Visibility-to-enquiry measurement template for industrial teams

Agree each event and who owns it

Each record needs a definition, a reliable source and an accountable owner.

Commercial records own commercial decisions

Analytics records behaviour. The authorised commercial system owns qualification, opportunity and revenue state. A browser event firing cannot overwrite those decisions.

For Plact's acquisition, a demo request and a confirmed demo booking are different events. For a supplier, an RFQ, a quotation and an accepted order are different. Retain the terms your team actually uses.

If your team uses Google Analytics

GA4 documents lead events including generate_lead, qualify_lead, disqualify_lead, working_lead, close_convert_lead and close_unconvert_lead. Match the documented meaning to your approved workflow before mapping them. In particular, contact with a representative is not the same as accepting an opportunity, and a converted-lead event is not an accounting record. Google recommended events.

Names alone are not instrumentation. Each event needs an exact trigger, source, duplicate rule and accountable owner. This template specifies no new event implementation.

Sources: Google Analytics Help

Keep the context, not the buyer's private message

For a measured resource or event, record approved public identifiers and controlled categories:

Keep names, email addresses, message bodies, drawings, filenames and confidential RFQ details out of analytics payloads and URLs. Google Analytics warns against identifiable data in those values. Opaque IDs still need an approved privacy and access design. Google Analytics privacy guidance.

• Resource ID, canonical URL, version and release date.
• Buyer task, such as inspect proof, compare supply options or send a requirement.
• Offering or capability ID and known market context.
• Public evidence type and document revision when relevant.
• Observed referrer class, source page and permitted join identifier.
• Qualification reason in the authorised commercial workflow, not automatically in analytics.

Sources: Google Analytics Help

What the provider reports actually mean

Google's 3 June 2026 announcement introduced separate generative-AI impressions reports for Search and Discover, initially for a subset of websites. The Search view measures AI-feature impressions, while ordinary Web reporting still includes AI Search activity; the Discover view has its own scope. Do not label ordinary Web clicks as AI clicks or infer an AI click-through rate from impressions alone. Availability must be checked on the actual property. Google announcement, report help.

Bing AI Performance aggregates citation activity across supported Microsoft and selected partner experiences and exposes sampled retrieval phrases. Those records are neither a complete prompt panel nor a lead report. Bing explanation.

OpenAI documents utm_source=chatgpt.com on ChatGPT search referral URLs. An observed tag is useful arrival evidence, not a complete history: the buyer may return directly, share a link or contact sales offline. OpenAI publisher FAQ.

These providers expose different populations. Their figures do not share a denominator and should not be added into one citation or conversion rate.

Sources: Google Search Central; Google Search Console Help; Bing Webmaster Blog; OpenAI Help

Worked example: synthetic, not Plact or customer data

Assume a hypothetical 28-day cohort of 1,500 consent-eligible sessions. Each has one observed arrival class. Distinct server-accepted enquiries are joined to those sessions, with repeated submissions for the same request deduplicated. Qualification is assessed 14 days after each request; opportunities at 30 days. This example is read after every included request has had the full 30-day follow-up window. Unresolved reviews stay pending. No revenue data is available. Newsletter signup is a separate action on this hypothetical site, not a required step before an enquiry.

The enquiry states reconcile: 21 qualified + 25 rejected + 18 pending = 64. Opportunities are a later subset of qualified enquiries, not eight additional enquiries.

Two legitimate questions have different denominators:

• Qualified enquiries per observed session: AI 4/100 = 4.0%; organic 12/900 = 1.33%; direct/unknown 5/500 = 1.0%.
• Qualification among reviewed enquiries: AI 4/6 = 66.7%; organic 12/28 = 42.9%; direct/unknown 5/12 = 41.7%. Eighteen pending reviews remain visible and are excluded from these rates.

Check landing-page mix before comparing channels

Do not conclude that AI caused better enquiries. In this invented example, arrival types have different landing-page mixes:

Commercial-page rates are identical. Most AI arrivals reached commercial pages; most organic arrivals reached education. The aggregate difference is compatible with that mix, not proof of a superior channel effect. Unobserved buyer differences and small counts remain. One additional or missing qualified AI enquiry changes its session rate by one percentage point. Zero out of 20 does not establish that a path never works.

A defensible summary is: “Four qualified enquiries were associated with 100 identified AI-referral sessions. Two became accepted opportunities by the review date, while two other AI-associated enquiries remain pending. Landing-page mix differs from organic search. Incremental impact and revenue are unestablished.”

“AI generated better leads” and “citations created eight opportunities” are not supported by these records.

Copy this resource review row

Keep the raw counts, missing evidence and the decision the review supports.

Attribution classes describe evidence, not causal credit. One opportunity may have several touches. Do not add assisted and direct counts and call the sum unique leads. Keep “I found you through AI” self-reports separate from verified referral records.

If a request cannot be joined safely, count it as unjoined. Do not infer identity from an IP address, email or message to fill the gap. Keep unobserved or non-consenting journeys outside the observed cohort and report coverage when known, without inventing a correction factor.

Where Plact fits

Search to Enquiry connects buyer-facing information with direct enquiry readiness. Where measurement is expressly agreed and connected, Plact can explain what changed, what was observed and which commercial records can be linked. Missing measurement should remain visible. Your team retains qualification and sales decisions.

The template is useful whether your team runs the work or buys managed help. Book a demo if you need Plact to manage the discovery-to-enquiry work, and ask to see the definitions, permissions and unknown states. A live integration or revenue join must be demonstrated; this guide does not promise one for every business.