Landing Page Message Match: Carry the Promise from Query to Ad to Page

Message match is the continuity of a specific promise from the search someone makes, to the ad they read, to the page they reach and the action they take. A visitor should be able to follow that promise without discovering that the service, eligibility, price condition or next step has quietly changed.

Repeating a keyword is not enough. An ad and page can both say “commercial cleaning” while disagreeing about whether the business serves a small office, offers a one-off visit or requires an annual contract. The words match; the decision does not. Conversely, different wording can preserve the same promise when the meaning and conditions remain clear.

This guide gives marketing leads and website owners a practical way to inspect that continuity. You will build a message continuity matrix, attach evidence to important claims, name the person responsible for each promise and write a validation criterion for each proposed repair. The companion worksheet includes a blank version and fictional examples you can replace with your own authorised material.

The task is deliberately narrow. Canada Create’s guides to calls to action, PPC returns and conversion rate optimisation cover related decisions. Here, the question is whether a particular journey preserves its promise. Finding a contradiction is useful evidence of a communication problem. It does not establish how much that problem affects enquiries or sales.

Define the promise before comparing the words

A promise has several parts: who the offer is for, what the business will provide, where and when it is available, what it costs or requires, why the reader should believe it and what happens next. You do not need every part in every headline. You do need an honest account of the parts that could change someone’s decision to click or continue.

Trace price and qualification as part of the offer. If a price appears, record its currency, unit, what is included, required fees and whether payment repeats. A starting price, an estimate and a fixed total make different promises. If eligibility depends on location, business type, plan or a minimum order, carry that condition to the point where someone decides to proceed. Do not invent a price when the real offer is a quote.

For example, a fictional “no-charge scope call” must not become a free cleaning visit. A fictional monthly price must not become a one-time total on the destination. Write separate rows for the cost and eligibility conditions so an aligned service name cannot conceal an altered commitment.

Start by completing one sentence: “A person looking for this needs to understand that we provide this service, under these conditions, and that the next action does this.” If the team cannot agree on that sentence, comparing headlines will conceal the disagreement. Ask the service owner to settle the offer first.

Separate three relationships. Query-to-ad match asks whether the ad addresses the apparent need. Ad-to-page match asks whether the destination preserves the advertised offer. Page-to-action match asks whether the button, form and confirmation deliver the next step that the page describes. A strong connection at one stage does not repair a break at another.

For example, “Request a consultation” can accurately lead to an enquiry form, followed by a message explaining that the team will contact the person to arrange a time. “Confirm your appointment” creates a different expectation. If it leads to the same unconfirmed enquiry, the problem is the promised action, even if every service keyword is repeated.

Also distinguish message match from Google Ads keyword match types. The former is a review method for meaning across a journey. The latter are platform controls governing how keywords relate to searches. They are not interchangeable concepts.

Google describes Quality Score as a diagnostic and identifies ad relevance and landing page experience among its components. It also says Quality Score is not a key performance indicator or an input to the ad auction. A completed message worksheet therefore cannot promise a higher score, lower advertising costs or better rankings. Use the worksheet to document what you observed and what needs verification. See Google’s Quality Score documentation.

Assemble a small, dated journey packet

Choose one meaningful journey rather than the whole website. A useful starting point is one service, one audience and one next action. If the same page receives substantially different promises, treat each promise as a separate journey. A page offering both a self-service download and a paid consultation may need two records even when its URL never changes.

Collect the query or intent statement, the relevant ad wording, the destination page version, the CTA label, the form introduction and the confirmation message. Record the observation date, language and any condition that changes the experience. Examples include mobile versus desktop, a location-specific page, or a promotion that expires on a stated date.

Label the basis of the query. It might be an authorised reported search term, a planned keyword, a statement from customer research or a fictional scenario. A planned keyword is not proof of what a person actually searched. A plausible intent interpretation is still an interpretation.

