SEO Metrics Explained

Four Kinds of SEO Numbers That Answer Different Questions

A report can put a forecast beside a recorded click and make them look equally certain. Before acting on either, give the number a plain label.

1. Estimates: a calculation of what may be happening, such as a competitor's likely search traffic. They help you choose what to investigate.
2. Recorded activity: an event a system recorded, such as a Google search click or a successful page request. It tells you what happened within that system's coverage.
3. Calculated scores: a summary made from selected inputs and a formula. It helps you compare those inputs under the same rules.
4. Business results: an inquiry, a qualified opportunity or a sale recorded by the business. These tell you how far the work has moved toward a customer.

None needs to impersonate another to earn its place in the report.

Read an SEO Metric Before You React to It

Semrush Traffic Estimates Are Research, Not Your Visitor Counter

Semrush explains that its Domain Analytics traffic estimates use keyword positions, search volume and expected click-through rates. Its Traffic Analytics estimates use clickstream data across traffic sources. Those are different calculations, so they can produce different totals.

That distinction changes the question. Before deciding which total is right, check whether both reports cover the same country, device and period. Then check whether both even claim to measure the same traffic.

Semrush's explanation of its two traffic methods makes the distinction explicit. Estimates help you study websites whose private records you cannot see. They do not become recorded visits because the chart looks precise.

There is an equally important correction to the broad criticism of SEO tools: some reports contain your connected first-party data. Organic Traffic Insights combines Semrush, Google Analytics and Google Search Console data. A number inside Semrush is not automatically an estimate. Read its source.

The same standard belongs in a comparison with Ahrefs, Moz, SpyFu or Similarweb. We have not run a controlled accuracy test of those products here. The fictional traffic totals in Kodi's recording illustrate the problem; they are not vendor benchmark results.

Search Console: Being Shown and Being Chosen Are Separate Results

Google Search Console records impressions and clicks from Google Search. Click-through rate, or CTR, is clicks divided by impressions. Average position summarizes the position of the topmost result from your site within the report's scope.

Those measurements help you find the next page to work on. They do not all describe a visit.

Here is a dated example from AISymantix's own page records:

Source: live Digital Karma Data Warehouse, checked November 7, 2026. These are page totals, not unique people. CTR uses total clicks divided by total impressions; average position is weighted by impressions. The window is fixed, so later results belong in a separate comparison.

The first page was shown more often. The second earned the recorded clicks. That is a reason to inspect their query and page relationships, not declare a winner from impressions alone.

It also does not tell us that four clicks became four customers. Inquiry records answer the next part.

We would look at the queries and their positions before rewriting a title. A page shown far down the results needs a different investigation from a page near the top that people pass over. Google's Performance report guide explains the measures and their reporting scope.

A Visibility Score Needs Its Formula Beside It

A score compresses information. That is its advantage and the reason you need to know what went into it.

A ranking-based index, an impression-weighted score and a technical page score are different calculations. The same number on their displays does not mean the same performance.

A score based on a selected keyword set can also change when that set changes. Removing difficult terms may improve the summary without bringing another visitor. Keep the comparison set and method stable before treating a rising score as progress.

AISymantix's visibility index and CTR guide explains the local calculation. The published CTR-by-position dataset shows the click curve behind that work. Neither is a promise that the next search result will receive those clicks.

For the definition and formula, use What Is Search Visibility? This article's job is to help you interpret a report without confusing its different measurements.

An AI Readiness Score Gives You Fixes to Make

The AI Search Visibility Checker examines the page itself. It helps you find access and clarity problems that you can change.

That is a different job from counting live AI recommendations. A score can show that a page passed more of the defined checks. It cannot establish that an AI service recommended the business afterward.

The AI Readiness Score explanation describes the checks. Use the failed items to decide what to fix, then use separate records to watch what happens next.

Good Bots: A Successful Fetch Is a Discovery Clue

A successful crawler request tells you that a request reached a recorded content page and received a successful response. That can confirm that a new article is being fetched or help explain an access problem after a site change.

Our Good Bots count applies the configured crawler classification and content rules. It excludes failed requests and non-content requests such as images or feeds. Search crawlers, AI search, training and user-requested fetching remain separate purposes in the analysis.

The crawler name alone is a claim, not an identity check. Nor does the count establish indexing, training use or an AI citation.

When crawler activity and search impressions move together, we have something to investigate. We have not established that one caused the other. The five-site AI bot visibility study puts those measurements beside one another so that relationship can be tested.

AI Citations: Read the Answer, Not Just the Link Count

An observed citation records a source link in a tested AI answer. The prompt, service and date belong with that record because the next answer may differ.

Then read what the answer says. A link to the right business beside the wrong service description is a citation with a meaning problem. A citation count on its own misses that.

This is where AISymantix's focus on entity architecture becomes concrete: can the AI tell which business it is describing and what that business actually offers?

Google also includes traffic from AI Overviews and AI Mode in Search Console's overall Web reporting, according to its AI features guidance. The total Web figure by itself does not tell you which exposure came from an AI feature.

An AI referral visit is another record again. It carries an observed referrer or campaign label. It is not a count of all AI citations, and a missing referrer does not mean no AI discovery occurred.

An Inquiry Is Progress. A Sale Is a Later Result.

A received form says someone submitted a question. A qualified inquiry says the business reviewed it and found a relevant opportunity. A sale needs its own record.

Keep those stages separate and the report starts helping you decide where to spend your time. A page may earn fewer visits yet bring better questions. That possibility needs inquiry data, not a guess based on the visitor total.

AISymantix's measurement page explains how visits, completed checks and inquiries are counted. The search visibility-to-inquiries study follows the next step after exposure.

Google's guide to combining Search Console and Analytics also explains why search clicks and analytics sessions can differ. Do not force them to match by renaming one of the counts.

The SEO Number Should Point to the Next Decision

If a competitor estimate changes, investigate the market. If search exposure grows without clicks, inspect the queries and pages. If people arrive but do not inquire, look at the offer and the next step. If AI describes the business incorrectly, work on the information it is using.

That is what makes the report worth opening. It points to a decision you can test, then keeps the record of what happened afterward.

For a page you control, run the free checker and read the specific findings. For a review that connects the page with search and discovery records, see what the AI Symantix Audit covers.

If the decision is whether to keep paying for an SEO subscription, the BeforeYouBuyX free-first buying guide handles the cost question. For a larger discovery network, AI Website Systems explains the shared reporting system. That full setup is usually overkill for one or two sites; its value becomes clearer when the network and the work grow more complicated.

Krisada owns AISymantix, BeforeYouBuyX and AI Website Systems. Those related resources address different decisions. This article's conclusion is the same whichever tool you use: know what the number means before you decide what to change.