadsheet.

Build the matching plan around the type of demand

The following scenarios are fictional decision examples. They recommend questions and boundaries, not universal account settings. None reports campaign performance.

Scenario one: a narrow specialist service

Suppose a business repairs a limited set of residential heating equipment and cannot accept installation or commercial jobs. It wants more of the same qualified work, with little capacity for unsuitable enquiries. A plan using narrower matching may be reasonable because the service boundary is constrained and the cost of handling the wrong requests matters.

The rationale should name those constraints. “Exact is always best for leads” is not a useful explanation. A better record says that the campaign is concentrating on a defined repair theme, that close variants remain possible and that expansion will be reconsidered once qualification evidence and capacity support it.

Review the wording of the offer as carefully as the matching choice. If a broader phrase accurately describes the specialist job, it may deserve consideration. If the company cannot serve the demand, adding brackets around the phrase does not solve the service mismatch.

Scenario two: an established offer with capacity to grow

Now suppose the offer is clearly defined, the landing page is appropriate and the business has reliable evidence of qualified outcomes. The team has capacity and wants to discover how customers describe the service beyond its existing vocabulary. This is a stronger setting for a bounded broad-match evaluation with suitable bidding.

Start with a coherent service theme. Expansion should answer a specific question, such as whether additional ways of describing a repair generate acceptable enquiries. It should not quietly introduce unrelated offers, new locations, new conversion definitions and new creative at the same time.

The decision may still be to retain the existing approach if additional reach produces insufficient value. Broad matching is an opportunity to investigate demand, not evidence that the business should accept every related request. The relevant outcome is the quality and cost of the work acquired.

Scenario three: a long buying process

A commercial maintenance provider may receive enquiries from property managers who need specifications, procurement approval or a scheduled inspection before making a decision. A query that sounds exploratory can still belong to a valuable buying process. Judging every phrase by whether it suggests an immediate booking would misrepresent the offer.

Here the matching plan needs an explicit intermediate outcome, such as an enquiry from an eligible organisation about supported equipment. The final contract may arrive later. Decide which stage is sufficiently meaningful to inform the evaluation, and document the uncertainty between that stage and a sale.

A narrow exact-match list does not remove this delay. A broader list does not make the delay unacceptable. The question is whether the team can observe useful evidence, fund the evaluation and avoid mistaking early activity for commercial success.

Choose matching and bidding as a connected decision

Matching determines the kinds of searches available to a keyword. Bidding evaluates how to compete for opportunities within the campaign’s setup. A broad-match keyword does not, by itself, define which business outcome deserves a higher bid.

For Search campaigns, Google’s Smart Bidding strategies optimise towards conversions or conversion value at auction time. They include Maximize conversions, Target CPA, Maximize conversion value and Target ROAS. Google’s June 2026 naming transition relabels the target-bearing Maximize strategies as Target CPA or Target ROAS without changing their bidding behaviour. Maximize clicks and Target impression share are automated strategies with different objectives; they are not conversion-focused Smart Bidding. Google’s bidding documentation

A match-type proposal should therefore name the current bidding objective and the outcome it uses. If the team is evaluating broader coverage, it needs to know whether the campaign is pursuing raw enquiries, qualified enquiries or measured value. Those are different questions even when the keyword text stays the same.

Check that the goal expresses the business decision

Smart Bidding requires conversion tracking. Having tracking enabled establishes a prerequisite, but it does not establish that the chosen action represents business value. Google’s Smart Bidding setup requirements

In your plan, describe the event in plain language. “A unique eligible homeowner requested a supported repair” is more informative than “lead conversion”. Then identify what the actual system records today. If it records any form submission, preserve the difference between the desired outcome and the available signal.

Check which conversion actions are used by the selected campaign goals. Google distinguishes primary and secondary actions, and custom goals have an exception that can make an included secondary action usable for bidding. A label alone is therefore insufficient to infer the campaign’s optimisation behaviour. Google’s conversion-action guidance

This guide does not prescribe a tracking implementation. Its requirement is narrower: the person selecting match types should understand what success currently means in the account. If that cannot be explained, resolve the measurement question before treating expansion as a qualified-lead growth strategy.

Recognise a legitimate role for narrower matching

Google identifies non-conversion objectives, such as visibility around a specific set of searches, among the use cases for exact and phrase match. The appropriate combination follows the goal and constraints; it does not require an account-wide rule that one match type must always win. Google’s customer-reach guidance

For an offer requiring tightly bounded language, document why that boundary matters and obtain the appropriate business review. Do not describe exact match as a guarantee of regulatory compliance or as a literal whitelist. Semantic matching and close variants still need to be understood.