Google’s search terms report documents terms that met its reporting conditions and triggered ads; it is not simply a copy of an advertiser’s keyword list. Its coverage is limited, so absence from that report does not prove that a search never occurred. Preserve the distinction between reported terms, inferred intent and missing information. See Google’s explanation of the search terms report.

Do not join separate reports into a supposedly exact individual journey unless the evidence actually connects them. A search term in one export and an ad combination in another can support a possible-path review, but their coexistence does not prove that the same visitor saw both. Name the packet “possible combination” when that is what you have.

Preserve short exact excerpts from your own authorised material alongside the interpretation. “The page feels generic” is difficult to verify. “The ad offers a one-off visit; the page only describes monthly plans” identifies the gap. Keep personal details and customer correspondence out of a worksheet that will be shared publicly.

Build the message continuity matrix

Use one row for one promise within one journey. Split a statement containing several promises into separate rows when different people must verify them. “Free same-day installation across Ontario” contains cost, timing, service and geographic commitments. A single green status would hide which of those commitments has evidence.

Read across each row from the query to the ad, page and action. Then read down each stage. Reading across finds a disappearing condition; reading down finds contradictions within a page, such as an introduction promising a free assessment and a form describing a paid visit.

The worksheet uses four match states. Aligned means the inspected statements preserve the same meaning. Ambiguous means more than one reasonable interpretation remains. Contradicted means the statements or evidence conflict. Unknown means a necessary part of the journey has not been inspected. These are editorial review states, not platform ratings or performance scores.

Keep evidence state separate. A claim can be repeated consistently while remaining unsupported. Marking that row aligned does not authorise the claim. Equally, an offer can be well supported by operations but poorly communicated on the page. That row needs a message repair, not new proof that the service exists.

Choose a state from the evidence available. An observed conflict remains contradicted even if another stage is uninspected; note the missing stage in the finding. Use unknown when the necessary comparison cannot be made, and ambiguous when inspected wording permits materially different readings. Do not use aligned as shorthand for an approved or substantiated claim.

Finish every finding with an observable acceptance condition. Replace “make the page clearer” with “the page and form both say this is a request, and the confirmation states that no appointment is reserved.” Someone else should be able to inspect the proposed version and decide whether that condition has been met.

A repair also needs an owner and a review date. The writer can change the sentence; the operations manager may need to confirm its truth. If those are different people, record the evidence owner in the worksheet and identify the person checking the proposed change in the reviewer field. Responsibility should survive a handoff.

Give every consequential claim an evidence owner

Evidence ownership answers a practical question: who can confirm that the business can still honour this promise? Choose a role close to the underlying fact. Operations can confirm service coverage, the product owner can confirm a feature, and the person responsible for fulfilment can confirm what happens after an order. A marketing approver may coordinate the decision without owning the operational truth.

The evidence register should identify the claim, supporting statement, scope, limitations, owner, the person or accountable role who checked it, date checked and review trigger. The owner maintains the underlying fact; the checker records the review. Those roles may be held by the same person. A source reference is useful only if another authorised reviewer can find the relevant information. “Website” is too vague. “Current service schedule, Toronto office-cleaning section, approved 8 October” points to a specific basis.

Do not mistake one kind of evidence for another. A testimonial about communication does not prove a response-time commitment. A company’s years in business do not prove that a particular service has been available for that entire period. A feature demonstrated in a test account does not establish that it is included in every plan.

Look for scope drift as evidence moves into copy. “Available for selected postal codes” may become “serving the GTA.” “Request a same-day callback” may become “same-day service.” Those changes are small in length and large in meaning. Record the original supported statement so the reviewer can compare the proposed wording with it.

Use evidence states of verified, missing, stale or not applicable. “Verified” means a named reviewer checked a relevant source for the stated scope and date. It does not mean the claim is permanently true. In the fictional companion, verified means supported within the supplied invented scenario only; none of its records is evidence about a real business.

