HELP CENTRE / UNDERSTANDING AN AUDIT

Website Health Score,
explained.

A transparent summary of the checks DeepSpider evaluated. Learn why an issue affects the number, why two audits may differ and what the score cannot tell you.

Current method: Scoring v2Guide updated:
01 / THE PURPOSE

What the number measures

The Website Health Score is a 0–100 summary of the audit checks that had enough evidence to be evaluated. An Error or Warning can reduce it according to the affected share of that check’s records and the check’s impact. It is a DeepSpider product metric, not a Google score or a prediction of rankings.

A high score can still need attention.

Review the actual findings, the proportion of affected records and the audit’s coverage. Informational insights and editorial review opportunities may matter even though they do not reduce the score.

4 categoriesTechnical · Structure · Content · Metadata
2 severitiesErrors and warnings contribute; Errors weigh more
0–100One decimal displayed, full precision used for calculations

Category weights

35%Technical
25%Structure
25%Content
15%Metadata

The overall score is a weighted average of the category scores. If a category has no evaluated scoring checks, it is omitted and the remaining category weights are proportionally rescaled. An unchecked check is never treated as passed.

02 / SCORING METHOD

How errors and warnings affect the score

The current scoring method gives every evaluated Error or Warning check a penalty based on how widely that issue appears. Each penalty reduces the remaining health of its category. Passing checks contribute a factor of one, so adding a clean check does not dilute an existing failure within that category.

PER CHECKaffected share = min(1, affected ÷ checked)check penalty = affected share × severity × impact factorPER CATEGORYcategory score = 100 × product(1 − each check penalty)

Product means multiplying together the remaining-health factors from the evaluated scoring checks. More affected records or another independently failing check lowers the score. The multiplication keeps the result between 0 and 100 and gives each further issue a smaller additional point loss.

v2 multipliers
Finding severityMultiplierCheck impactMultiplier
Error1.00High0.85
Warning0.30Medium0.60
OK / no finding0Low0.30

For example, a High-impact Error affecting 50 of 100 checked records gives a 0.5 × 1 × 0.85 = 0.425 penalty, leaving that category at 57.5 before other scoring checks. The same finding as a Warning gives 0.1275, leaving 87.25.

03 / PUTTING IT TOGETHER

A widespread issue in practice

Take a synthetic audit of 1,000 pages. In Technical, one High-impact check finds an Error on all 1,000. Nineteen other Technical checks are clean, and the other three categories score 100.

TECHNICAL CATEGORYWidespread Error15 / 100

The affected share is 1. The check penalty is 1 × 1 × 0.85, leaving 15 in this category.

OVERALL HEALTHFour categories combined70.25 / 100

Technical contributes 35% of the overall score. The other three categories each score 100.

If the same finding were a Warning, the Technical score would be 74.5 and the overall score 91.075 (shown in the application as 91.1). These are illustrative calculations, not scores measured for a customer website. Additional failing checks can reduce the score further; the result depends on severity, impact and affected shares.

04 / READING THE RESULT

Counts, coverage and saved audits

What “affected” means

For most URL checks, a URL counts once within that check. The same URL can appear in multiple checks, so a total of 1,000 findings is not necessarily 1,000 different pages. Image checks use their own eligible records. Sitemap last-modified findings count supplied date records; missing sitemap and missing robots.txt sitemap declaration each use a single domain-level result.

What is excluded

Informational and Needs review findings, deliberately score-excluded checks, unavailable checks and checks without evaluated records do not contribute a penalty. A category without evaluated scoring checks receives no score. A score of 100 describes evaluated checks, not everything about a website.

What the label tells you

The detailed audit displays its scoring method alongside the score. An unfinished audit’s score is provisional. Check the method label and coverage when interpreting the result.

How history works

Saved audit history retains each score as originally recorded. Opening a detailed snapshot calculates its score using the selected method. Compare Audits applies that same method to both full snapshots and still checks comparable coverage. A summary-only audit cannot be recalculated without retained details.

Scope of the score

These are transparent DeepSpider product weights, not search engine ranking factors. The scoring weights are periodically reviewed against representative real audits.

05 / COMMON QUESTIONS

Questions & answers

Do more errors and warnings lower the score?

Yes. For a given scoring check, a larger share of affected records gives a larger penalty. Additional independent failing checks compound. An Error has more influence than a Warning of the same impact and reach. Raw totals alone do not determine the score.

Why can a site with thousands of findings still have a high score?

A large site can have many findings on a small proportion of its checked records. The same URL may also be affected by several different checks. Look at the affected and checked numbers for each issue to understand how widely it occurs.

Are 1,000 findings equal to 1,000 affected pages?

No. The totals can include findings from several checks on the same URL as well as image, sitemap and domain-level checks. Open the individual check to see its affected and checked records. Within most URL-based checks, repeated entries for the same URL count once.

What happens if an audit is Partial or checks are unavailable?

An incomplete audit score is provisional. Checks that were not run or were added after an older audit do not count as passes. A category without evaluated scoring checks receives no category score; the overall weights of available categories are proportionally rescaled.

Can I compare scores across audits?

Compare like with like: inspect both audits’ coverage, evaluated checks and the scoring method shown alongside the score. Detailed Compare Audits applies the selected scoring method to both retained full snapshots. A summary-only audit cannot be recalculated without its underlying findings.

Does 100 / 100 mean my site is perfect?

No. It means the scoring checks evaluated in this audit did not report penalized findings. It does not prove full site coverage, guarantee search rankings, or replace reviewing informational and Needs review findings.

What should I do after receiving a high score?

Review remaining high-impact findings and informational opportunities, confirm the audit covered the pages and resources you care about, and run another audit after meaningful site changes to watch for regressions.