The performance of an AI-built landing page should answer three questions: Are the intended users arriving? Are they actually making inquiries? Do those inquiries generate new leads that meet business requirements? Traffic helps explain visits, inquiries indicate willingness to act, and qualified leads indicate lead acquisition quality. Page launch speed and the business results a page generates need to be evaluated separately.
The starting point for measuring landing page performance is to document what each metric counts, what triggers it, how duplicates are handled, and the observation period. Start by establishing the sequence of “landing page entry sessions → sessions with inquiries → new leads → qualified leads,” then analyze channels and page versions. These stages use different counting units, so their numbers cannot all be treated as a declining funnel representing the same group of people.
These three metrics answer different questions. Page views (PV) describe how many times a page is viewed, unique visitors (UV) describe the distinct visitors identified during a reporting period, and sessions describe continuous browsing activity. The Alibaba Cloud analytics documentation defines these metrics separately and explains that multiple views by the same visitor increase PV, while UV is deduplicated within the reporting period.
To evaluate how effectively a landing page handles incoming traffic, use “sessions that begin on this page” as the primary traffic baseline, with PV and UV as supplementary metrics. This reduces confusion between repeated browsing within the site and new arrivals. If the tool only provides page view data, clearly state that limitation in the report, use the available measurement rules initially, and add session entry tracking later.
UV should not be interpreted as an exact count of real people. Situations such as cross-device browsing and changes in browser identifiers should be documented in the tool’s measurement definitions. Do not force a one-to-one relationship between anonymous visitors and contacts who have submitted their details. When comparing different periods, also use consistent time zones, domain scope, and rules for excluding internal visits.
2. How Do You Create a Metrics Table the Whole Team Can Use?
The following suggested definitions suit landing pages intended to generate appointments, quote requests, or demo applications. Marketing, sales, and the colleagues responsible for data should agree on them before launch. Record the effective date of any subsequent changes.
| Metric | Suggested Counting Unit and Trigger | Main Exclusion or Deduplication Rules |
|---|---|---|
| Page views | Number of times the landing page is opened | Flag internal testing and identified abnormal visits |
| Entry sessions | Sessions that begin on the target landing page | Count according to the analytics tool’s session rules |
| Inquiry entry point clicks | Events involving clicks on online chat, phone, or appointment buttons | Retain both event counts and the number of sessions with clicks |
| Inquiry sessions | Landing page entry sessions in which a substantive inquiry occurs | Count multiple messages within the same session once; list inquiries that cannot be linked separately |
| New leads | Records entering the lead database for the first time that can be used for subsequent contact | Deduplicate using agreed contact details; list existing customers separately |
| Qualified leads | New leads that meet business eligibility criteria after review | Exclude duplicates, spam, and clearly unsuitable requests |
Retain both the raw event table and the deduplicated lead table so differences in the numbers can be explained. One inquiry may contain multiple messages, or it may leave no contact details. The same contact may also submit repeatedly through different entry points, so inquiry volume and new lead volume do not have to match.
3. Does Clicking an Inquiry Button Count as an Inquiry?
Record a click as an “inquiry intent action.” Record an actual business message, a connected phone call in which a need is expressed, or a successfully submitted appointment request as a “substantive inquiry.” For forms, use confirmation that the server has successfully saved the submission. Record button clicks and frontend success messages separately to help troubleshoot submission failures.
Each entry point should have its own confirmation conditions: for live chat, check whether the visitor sends a business-related message; for phone inquiries, check whether the call connects and a need is recorded; for appointments, check whether the request is successfully created. Merely opening a chat window or dialer, or copying contact details, does not establish that contact has been made.
Cross-platform inquiries especially need to be listed separately. If a user clicks through to another app and subsequent inquiry records are unavailable, report “outbound clicks” and “subsequent inquiries confirmed manually.” Do not use click counts to fill gaps in inquiry data. Customer service should also retain source information, link it where possible, and mark it as unknown where a link cannot be established.
4. What Counts as a Qualified Lead, and Who Confirms It?
Qualified leads need explicit business eligibility criteria. For business services, these might include usable contact details, a need within the scope of the service, a customer within the target market, and a sales assessment that there is a reason to continue the conversation. Whether budget and purchasing timeline are required should depend on the business. Do not classify every inquiry without a confirmed budget as unqualified.