Google’s misrepresentation policy addresses misleading or omitted information, unclear relevance and unavailable offers. That makes a claim’s availability and discoverability a separate concern from whether two sentences sound similar. The worksheet is not a policy clearance or legal review. For platform requirements, use the current Google Ads policy and obtain appropriate review for the business’s actual claims.

When evidence is missing, the practical choices are to obtain support, narrow the statement to something supportable or remove the claim. Moving an unsupported promise from an ad into a prominent page heading merely reproduces the problem.

Worked example: an office-cleaning enquiry with five different breaks

Fictional example: Alder Quay Office Care is an invented company used only to demonstrate the method. Its invented offer is a one-off evening office-cleaning service for Toronto addresses. A no-charge scope call is available, but cleaning is paid and scheduling follows a scope review. No client, advertising result or operational test is represented.

The imagined search is “one off office cleaning Toronto evening.” The draft ad contains three headline ideas: “One-Off Office Cleaning,” “Evening Visits in Toronto” and “Request a Scope Call.” Its description says, “Tell us your office needs. Request a no-charge scope call before booking cleaning.” These are specimen assets for review, not a record of a served ad.

The fictional destination instead opens with “Commercial Cleaning Across the GTA.” It discusses monthly plans, calls its enquiry a “free clean assessment,” displays “Toronto’s best-rated office cleaners” without a source, and ends with “Book your clean.” Its form confirmation says only that the request was received. Each issue belongs in its own row.

Rebuild the promise using the supported offer

A revised opening could say: “One-off evening office cleaning in Toronto.” The supporting sentence could read: “Tell us about your space. We offer a no-charge scope call, then quote the cleaning work and confirm an available time.” That retains the service and location while clarifying the sequence.

The offer explanation should distinguish a requested time from an available appointment. A suitable fictional sentence is: “Evening visits are subject to availability and confirmation after the scope review.” This matters because an evening preference field does not, by itself, reserve evening capacity.

The CTA becomes “Request a scope call.” The form introduction explains that the team will review the office location and requirements before arranging the call. The confirmation says: “Your request has been received. No cleaning visit is booked yet.” The wording describes the same action at all three points.

The revised page does not need to hide monthly plans. It can explain that ongoing cleaning is a separate option. The acceptance condition is that a reader pursuing the one-off offer can understand and request it without assuming a recurring commitment.

Review the repair without inventing its impact

The operations owner checks the fictional service specification. The reviewer then checks whether the revised copy preserves Toronto-only coverage, one-off availability, a no-charge call and a later booking decision. The unsupported rating claim is removed from the proposed version.

That is a completed reasoning exercise, not a completed campaign test. The worksheet’s original findings remain visible and its repair-validation fields remain marked not run until the proposed wording and behaviour are actually inspected in the relevant environment. An example can demonstrate how to write acceptance criteria without pretending those criteria have passed on a real site.

The correct result statement is: “The proposed wording resolves the five identified issues within this fictional scenario.” It is not “the page will convert better.” Whether real people understand the revised offer, and whether that affects suitable enquiries, are separate questions.

Use contradiction pairs to expose conditions that disappear

Some mismatches are easy to miss because each sentence seems reasonable on its own. Put the statements next to each other, then write what a reader would have to assume to reconcile them. If the required assumption changes eligibility, cost or commitment, the condition deserves explicit review.

Fictional retail example: delivery timing

An invented retailer, Cedar Box Supply, advertises “Dispatch within two business days.” Its landing page says “Delivered in two days across Canada.” The fictional warehouse rule supports dispatch after payment and stock confirmation; it contains no carrier delivery guarantee. The page has changed both the event being timed and the geographic promise.

The repair is to carry “dispatch” through the page and identify where delivery estimates become available. Do not substitute an invented arrival estimate to make the offer more attractive. Ask the fulfilment owner to confirm the start of the dispatch clock, relevant stock conditions and any exclusions before choosing the final wording.

