SEO measurement · 10 min
Which organic landing pages generate qualified sales enquiries?
Eight qualified projects can justify a closer look at a landing page. They cannot tell you which query each customer typed. Here is how to keep the evidence useful from the first visit to sales review.

Start with the decision your sales director needs to make
Which pages bring in projects worth quoting for? The most-read article is not necessarily one of them. To answer, capture the observed acquisition context when an enquiry is received, carry it into the sales record and ask sales to qualify the project using an agreed definition. Leave the source unknown when the evidence is missing. A Search Console query report cannot tell you which search term a named customer used.
Consider a commercial door supplier. Its hinge adjustment guide attracts existing owners looking for help. Its fire-door specification page attracts fewer readers but helps an architect assemble a tender. Ranking these pages by traffic would miss the purpose of both. Give the support guide a service objective and evaluate the specification page against relevant project enquiries. Your first measurement decision is which pages belong in a sales comparison at all.
Three records with different units
Search Console describes visibility and clicks from Google Search. Analytics describes the visits and actions it can observe on the site. Your enquiry system records what the business received, and the CRM records what sales accepted. Put them alongside one another for context, but label the units: clicks, sessions and unique enquiries. A user might click twice before opening one project; two colleagues might work on that project from different devices.
Google's guidance on using the two reporting systems together recommends investigating trends in light of their different collection methods. If the trends diverge, inspect the date range, tagging coverage, hostname and filters before interpreting the business impact. Multiplying one series by a constant to match the other hides the reason for the difference.
Decide which search engines are in scope. An Organic Search channel report can include engines other than Google. Filter the analytics source when comparing specifically with Google Search Console. Keep hostname as part of the landing-page key: the same path on a product subdomain and a consulting website can serve different customers. Adding a domain-property total to a subdomain total would also count overlapping traffic.
Choose an attribution question you can answer honestly
First observed acquisition, the enquiry session and a prospect's answer to “How did you hear about us?” are separate fields. Imagine an architect who discovers a specification guide in January and opens a link in a supplier email in February. The first observed source may be organic search while the enquiry session is associated with email. Neither observation proves what caused the purchase.
A defensible starting measure is “unique received enquiries with an observed organic source for the submission session.” Keep the first known source separately where you can appropriately capture it. Select the corresponding traffic-source scope in Analytics. A first-user dimension answers a different question from a session dimension. If your CRM derives a classification from a referrer, label it as your own rule. Do not describe it as GA4 attribution or assume that the two classifications agree.
Do not classify a contact as organic merely because the enquiry arrived from an SEO article. The reader may have opened it from an email, bookmark or colleague's message. Equally, a buyer who says “Google” may not distinguish an ad from an organic result. Preserve the statement as self-reported evidence. Overwriting the observed field with that answer makes it impossible to assess the two sources independently.
Specify the handover from website to CRM
Draw one enquiry journey before choosing a connector. A reader opens a guide, moves to a service page, submits the form and receives confirmation. The receiving system assigns a stable internal request ID. The CRM transfer uses that ID so that retrying a failed transfer does not create another enquiry. Clicking Submit is not the same event as successful receipt.
A practical record includes receipt time, request ID, service, landing hostname and cleaned path, submission page, observed source and how it was established. Add sales status, reviewer, review date and decision reason. Give each field a purpose. Names, email addresses and free-text descriptions belong in the appropriate business system, not in general analytics event parameters or URL query strings.
The landing page must survive the move to the contact page. If the form always stores its current URL as the landing page, the contact page will appear to originate every opportunity. Test embedded forms and third-party booking systems explicitly. Where the context does not travel, store an unknown value and document the limitation. A missing analytics identifier should never prevent a genuine customer from contacting sales.
GA4 provides recommended events including generate_lead and qualify_lead. Names do not provide an integration. Specify the trigger, responsible system, duplicate handling and acceptance evidence for each event you implement. It is perfectly possible to build the business report from CRM outcomes without sending those later events back to Analytics. Start from the business decision rather than a wish to fill every analytics report.
| Field | Owner | Acceptance evidence |
|---|---|---|
| Enquiry ID | Website | Retry updates the same request. |
| Landing context | Analytics owner | Hostname and cleaned path precede the contact page. |
| Qualification | Sales | Review date, status and decision reason are present. |
Agree on qualification before rating the leads
For a commercial fit-out company, “requested a quote” is too broad. One workable definition is a business premises project within the service area, with a verified need and an agreed next step. Avoid requiring a final budget where the prospect reasonably expects a survey before costing the work. Test the rule on borderline examples with the people who will apply it.
Allow pending review, qualified, unsuitable with a reason, and duplicate. An unanswered call is not a negative qualification decision. A returning customer requesting work on another building may be a new opportunity, not a duplicate. Sales should own the judgement; marketing should be able to question a pattern of rejections without quietly editing the outcomes.