Also avoid changing bidding and matching together merely to simplify a presentation. If several conditions change at once, the team will have less clarity about which decision contributed to the result. A larger redesign may be justified, but its outcome should be described as a combined change.

Read the measurement uncertainty before choosing a winner

A matching decision is only as useful as the evidence used to judge it. Separate what is observed from what is inferred. Clicks and enquiries can be counted; the future value of an unresolved enquiry remains uncertain. A search that sounds commercial does not close that evidence gap.

Google documents that conversions can arrive after the initial interaction. Recent spend may therefore be visible before all associated outcomes are recorded. Your own qualification process can add another delay after the enquiry. Record both when deciding when the evidence is mature enough to use. Google’s conversion-delay guidance

A fictional example: cheaper enquiries, unresolved quality

The following numbers are entirely synthetic. They illustrate interpretation only and are not Google Ads benchmarks, client results or an executed experiment. Assume the two groups use a consistent cost basis and contain unique enquiries from a fixed cohort.

The broader approach has a lower cost per enquiry in this invented example. Its cost per currently qualified enquiry is higher, and more enquiries remain unresolved. Neither observation establishes what the final business comparison will be.

Cost per qualified enquiry is spend divided by qualified enquiries. If all pending enquiries qualify, the existing group’s final cost is $1,200 ÷ 10 = $120; if none qualify, it remains $1,200 ÷ 8 = $150. The broader group’s corresponding range is $1,320 ÷ 12 = $110 to $1,320 ÷ 6 = $220. These are completion bounds for this fixed cohort, not confidence intervals or forecasts. They assume no additional enquiries, unchanged costs and no reclassification of resolved records. If the qualified count is zero, report the cost per qualified enquiry as undefined, never $0.

The practical lesson is to preserve pending outcomes. Counting them as rejected could penalise a group with a slower qualification process; assuming they will all qualify would be equally unsupported. Use a comparable maturity window and the same qualification standard before making a durable matching decision.

Visibility and representativeness are separate limits

Some low-activity queries are omitted from Google’s search terms report for privacy. Visible query examples therefore cannot prove that every search reached was relevant. The report’s search-term match label can also differ from the keyword’s configured match type because a broad keyword can match a search in a narrower way. Google’s search terms report documentation

For a match-type evaluation, use examples to explain the kinds of demand encountered and use aggregate business outcomes to judge commercial performance. Do not extrapolate a handful of memorable queries into a claim about all traffic. Equally, a favourable aggregate should not conceal a known offer conflict that needs attention.

If the campaign has too little mature outcome evidence, say so. “Inconclusive” preserves the information honestly. It is different from proving that two approaches perform equally, and it does not oblige a small business to keep paying indefinitely for an answer it cannot afford to obtain.

Check the settings that sit around your keywords

A keyword list is not a complete account specification. Before attributing reach to a match type, record the campaign’s active targeting and expansion features. Otherwise, two people can look at the same list and make different assumptions about what is eligible to serve.

Keep matching, routing and delivery separate

Google’s standard prioritization rules favour an eligible exact keyword identical to the search term, with further rules covering relevance and Ad Rank. For these rules, identical includes spell-corrected searches but not plurals or synonyms. Eligibility exceptions include budget restrictions, low-search-volume status, disapprovals and targeting differences. This priority is not a promise that every semantically related search routes to the exact keyword. Google’s account prioritization guidance

For planning, separate three questions: could the keyword match the need, which eligible ad group would be considered, and would an ad actually appear? The public worksheet addresses the first question conceptually and the business value of the demand. It does not simulate the account’s routing or the auction.

Group keywords around offers that share an appropriate destination and customer expectation. If repair and replacement need different explanations, make that distinction intelligible in the plan. Avoid adding duplicate structures solely to make the keyword list look controlled; every additional structure creates another thing the team must understand and maintain.

Document AI Max separately from keyword syntax

AI Max is a feature suite within Search campaigns, not a fourth positive keyword syntax. Its search term matching can expand reach through broad match and keywordless technology informed by keywords, creatives and URLs. Turning on AI Max enables search term matching, which can be turned off at ad-group level. An exact-keyword list therefore does not, by itself, describe all the campaign’s matching routes. Google’s explanation of AI Max

Bidding is a condition here: Google’s AI Max FAQ states that search term matching does not work with Manual CPC and recommends conversion- or conversion-value-based bidding for the full capabilities. Record the bid strategy and the actual ad-group setting, not simply “AI Max on”. Google’s AI Max bidding FAQ

Record text customization and Final URL expansion separately. These campaign-level features can change the ad message and destination, so enabling them alongside broader matching changes more than the query coverage.