A useful acceptance criterion is: “The ad and page describe dispatch, the clock starts after the stated conditions, and the checkout does not label the estimate as guaranteed delivery.” This checks continuity. It does not calculate shipping performance or establish consumer-law compliance.

Fictional software example: access versus a sales conversation

An invented software provider, Larch Workflow, uses “Try the scheduling tool” in an ad. The page says “Explore our scheduler,” but its button opens a sales-demo request. The fictional product specification says there is no self-service trial. A person expecting immediate access is being offered a different action.

If the actual offer is a demo, the ad and page should explain that offer. “Request a scheduling demo” preserves the service while naming the commitment. Changing the button alone would leave the earlier expectation unresolved. Creating a real trial would be a product decision with a much larger scope.

The validation criterion should identify the endpoint: “Before submitting contact details, the reader can tell this is a demo request, and the confirmation does not promise immediate account access.” The evidence owner is the product owner; the reviewer needs to inspect the form and confirmation as well as the page introduction.

When uncertainty is the finding

Suppose a repair business advertises weekend availability, but the only evidence is an undated brochure. The page repeats the claim accurately. The wording is aligned, while availability remains unverified. Record that combination rather than forcing the row into either a clean pass or a contradiction.

This separation prevents two common errors: blaming the landing page for a promise created upstream, and approving a promise merely because it appears everywhere. It also gives the right person a precise task. The next step is to confirm weekend capacity, not to conduct a general copywriting session.

Review combinations and alternative paths

A single polished ad mock-up is not always a complete representation of the messages that can appear. Google’s responsive search ad documentation explains that headlines and descriptions can be assembled in different combinations and orders. It also notes that previews do not show every combination and that some formats may shorten text. See Google’s responsive search ad documentation.

For message review, inspect whether individual assets depend on another asset to qualify their meaning. “No charge” is risky as an isolated promise if only a different line explains that the free item is a short scope call. Build wording whose meaning remains supportable within the combinations and formats being reviewed. Have the account owner confirm any required display controls.

Google provides a combinations report that can show common asset combinations and their impressions. It cautions against assuming a static recreation will perform the same way. Use the report as evidence about reported combinations, not as a universal picture of every visitor’s experience or proof of a causal outcome. See Google’s combinations report guidance.

Include alternative routes when they carry a different promise. A sitelink may open a pricing page, while the main ad opens a service page. A returning visitor may use a saved URL. A location selector may send someone to a page with different availability. Record these as distinct journeys when they change the offer or next action.

If the account uses features that alter ad text or destinations, ask its authorised owner which settings apply and inspect the available evidence for those paths. Do not assume a fixed destination from a planning document. Equally, do not declare an automated path broken without knowing what was actually eligible or observed.

Keep a clear coverage note: the combinations and paths inspected, their evidence dates and what remains unknown. That note is more useful than a statement that “all ads match” after reviewing one screenshot.

Choose the smallest repair that restores the truth

Start with the verified offer. If the ad promises something the business cannot provide, changing the page to echo the ad is the wrong repair. Narrow or remove the upstream promise. If the ad is accurate and the page has dropped an important condition, bring the page back into alignment. If the business has changed the offer, update every affected stage together.

Give priority to contradictions about eligibility, availability, payment and commitment. A visitor who cannot obtain the advertised service faces a different problem from a visitor who encounters an awkward sentence. Record the reason for priority in plain language rather than creating a weighted score that implies unsupported precision.

Next, identify missing explanations that can change the interpretation. “Request a quote” may be accurate but incomplete if the business first requires an on-site survey. The message need not describe every operational detail. It should disclose the next material step before the person commits to it.

Decide whether different intents require different destinations. Two audiences can share a page if the relevant offer is clear and the route preserves their expectations. They may need separate destinations when their eligibility, deliverables or next actions conflict. Do not create a page for every keyword merely to achieve identical wording.

