Before you approve another month of work, you should be able to see which enquiries were worth pursuing and what the website needs next.
Start with the business’s received enquiries. Put the search and website records beside them. That gives you a more useful decision than a report ending with traffic, rankings or mentions.
Download the visibility and enquiry scorecard (XLSX). It includes a blank scorecard, an enquiry log and a worked example. You can use it without buying a CRM or another reporting tool.
Looking for industry research to put beside your own records? Our AI SEO statistics reference explains what published studies measured, when they observed it and where comparisons stop being useful.
Still deciding whether to buy? The AI SEO purchase guide models a prospective commitment. This scorecard is for reading the actual evidence once work is underway.
Decide what a suitable enquiry means
Write down the service, market and practical fit you want before classifying the enquiries.
For an HR advisory business, that might mean employers seeking outsourced HR support in its service market, with a relevant business need. A job application or a request for individual career advice would be a different need. An unclear request stays Pending until someone checks it.
Use the same definition across reviews. Record why an enquiry does not fit, including a service mismatch, unsupported location or unsuitable project scope. Those reasons can tell you more about the page than a total enquiry count.
An opportunity is a suitable enquiry with an agreed commercial next step. A call request alone does not establish that stage. Record Won or Lost only for an opportunity you actually pursued.
Read four kinds of evidence together
| Record | What it helps you decide |
|---|---|
| Search visibility | Whether relevant pages appear for the services and markets you want. Inspect the queries and pages behind the total. |
| Recorded website visits | Where measurable visits begin and whether people reach a useful next action. |
| Received enquiries and fit | What people asked for, whether the business received it and whether it matched your service. |
| Opportunities and outcomes | Which suitable enquiries progressed and what happened after the conversation. |
Search Console and Analytics have different collection and reporting rules. A search click is not the same unit as a session, and neither identifies a qualified buyer by itself.
Google now provides dedicated generative AI visibility reports. Use impressions from the Search report for the optional scorecard row, keeping Discover separate. Those Search impressions are already included in overall Web performance, so do not add them to the Web total. A missing report or unavailable figure should stay unavailable in your record.
For a report-specific walkthrough, use the Search Console generative AI guide. Its exercise separates property totals, page rows and unavailable values. For server-observed crawler activity, the AI bot page-role guide keeps request counts and page coverage separate from visits and enquiries.
For ChatGPT visits, use the current GA4 tracking steps. For the questions and pages appearing in sampled AI answers, use the monthly citation log. Keep those observations beside the enquiry record.
Use the scorecard in three passes
1. Set the period and source details
On Scorecard, enter the period start, inclusive end date and review date. Record the page, market and channel filters, export dates and GA4 reporting timezone. Search Console uses Pacific Time.
Enter the available search and analytics totals in the amber cells. Leave a missing figure blank. Enter zero only after checking the source. The optional Google Search generative AI row uses the dedicated Search report, not Discover. Keep it beside the overall Web total without adding the figures together.
For the lead key-event row, record a tested completion event and its exact filter. Do not use every button click as an enquiry. Check that the event represents the intended completed action and that an agreed test reaches the receiving business. Google’s key-event guidance explains the analytics check; receipt and business fit need their own checks.
2. Review the actual enquiries
On Enquiry log, use one row per received enquiry. Exclude spam and tests, and update the same record when the conversation progresses.
The fields are:
- Enquiry ID and received date: an internal reference and when the business received it.
- Service needed, fit and reason: what the buyer wants and why it does or does not match.
- Stage: Enquiry, Conversation, Opportunity, Won or Lost.
- Recorded source and source evidence: only what a captured record actually supports.
- Landing page: the known entry page, or Unknown.
- Buyer-reported discovery: what the person says brought them to you.
- Next action and owner: what happens next and who handles it.
Keep recorded source and the buyer’s answer in separate fields. Someone may arrive through ChatGPT after a colleague first mentioned you. Both facts can be useful. Do not overwrite one to make the story simpler, or guess an individual enquiry’s source from an aggregate GA4 report.
The workbook checks for duplicate IDs and missing core fields. Mark log coverage Complete only after you have checked the receipt records. Until the setup and log checks are complete, the enquiry totals remain blank.
3. Choose one action
The scorecard counts enquiries by their received date. Fit and stage reflect the review date. A later win can therefore change an older period’s enquiry outcome; save a dated copy at each review.
The suitable-share calculation excludes enquiries still awaiting a fit decision. Read that pending count alongside the percentage. Won counts refer to those enquiry records, not revenue collected during the reporting month.
The blank log has space for 100 enquiries. Before exceeding that range, start a new copy or extend the formulas and validation together. The workbook does not connect to your accounts or automatically join people across systems.
Worked example: more recorded activity, one clearer decision
This is an illustrative HR advisory business. Every figure and record is fictional, including the matching example in the workbook. It is separate from our client case studies.
| Record | August 2026 | September 2026 |
|---|---|---|
| Google Web impressions | 1,800 | 2,600 |
| Google Web clicks | 45 | 58 |
| GA4 Organic Search sessions | 42 | 52 |
| GA4 ChatGPT sessions | 3 | 6 |
| GA4 completed lead key events | 6 | 8 |
| Received enquiries | 6 | 5 |
| Suitable enquiries | 2 | 3 |
| Not suitable | 3 | 1 |
| Pending fit review | 1 | 1 |
| Reached opportunity | 1 | 1 |
| Won / Lost | 0 / 0 | 0 / 0 |
The enquiry stages are recorded as of 5 October. These calendar months have different lengths, and the counts are small. The example is a decision exercise, not a growth comparison or an attribution result.
The first action is to investigate measurement before increasing spend. September has eight recorded lead key events and five received enquiries. Check the completion trigger and receiving process to establish why the counts differ. The difference alone does not tell you the cause.
The three suitable enquiries deserve attention. The increase from two is one additional enquiry, while one enquiry still needs a fit decision. Follow the opportunities and read the requests before claiming a trend.
One request is for recruitment, which this example business does not offer. Inspect how its HR support page explains scope. If that expectation comes from the page, clarify the service and supporting proof.
The workbook also shows a captured ChatGPT source beside a buyer’s answer that a colleague recommended the firm. The citation CSV uses the same fictional HR context. Neither record joins a sampled answer to a sale.
Turn the reading into a specific next step
| What you find | Next action |
|---|---|
| Recorded completions and received enquiries do not reconcile | Check the event trigger, duplicate firing, date scope and receipt process. |
| Repeated requests for work you do not offer | Inspect the relevant page’s fit, scope and route to contact. |
| Suitable enquiries with no agreed next step | Review follow-up and the questions buyers still need answered. |
| Too few completed observations | Keep the definitions consistent and collect more evidence before expanding the conclusion. |
For the fuller diagnosis, use why a website gets traffic but few useful enquiries. For our delivery and comparison windows, use the AI visibility timeline. A monthly log can support that reading without replacing the agreed review cadence.
Bring the decision to a conversation
Our AI SEO plan connects the work to the service you want to sell and the buyers you need to reach. Measurement helps decide what should change next.
Request your free Clarity Call. Bring the website and the question you need to resolve. You do not need to finish the workbook first. If you prefer to begin with a written public-site review, the free Outside View is optional.