Google’s current notice schedules automatic upgrades of the legacy campaign-level broad-match setting to AI Max starting in September 2026. That announcement does not establish an individual campaign’s migration status. The legacy setting also gave broad keywords exact-like priority for identical searches; that special priority does not carry over when moving to AI Max. Check the actual settings and routing assumptions before implementation. Google’s broad-match campaign-setting notice

If the aim is to learn about adding broad keywords, avoid introducing an AI Max feature bundle as an unrecorded second change. If the business chooses the broader bundle, name it as the thing being evaluated and adjust the question accordingly.

Record dependencies without turning them into assumptions

Write down the approved geography, hours, offer boundaries and existing exclusion policy. A city name in a query does not establish the person’s location, and an urgent phrase does not create appointment capacity. These are dependencies that the implementation owner must check against the real campaign and business.

Positive matching also does not replace the need to manage unsuitable demand. Keep exclusion decisions in their own documented process: the relevance rubric can explain a conflict, but it does not specify a negative keyword, match rule or attachment scope. That separation preserves the reason for the decision without improvising its implementation.

Use a bounded brief to evaluate broader matching

A small match-type brief should be specific enough for an account owner to assess. It is a handoff into a properly designed experiment, rather than a complete experiment protocol or an instruction to make changes immediately.

Example hypothesis: adding broad-match versions of Cedar Heating’s established residential repair theme will produce more qualified repair enquiries while keeping cost and customer quality within the business’s agreed limits.

Google’s newer Broad Match experiment flow allocates traffic and budget within one existing campaign. The control retains the original keywords; the trial retains them and adds broad versions. This evaluates added coverage, not wholesale replacement of the original list. Google’s Broad Match experiment documentation

The native flow requires suitable Smart Bidding and has restrictions involving shared budgets, portfolio bidding, Display targeting, active trials and certain legacy features. Confirm eligibility from the current experiment requirements; a custom setup needs its own review.

Use available experiment-power guidance to assess whether the proposed campaign and duration can answer the question. The estimate is not a guarantee of a conclusive result. Google’s Campaign Guidance

For this proposed brief, keep automatic application off and require the owner to review business quality. Treat the spend limit as a monitored operating rule unless a verified platform control enforces it. Stop and investigate broken measurement or a breached boundary. An inconclusive result should state what remains unknown and whether further evaluation is worth funding.

Leave a query-control record someone else can use

Download the match-type query-control kit (ZIP).

A useful handoff has a short decision statement and enough evidence to challenge it. It should not require a colleague to infer the intended strategy from punctuation, screenshots or an unexplained list of keywords.

For each service theme, record the following:

• The offer and customer boundary, including any timing or equipment limits.
• The intended outcome, current measurement signal and acceptable acquisition cost.
• The proposed match type and why its reach suits this particular decision.
• A few illustrative queries, including ambiguity and a known offer conflict.
• The rubric labels, reasons, evidence gaps and the owner of each unresolved fact.
• The relevant campaign-setting dependencies, review date and condition that would justify a change.

Keep example wording separate from observed wording. If you later add a real observation, identify its source and date and preserve any visibility limits. A worksheet should never make a hypothetical pair look like proof that Google served it.

Use versioned offer descriptions when the business changes. Suppose Cedar Heating adds supported evening appointments. The update should say which availability limits changed and which remained. It should not automatically turn every urgent phrase into suitable demand.

Likewise, a favourable result for one service theme is evidence for that tested context. It does not establish that the same matching choice will work for an unrelated service, another location or a different season. Write the conclusion narrowly enough that the next decision can learn from it.

Questions to settle before changing match types

Does a small budget mean exact match only?

A small budget increases the importance of prioritising a clear offer and limiting unaffordable uncertainty. It does not establish a universal match-type rule. Consider the service boundary, bidding objective, measurement quality and whether a broader evaluation could produce enough useful evidence within the approved spend.

Does exact match mean the person is ready to buy?

The match relationship and the person’s buying stage are different questions. A closely matching search can still come from someone comparing options or seeking information. Use the offer and outcome evidence to judge value, and preserve uncertainty when the words do not establish intent.

Should every variation become another keyword?

A variation deserves attention when it exposes a distinct offer, customer need or management decision. Merely changing spelling or word order is not a business strategy. Review the intended coverage and the existing theme before adding complexity, especially when the new row has no separate purpose.

What should a successful review produce?

A decision with a reason: retain the current approach, investigate broader demand within limits, resolve a prerequisite, or remove an unsuitable offer assumption from the plan. The value of the review is that a colleague can explain the decision and identify what evidence would change it.

Bring that record to your next campaign discussion. For help connecting keyword matching to your offer, measurement and acquisition goals, discuss Google Ads management with Canada Create.