Give the team a short calibration exercise. Have two reviewers independently assess the same anonymised enquiry. If one counts a requested brochure as a project and the other waits for a specification, no dashboard can resolve the disagreement. Record the rule that separates the two cases. When the rule changes, date the change rather than silently comparing the new standard with old results.
Worked example: eight qualified enquiries and twelve unknown sources
This is illustrative arithmetic, not client performance. Suppose the business receives 30 distinct enquiries. Submission-session source is known for 18 and unknown for 12. Ten of the 18 meet the organic classification rule. Sales qualifies eight of those ten, rejects one and has not reviewed one. Source coverage is 18 divided by 30, or 60%. The observed organic cohort has eight confirmed qualifications. That is not an estimate of all opportunities organic search might have influenced.
Eight divided by ten gives an 80% confirmed qualification share among all observed organic enquiries. Eight divided by nine gives about 89% among reviewed organic enquiries. Both are correct; they answer different questions. Publish the counts beside the rates and state which denominator you use. An improving review backlog can otherwise look like an improving content strategy.
Resist distributing the twelve unknowns in proportion to the known sources. They might all come from a booking widget that drops acquisition context. Repairing it changes the source mix even if it produces no additional opportunity. Mark the repair date and compare coverage before drawing a conclusion about results. Do not backfill historical source values with guesses.
Turn the result into an editorial decision
For each service group, report page intent, landing page, received enquiries, qualified cases, unresolved reviews and leading disqualification reasons. Group enquiries by receipt date and allow comparable time for follow-up. A cohort from last week has had less time to reach a site survey than a cohort from last quarter. Show the review cut-off as well as the acquisition period.
A low-traffic page producing relevant projects may deserve stronger internal links or a companion guide that answers a recurring specification question. A busy support guide may already be doing its job. Offer a route to paid assistance where appropriate without converting the whole guide into a sales pitch. If prospects repeatedly request work outside the service area, clarify geographical coverage before increasing content production.
Review a handful of actual outcomes before acting on a percentage. Two excellent industrial projects may matter more to the business than twenty small requests it cannot fulfil. But a proposal value is not booked revenue, and attributed revenue is not proof of incremental demand. Keep those distinctions visible when deciding whether to commission another article or repair a service page.
Acceptance tests for the integration
Run a normal landing-page-to-form journey, a journey with no source available, a double submission, a temporary CRM outage and a retry after recovery. For each case agree on the expected record count, source context and visitor message. During an outage, distinguish a request safely queued for delivery from a request the site failed to receive.
Test the same path on two hostnames. Change the sales status from pending to qualified and confirm that the existing record changes rather than a second enquiry appearing. Have someone outside the implementation team reconcile a small sample of accepted requests with CRM records. If only the developer can explain missing rows, handover is unfinished.
Who signs off on the first monthly report?
Ask sales to pick one won project, one unsuitable request and one pending review. Trace each back to the intake record and the reported page group. Then reverse the exercise: take a row from the page report and ask whether sales can explain what happened to the enquiries behind it. A total that reconciles in one direction can still conceal a missing form or two projects merged under one contact.
Agree on an owner for exceptions. A brief review note should record the affected form, dates, evidence, correction and the earliest comparable period. A gap that cannot be reconstructed should remain visible. A manager deciding which page to fund next needs to know where evidence ends more than they need a tidy chart.
What to include in a measurement brief
Supply two representative landing pages, the form location, the name of the sales system and anonymised descriptions of a suitable and unsuitable project. Add the management question that remains unanswered. Ask for separate estimates for capture, CRM delivery, qualification workflow, reporting and monitoring. That makes it possible to commission the smallest piece that will change a decision.
With a handful of enquiries, a maintained manual register may be enough. Invest in integration when matching consumes meaningful time or loses important cases. Ask suppliers to name the journeys that will remain unobservable. That answer is more useful than a promise to assign every sale to a single keyword.
Sources and methodology
- Using Search Console and Google Analytics data for SEO (Google, accessed 2026-11-08)
- Scopes of traffic-source dimensions (Google, accessed 2026-11-08)
- Recommended events (Google, accessed 2026-11-08)
Frequently asked questions
Can we identify the keyword behind a closed deal?
Not from Search Console alone. Use aggregated queries to understand demand for a page group. Join an enquiry to an observed visit only where the necessary context is available.
Should we label every direct return as organic?
No. Record the first known touch separately and keep the submission-session rule consistent. A direct return does not reveal every earlier influence.
What if a prospect phones instead?
Record the call in the sales system. Without a reliable source link, keep its source unknown or self-reported. Do not allocate calls among pages in proportion to clicks.
When is a spreadsheet enough?
When someone can maintain distinct requests, statuses and review dates, and a sample can be reconciled without guesswork. Automate when the manual work becomes the limiting factor.