Write the repair in the worksheet before implementation. Include the affected surfaces, supported statement, responsible owner and acceptance criterion. Then check neighbouring promises that the change could disturb. Narrowing “across Ontario” to “Toronto” should trigger a review of location copy elsewhere in that journey, not an unrelated website redesign.

Keep the original observation. A dated before-and-after record lets another reviewer understand why the change was made and prevents a later campaign refresh from quietly restoring the same unsupported promise.

Validate the promise in three separate layers

First, check the documents and proposed wording

Compare the relevant statements against the evidence register. Confirm that each material condition survives, evidence owners are named and acceptance criteria describe observable behaviour. A colleague should be able to repeat this review from the packet without relying on the writer’s memory.

The companion examples support this kind of exercise. You can compare their invented evidence statements with the worksheet findings and proposed repairs. Successful checks of CSV structure or fictional references establish that the files are usable and internally consistent. They do not establish that an advertising account, website or booking integration works.

Second, check what people understand

For a real proposed journey, ask people who resemble the intended audience to review the ad and destination. Use neutral questions: “What service is being offered?”, “Who can use it?”, “What is free or paid?”, and “What would happen if you selected this button?” Do not tell them the intended answer before asking.

Record the wording they use and where they found it. If they infer a confirmed booking from a request form, capture the specific cue that produced that interpretation. If they cannot find a condition, distinguish absent wording from wording that exists but was not noticed. Those findings call for different repairs.

A small comprehension exercise can reveal a misunderstanding worth investigating. It cannot provide a population-wide conversion forecast. Report the participants, tasks and limitations honestly. Do not turn three helpful conversations into a percentage claim about all prospective customers.

Third, check the authorised action path

In an approved test environment or an explicitly authorised live check, follow the intended route and inspect the resulting state. A request should produce a request confirmation. A confirmed booking should identify the reserved arrangement. A download promise should lead to the intended resource under the stated access conditions.

Define the expected result before submitting anything. Use authorised test details, identify test records and arrange their handling with the responsible team. A button click is not sufficient evidence of form delivery, a CRM handoff or an appointment reservation. Each claimed endpoint needs its own observable confirmation.

Record pass, fail or not run for each acceptance criterion. For a performed check, also record its date, environment and version, and the actual result or a findable evidence reference. “Not run” is the correct status when the environment or permission is unavailable. Leave its check date empty and record the blocker rather than inventing an observation. A reviewer can approve a proposed sentence while leaving the functional criterion open. If a criterion combines several required checks, pass requires all of them; a known failed part means fail, while incomplete checks without a known failure remain not run. Record the completed parts separately in the validation evidence. Avoid merging those decisions into a single completed checkbox.

Inspect the mobile and language variants that matter to the offer. For example, a qualifier present in one language but absent in another creates a different promise. This is a focused continuity check; it is not a substitute for a full accessibility, translation or technical audit.

Measure impact without confusing it with a corrected mismatch

A contradiction can be documented directly: one statement says the visit is free and another says it is paid. Its business impact requires additional evidence. Keep those two conclusions separate in your reporting. You can justify correcting an inaccurate promise without claiming that the correction generated a particular number of sales.

For a comprehension improvement that is suitable for testing, write a bounded hypothesis. For example: “Clarifying that this is a scope-call request may reduce enquiries from people who believe a cleaning visit is already booked.” That identifies the misunderstanding and an outcome worth observing without promising improvement.

Define the outcome before looking at results. A form submission, an eligible enquiry and a confirmed appointment answer different questions. Choose the one relevant to the hypothesis and specify how duplicates, test submissions and delayed outcomes will be handled. Keep the definition consistent across the comparison.

Also name a guardrail. A change that reduces mistaken booking expectations might reduce total submissions while leaving suitable enquiries unchanged. Another change might make the offer clearer but accidentally hide a valid service option. Reviewing only total form counts would miss those distinctions.