A Jiandaoyun article on lead management discusses the distinctions between qualified leads, qualification assessment, deduplication, and status logs. In practice, translate your company’s criteria into fields and review actions instead of giving sales only a vague “high or low quality” option.
Suggested statuses include “pending confirmation,” “qualified,” “not a fit,” “duplicate,” “unreachable,” and “existing customer.” Also retain the reviewer, decision time, and reason. Keep “unreachable” separate from “clearly not a fit,” as the former may still require follow-up. A qualified lead is not yet a sales opportunity, much less a closed deal; move it forward only after it meets the agreed conditions for the next stage.
For deduplication, first identify contacts using standardized phone numbers or email addresses, then associate them with companies or projects. Different contacts at the same company may have different needs, so do not merge all records based solely on the company name. Retain activity records for return visits, repeat quote requests, and new projects while distinguishing new leads from existing contacts.
5. Which Denominator Should You Use for Conversion Rates?
The denominator should match the question, and its full name should appear in the report. Start by standardizing the following formulas:
- Session inquiry rate = landing page entry sessions with substantive inquiries ÷ landing page entry sessions.
- Lead qualification rate = confirmed qualified leads within the same new-lead cohort ÷ all new leads in that cohort.
- Qualified lead yield per session = attributable qualified leads ÷ corresponding landing page entry sessions.
- Cost per qualified lead = lead acquisition cost within the same scope ÷ corresponding qualified leads.
“Qualified lead yield per session” reflects output density. It cannot be interpreted directly as the probability that a visitor converts, because sessions and contacts are different counting units. The lead qualification rate should also show the number awaiting confirmation; otherwise, differences in review progress will affect channel comparisons.
Suppose a page receives 1,000 entry sessions in one observation cohort, with inquiries occurring in 40 sessions, producing 20 deduplicated new leads and ultimately eight confirmed qualified leads. All of these figures are hypothetical and serve only to demonstrate the calculations. The session inquiry rate is 40 ÷ 1,000, the lead qualification rate is 8 ÷ 20, and the qualified lead yield per session is 8 ÷ 1,000. These results answer questions about behavior, quality, and overall output, respectively.
Do not directly multiply these metrics to extrapolate results, because sessions and contacts do not have an inherent one-to-one relationship. When there are zero qualified leads, display the count and total cost, and mark the unit cost as not calculable. Do not display it as zero cost.
6. How Should You Handle Channel Attribution and Conversions Across Reporting Periods?
Retain the landing page, channel, campaign identifier, page version, first identifiable source, and source at submission in the lead record. Then select one primary attribution rule for routine comparisons. Use the first source to analyze initial contact and the source at submission to analyze the final interaction leading to submission, displaying these two views separately.
Build a showcase site and grow leads in minutes
Describe your idea once, and We0 AI can generate a showcase site, pages, and CMS, then help you attract customers and traffic after launch.
One complete project generation for free registration
Best for trying one complete generation flow and seeing a first project draft quickly.
The Jiandaoyun article lists consistent time windows, channel attribution, and status logs as steps for implementing data tracking. For small teams, applying one rule consistently is more useful than simultaneously enabling several models that cannot be explained. Keep visits without source parameters as unknown or direct traffic; do not fill in the source arbitrarily based on sales impressions.
Reports should provide two views: the operational view asks, “How many new leads were added this period, and how many leads were confirmed as qualified this period?” The page evaluation view asks, “What results did a particular cohort of page entry sessions and new leads produce within the agreed observation window?” A record submitted last month and confirmed as qualified this month can count toward this month’s workload while remaining assigned to its original new-lead cohort.
When comparing channels, ensure their observation windows are equally complete. A newly launched page with many leads still awaiting confirmation should not be compared directly with a historical page whose reviews are complete. Businesses with long sales cycles should continue tracking opportunities and closed deals, but should not attribute every subsequent change to the landing page.
7. How Do You Connect Event Tracking, Forms, and Sales Records?
A minimum viable data workflow must make it possible to cross-reference page events, submission records, and sales statuses. You do not need a complex system from the outset, but you should identify who owns each type of record and where it is stored.
- List the key events: page entry, inquiry entry point click, actual inquiry, successful form submission, and lead status change.
- Generate a unique submission identifier for each successful submission, and record the landing page, page version, submission time, and source information permitted to be retained.
- Preserve the relationship between lead identifiers and submission identifiers in the lead table. Link repeat submissions to the original contact without overwriting the original records.
