Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

SEO › SEO Measurement

SEO KPIs and Reporting

SEO KPIs that connect visibility to outcomes: the KPI chain, segmentation, baselines and targets, attribution, annotations and reports that decide.
Guideintermediate8 min readUpdated

The KPI chain

LinkIndicatorsSourceWhat it diagnoses
CoverageImportant pages indexed; pages excluded by reason; new pages' time to indexSearch Console (GSC) Page IndexingTechnical blockage
VisibilityImpressions and average position for priority clusters; share of clusters with a page in the top positions; results features wonSearch Console, rank trackingRelevance and authority
ClicksNon-brand organic clicks by cluster and page type; click-through rate (CTR) by position bandSearch ConsoleSnippet quality; result features
EngagementEngaged sessions, scroll, next-page clicks from organic landing pagesAnalytics (GA4)Intent match; page quality
OutcomesConversions, leads, revenue, assisted conversions from organic landing pagesAnalytics, CRMCommercial value
InputsPages published or refreshed; links earned to target pages; technical fixes shippedProgramme records, backlink toolsWhether the work is happening

Reading the chain top to bottom locates problems: visibility without clicks is a snippet or result-feature problem; clicks without engagement is an intent problem; engagement without outcomes is a page or offer problem; nothing moving with inputs flowing is a time or competition question. The objectives and KPIs guide explains the general discipline of chaining indicators to objectives. Where the outcome row is enquiries, the guide to organic search for lead generation narrows it to qualified leads by landing page. Where AI-generated answers on the results page reduce clicks while visibility holds, the guide to AI in search results covers reading and reporting the chain.

Example KPI tree from objective to SEO indicators

Segmentation

Totals hide everything. The segments that make SEO data readable:

  • Brand versus non-brand. Brand queries measure brand demand, which SEO rarely drives; non-brand measures the programme. Define the brand terms once and apply the filter everywhere. Brand demand still matters to the brand team, and the tutorial on tracking share of search reads it against competitors' brand searches rather than as SEO performance.
  • By cluster from the keyword map: each priority cluster's owner URL and head term, so progress is read where the work was done.
  • By page type: category, product, guide, location, hub. Template changes move whole types.
  • By intent: informational traffic and transactional traffic have different volumes, click-through and conversion, and should never be averaged together.
  • By country and device where the site serves more than one.
  • New versus existing pages, so that growth from new content is separated from improvement on old.

Baselines and targets

A baseline is the segment's value over a period long enough to cover seasonality — the same months last year where the data exists, or a trailing period with the seasonal pattern noted. A target is derived from the gap, per cluster: the achievable position given the results page's competition, the clicks that position would yield from the cluster's impressions at a realistic click-through for that position band and result features, the conversion rate the page type achieves, phased over the months that rankings take to move. Summed, that is the programme's forecast, with assumptions listed. Targets handed down as a flat growth figure are met by reporting brand traffic, which is why the segmentation comes first.

Attribution and the measurement boundary

Organic search is frequently a first or middle touch. Last-click models (Attribution) in analytics credit the channel that closed, which understates search; data-driven and assisted views credit more generously. Report under one model, show the assisted figure beside it, and explain the difference once rather than every month. Assisted Conversions are the usual form of that figure: conversions in which organic search appeared on the path without being the last touch. Offline and cross-device outcomes are outside the data; say so. Campaign tagging (UTM) on links the site controls keeps paid and email traffic from leaking into organic, which is the most common cause of inflated organic figures. The measurement plan guide covers the wider plan.

Annotations

Every movement in the chain has a cause, and the report's job is to name it. The annotations log records site changes (releases, template changes, content published or consolidated, redirects), programme actions (links earned, briefs delivered), external events (campaigns, press, seasonality, outages), and known search engine updates with dates. Charts carry the annotations. A report without them is a weather report.

Report structure

A monthly report that leads to decisions has, in this order:

  1. Summary: the three things that matter this month and the decisions requested, in a few sentences.
  2. The chain by segment: coverage, visibility, clicks, engagement, outcomes for non-brand, by cluster and page type, each against baseline and target, each annotated.
  3. What changed and why: the annotated movements, including the ones that went the wrong way.
  4. Inputs delivered: pages, links, fixes, against the plan.
  5. Risks and issues: indexing problems, technical debt, competition, dependencies on other teams.
  6. Next actions: specific, owned, dated.

An appendix carries the detail; the body carries the argument. The weekly version is the operational subset (coverage, vitals, priority positions, issues); the quarterly version adds cluster-level progress against the map and the plan for the next quarter.

Report structure demonstration

Illustrative example — not a real result

A monthly report that changed a decision

A B2B software company's report shows non-brand clicks to the product-comparison cluster rising with impressions and position; engagement on those pages is good; trial sign-ups from them are flat. The chain locates the problem below engagement: the comparison pages end without a path to the trial, and the one link present goes to a generic page. The report's requested decision is a template change adding a contextual next step, owned by the web team, with the sign-up rate from the cluster as the indicator to watch. The following quarter's report shows whether it worked, against the same baseline. Nothing in the report mentions total organic traffic.

