Define the business outcome before choosing metrics
The phrase “more leads” is too vague to guide search strategy.
Raw inquiry
Any recorded call, form, chat, booking, trial, or purchase event. This count often includes spam, duplicates, wrong numbers, vendors, job seekers, and existing customers.
Valid lead
A real person or organization seeking something the business offers, even if timing, location, value, or fit is not yet known.
Qualified lead
A valid lead that meets agreed criteria such as service, market, budget range, urgency, project type, business size, or decision role.
Opportunity
A qualified lead accepted into the sales process with a defined potential value or next stage.
Won outcome
A completed sale, signed engagement, booked job, retained customer, or other agreed business result.
Create a tracking map for every conversion path
Each action needs an event, destination, owner, and test procedure.
| Action | Evidence to record | Quality follow-up |
|---|---|---|
| Phone call | Clicked number, call source where approved, connected call, duration | Call outcome, service need, location, booked next step |
| Contact or proposal form | Successful server submission, form type, landing page, campaign context | Spam status, response, qualification, opportunity stage |
| Booking | Completed booking and appointment type | Attendance, qualification, reschedule or cancellation |
| Purchase or lead transaction | Completed order, product or service, value where permitted | Return, cancellation, repeat purchase, margin context |
| Tool or report | Tool started, email captured, report generated, proposal action | Whether the report user becomes a valid conversation |
The event fires only after a successful action, not when a button is merely clicked.
The form success state is visible and notification delivery is monitored.
Phone links are tappable and the displayed number is correct.
Cross-domain booking or payment flows preserve approved attribution data.
Consent, privacy, retention, and sensitive-data requirements are documented.
Test submissions are labeled and excluded from production reporting.
Use a measurement chain instead of one headline number
Each layer answers a different question and helps diagnose where the journey is breaking.
| Layer | Core question | Example measures |
|---|---|---|
| Implementation | Did the planned work launch correctly? | Pages published, fixes validated, links added, tracking tested |
| Technical health | Can priority pages be accessed and interpreted? | Index status, crawl errors, canonical quality, template performance |
| Visibility | Are relevant pages appearing for useful demand? | Impressions, query coverage, landing-page visibility, local presence |
| Engagement | Do visitors appear to find the page relevant? | Clicks, engaged visits, paths, tool usage, key page interactions |
| Conversion | Do visitors take the intended action? | Calls, forms, bookings, trials, purchases, assisted actions |
| Quality | Are those actions commercially useful? | Valid leads, qualified rate, opportunities, booked jobs, sales feedback |
| Business contribution | What value is associated with the observed outcomes? | Pipeline influence, revenue context, cost per qualified lead, customer value |
Treat attribution as a model, not certainty
Customers use multiple searches, devices, channels, recommendations, and offline interactions. Most reports observe only part of that journey.
Last-click view
Useful for understanding the final recorded source, but it can understate earlier research, branded validation, AI answers, referrals, and offline influence.
First-touch view
Highlights discovery but may over-credit an early visit that was not decisive.
Assisted view
Shows channels and pages that contributed before the final action, although identity, consent, and device gaps remain.
Sales-confirmed view
Adds customer language and opportunity outcome from the CRM or team. This is often the most valuable context even when it cannot reconstruct every touchpoint.
Directional testing
Market, page, or time-based comparisons can strengthen confidence when perfect attribution is impossible, but they need documented assumptions and outside influences.
Build a decision-focused monthly scorecard
A useful report explains what changed, why the team believes it changed, what the business learned, and what happens next.
State the business goal and qualified-lead definition at the top.
Separate branded and non-branded search where the data allows.
Show priority pages and query groups rather than only sitewide totals.
Report calls and forms by validity and qualification status.
Explain completed work and which metrics it was intended to influence.
Annotate launches, outages, promotions, seasonality, tracking changes, and major external events.
Identify wins, losses, uncertainty, and required investigation.
End with a small number of prioritized decisions, owners, and validation steps.
Use the chain to diagnose performance
The same traffic trend can require very different responses.
| Observed pattern | Likely questions | Next investigation |
|---|---|---|
| Impressions rise, clicks do not | Is the page ranking too low, appearing for the wrong intent, or presenting a weak result? | Queries, position ranges, titles, descriptions, SERP features, page intent |
| Clicks rise, conversions do not | Is the traffic relevant and is the landing experience clear? | Query intent, mobile usability, offer, proof, CTA, form errors |
| Conversions rise, quality falls | Did targeting broaden or qualification weaken? | Lead sources, pages, search terms, service area, spam, sales disposition |
| Qualified leads rise, wins do not | Is the issue sales response, pricing, capacity, competition, or customer fit? | Response time, call quality, proposal stages, objections, lost reasons |
| All metrics fall suddenly | Did tracking, indexing, availability, demand, or the site change? | Analytics health, Search Console, releases, server logs, seasonality, profile status |
A practical implementation sequence
Build the smallest trustworthy measurement system first, then add sophistication where it changes decisions.
Agree on definitions
Document valid lead, qualified lead, opportunity, and won outcome with marketing and sales.
Inventory conversion paths
List every phone, form, chat, booking, purchase, and tool flow, including third-party destinations.
Test the evidence
Submit and call from multiple devices, verify website states, database records, emails, analytics events, and CRM creation.
Connect quality feedback
Add the simplest reliable sales disposition process before building an elaborate attribution model.
Create the baseline
Record recent performance, data gaps, seasonality, and major business changes.
Review and improve
Use the monthly scorecard to prioritize fixes and content based on qualified outcomes, not only visibility.
Primary sources and further reading
Platform behavior can change. These official sources support the platform-specific statements in this guide and provide the best place to verify current requirements.
Frequently asked questions
What is a good SEO conversion rate?
There is no universal benchmark. Page intent, offer, industry, device, market, brand familiarity, and the definition of a conversion all change the rate. Compare relevant page groups and prioritize qualified outcomes.
Should phone-call duration determine lead quality?
Duration can help triage calls but is not a reliable definition by itself. A short call can be highly qualified and a long call can be irrelevant. Use call outcomes or sales disposition when permitted.
Can SEO revenue be tracked perfectly?
Usually not. Identity, privacy, devices, offline interactions, long sales cycles, and multi-channel journeys create gaps. Use transparent models, CRM outcomes, and directional evidence rather than false precision.
How should AI-search traffic be measured?
Record referrals where platforms expose them, branded search and assisted journeys where useful, and a controlled question set for observed mentions and citations. Keep platform and sampling limitations visible.