- When sales updates a status, save the decision time, reason, and person responsible so qualified leads can be traced to specific records.
- Use test records to complete successful submissions, failed submissions, duplicate submissions, and inquiries across entry points, then check each item against the reports.
Recommended acceptance criteria are to trigger a success event only after the server successfully saves a submission and to prevent duplicate counting with submission identifiers. Whether this can be implemented must be confirmed by the platform or developers actually involved. Do not assume the capability exists simply because the page has a form.
When collecting contact details, collect what the business needs and explain its purpose. Internal identifiers are sufficient for linking analytics reports; avoid placing phone numbers or email addresses directly in URL parameters. The Alibaba Cloud documentation also explicitly states that its analytics count production traffic and exclude development previews. Confirm the environment scope for other tools as well. Source
8. When Data Looks Abnormal, Should You Adjust Traffic or the Page First?
First identify the stage where the change occurs, then decide what to do. Break down the data by channel, device, and page version, and check that data collection is working correctly.
- Entry sessions increase, but inquiry sessions do not: Check whether the traffic matches the target audience and whether the page’s promise is consistent with the channel messaging.
- Inquiry button clicks increase, but substantive inquiries do not: Check redirects, customer service handling, and business hours, and rule out entry point failures.
- New leads increase, but qualified leads do not: Check for spam submissions, mismatched needs, duplicate records, and sales review progress.
- Qualified leads increase, but opportunities progress slowly: Review response speed, needs assessment, and subsequent proposals to identify issues in the sales process.
Before each change, write an observable hypothesis. For example: “Target customers do not understand the implementation scope, so clarify the delivery boundaries before the form.” Retain the version and modification time, and monitor both inquiries and qualified leads. Avoid declaring an optimization successful solely because clicks increase.
When traffic is limited, use customers’ actual inquiry wording and failed submission records to help identify problems. When comparing page versions, keep channels, audiences, device mix, and observation windows as similar as possible. If advertising creative changes at the same time as a major page revision, state that the page changes’ impact cannot be isolated.
9. FAQ
Does High Traffic Mean an AI-Built Landing Page Is Effective?
It only shows that the page received visits. You also need to determine whether entry sessions generated substantive inquiries, whether those inquiries produced new leads, and whether those leads met business criteria. Repeated growth in PV may also come from multiple views by the same visitors.
If One Person Submits a Form Twice, How Many Leads Does That Count As?
Retain both submission records and merge them into one new lead according to the agreed contact identification rules. If the second submission concerns a separate new project, create a separate needs record while retaining its relationship with the original contact.
Can We Start Tracking Without a CRM?
Yes. Start with a controlled spreadsheet that records submission identifiers, sources, lead statuses, decision times, and responsible owners. Ensure someone maintains it, change history is accessible, and records are regularly reconciled against form and customer service records before considering a more systematic solution.
What Is a Good Inquiry Conversion Rate?
A universal benchmark is inappropriate without considering the business type, channel, average transaction value, and definition of an inquiry. First compare your own historical performance and page versions using consistent measurement rules, then assess the results alongside qualified lead volume, costs, and subsequent progress.
Related Tools
The we0 website describes a workflow covering natural language requirements, multi-agent website building, visual adjustments, and domain deployment, making it a tool to consider for building and iterating landing pages. Adding this article’s event list and acceptance criteria to project requirements helps teams plan page launch and performance evaluation together. The specific methods for data collection, linking external inquiries, and CRM integration should be confirmed against the project implementation plan.
Web analytics tools record pages and sessions, form or customer service tools confirm actual actions, and a CRM or controlled spreadsheet manages deduplication and qualification. When selecting tools, focus on whether data export, unique identifiers, source retention, and status history meet your requirements.
Related Links
- Alibaba Cloud: Website Analytics Definitions for PV, UV, Sessions, and Environment Scope
- Jiandaoyun: Metrics, Deduplication, and Time Windows for Converting Leads into Opportunities
- we0 Website: Website Building, Adjustment, and Publishing Workflow
Summary
To measure landing page performance, first standardize the definitions of visits, inquiries, and qualified leads, then connect page events, submission records, and sales statuses. Use sessions to evaluate how effectively the page handles incoming traffic, deduplicated leads to measure new lead acquisition, and qualification decisions to assess quality. Consistent attribution rules and complete observation windows are necessary to use these findings to guide page iterations, channel investment, and sales follow-up.