Leading and lagging indicators

Outcomes lag the work by months, which is why a report that shows only outcomes cannot tell whether the programme is on track. The chain supplies the leading indicators: inputs delivered (pages, links, fixes) lead coverage; coverage leads visibility; visibility leads clicks; clicks lead outcomes. A programme whose inputs are flowing, whose coverage is clean and whose visibility on priority clusters is rising is on track even while outcomes are flat, and the report says so with the expected lag. A programme whose visibility is flat after two quarters of inputs has a problem the lag does not explain, and the report says that too. Reading the chain in order is what makes the monthly report an early-warning system rather than a scoreboard. Where the programme was approved in stages, these leading indicators are the natural review criteria, as the guide to building an SEO business case explains.

Forecasting

A forecast is the sum of the per-cluster targets over time, and it is only as good as its assumptions, which are listed with it: the achievable position per cluster and why, the click-through assumed for that position given the result features, the conversion rate assumed from the page type's history, the months assumed for rankings to move, and the seasonality applied. Forecasts are revised quarterly as clusters arrive or stall, and the revisions are explained. A forecast presented as a single number with no assumptions is a promise; one with assumptions is a plan that can be discussed. The SEO forecasting guide builds such a forecast end to end, from a seasonal baseline and click-through measured on the site's own queries to low, central and high scenarios tracked against actual.

Reporting a loss honestly

Rankings fall, updates move the results page, a competitor publishes better pages, a migration goes wrong. The report that survives these is the one that names the loss on the first page, segments it (which clusters, which page types, which countries), dates it against the annotations, states the most likely causes in order of evidence, and proposes the investigation or the fix with an owner. A report that buries a loss under brand traffic growth is found out within a quarter and never trusted again. The measurement hub sets out what the data can and cannot prove, which is the frame for an honest explanation.

The definitions sheet

Every KPI in the report has a written definition kept with it: the source, the filters (the brand term list, the country, the device), the date range convention, the attribution model, the segment boundaries, and the date the definition was last changed. The sheet is what lets a new analyst reproduce the numbers, what finance checks against, and what prevents the quiet redefinition that makes a "recovery" out of a filter change. It is a page, not a project, and it is the first thing written when the reporting is set up.

Reporting to different audiences

  • The team needs the operational detail and the cluster-level data to act.
  • Marketing leadership needs the chain, the forecast, the risks and the decisions.
  • Executives need outcomes against objective, the trend, and what is required of them.

The same data serves all three at different depths; building three reports from one dataset is the usual arrangement. The SEO KPIs expertise page describes the author's framework and includes an interactive KPI tree; the Search Console tutorial covers the source most of the chain depends on.

Before reporting a loss, check that it is real; the troubleshooting page on rankings that dropped suddenly includes the tracker and reporting changes that mimic a drop.

When organic click-through appears in a report, pair it with position as described in reading organic CTR with the CTR calculator so readers do not compare unlike results.

Where traffic is the business objective rather than a step towards leads or sales, the website traffic objective hub sets organic visits beside paid and social routes on the same engaged-session measures.

To decide which links of the chain come from Search Console and which from analytics, the Search Console versus GA4 comparison sets out what each tool measures before and after the click.

When organic visits are one branch of a site-wide traffic goal, the KPI tree for website traffic splits organic clicks into search demand, visibility share and click-through so seasonal swings are not reported as wins.

Common mistakes

  • Reporting totals. Brand and seasonality do the talking.
  • Position as the headline. Average position across all queries is meaningless; position per priority cluster is a KPI.
  • No baseline. Every number is "up" or "down" from nothing.
  • Unannotated charts. Every movement a mystery.
  • Revenue under an unstated model. Credibility lost when finance checks.
  • Charts without decisions. The report is read and nothing happens.
  • Changing definitions. A KPI redefined mid-year cannot be tracked.

Frequently asked questions

Is organic traffic a good KPI?

As a total, no: it mixes brand demand the SEO work did not create with non-brand demand it did, and it rises and falls with seasonality and campaigns. Non-brand organic clicks to the clusters the programme targets, against a seasonally comparable baseline, is the useful version. Traffic is also a middle link in the chain; the outcome is what the visits produce.

How do I report SEO revenue when attribution gives credit to other channels?

Report under a stated attribution model, show the organic-assisted figure alongside the last-click one, and explain the difference once. Organic search is often the first touch and rarely the last, so last-click understates it; claiming all assisted revenue overstates it. The honest position is both numbers with the model named.

What is a realistic target?

One derived from the gap: for each priority cluster, the current position and click-through, the position achievable given the competition on the results page, and the clicks that position would bring from the cluster's impressions. Summed across clusters and phased over the time rankings take to move, that is a target with a rationale. A percentage pulled from last year is not.

Terms used on this page

Frameworks

Where this page is used

Part of these learning paths

Sources

  1. Performance report (Search results) — Google Search Console Help

Related