Where traffic, tooling and authorisation support a controlled experiment, plan the comparison with an appropriately skilled analyst. Google recommends focusing experiments on one variable so the effect can be interpreted; see its experiment guidance. A coordinated rewrite across several stages is a bundle of changes. If tested as a bundle, report the result at that level.

Specify the eligible audience, assignment method, primary outcome, observation period, treatment of delayed outcomes and decision rule before starting. This guide does not supply a universal sample size or test duration because those choices depend on the actual outcome, traffic and effect being investigated.

When only a before-and-after comparison is feasible, record concurrent changes such as price, staffing, campaign targeting or season. An observed difference is not automatically caused by the message repair. Low traffic may support an honest conclusion that comprehension improved in the reviewed sessions while commercial impact remains unknown.

A suitable review note has three parts: the mismatch observed, the correction verified and the outcome evidence available. For the fictional cleaning case, only the first two can be explored on paper. No conversion lift, ranking change, Quality Score gain or client result has been measured.

Use the companion worksheet without turning it into a project

Download the message-match worksheets (ZIP)

Start with trace-blank.csv and add one row for each consequential promise. Use trace-fictional.csv as a worked reference, not a set of claims to copy into a real campaign. The blank evidence register and its fictional counterpart use JSON; the instructions explain how to read or replace the fields without writing software.

The dictionary defines the columns and allowed states. Keep the identifiers stable so a finding still points to the right evidence after sorting. If several rows share an evidence record, confirm that the record’s scope supports every linked claim. A broad reference should not conceal a narrow exception.

The files contain no macros, live integrations or performance calculator. The CSV states are manually entered labels; the file does not enforce approval or calculate whether a repair has passed. They are a compact record of statements, evidence, decisions and checks. Import CSV fields as text in your preferred spreadsheet application, preserve commas and quotation marks, and keep sensitive operational evidence in an appropriately restricted working copy.

To run a first review, choose one journey, capture its current wording, ask the relevant owner to verify the offer and write a testable repair for each supported finding. A short packet that resolves one important contradiction is more useful than a large inventory in which every evidence field says “to be confirmed.”

Recheck when the offer changes

Message match can deteriorate without a page redesign. A service area contracts, a free consultation becomes paid, a product feature moves to a different plan or a booking provider changes its confirmation behaviour. Make those events explicit review triggers in the evidence register.

Assign a review date based on how volatile the promise is. A seasonal promotion needs attention around its end date; a stable service description may need review when operations changes. No universal weekly cadence can replace knowing who owns the offer and what would invalidate it.

At each review, verify the underlying fact first, then follow it through the relevant ad assets, page copy and next action. Retain the previous wording and record what was changed. Close a finding only when its stated criterion has been checked, and reopen it if new evidence conflicts.

The useful end product is a traceable decision: this is the promise, this is why the business can make it, this is where the reader sees it and this is what happens next. That gives the marketing and delivery teams something concrete to maintain.

Common questions about message match

Must the landing page repeat the ad headline exactly?

No. Preserve the meaning, offer and conditions. Exact repetition can help a reader recognise the subject, but repeating an unsupported or ambiguous claim does not make it accurate. Judge the complete promise, including the next action.

Can several ads use one page?

Yes, when the page clearly supports their respective promises and routes. Review each distinct intent and offer. If one ad promises immediate access while another promises a consultation, a shared destination needs to explain and deliver both paths or the journeys need to change.

What if everything matches but enquiries are still weak?

Message continuity is only one question. Demand, offer suitability, usability, measurement and follow-up can still matter. Record that the inspected promise is consistent, then investigate the next supported question. Do not keep rewriting matched copy simply because the business outcome is disappointing.

Sources and method: Public documentation was reviewed on 8 October 2026. The worksheet, review states and examples are original practical methods; fictional companies and figures describe no real campaign. Public source checks and synthetic file checks are distinct from live integration testing.

Discuss a conversion review with Canada Create.