Expert judgment, built into every report
Auditaar is website intelligence built by growth marketers. From a public URL it crawls and renders your site, runs the applicable checks from a library of 200+ across six pillars, and reports one A-Score out of 100 in which every point traces to a check.
Auditaar needs only a public URL: no logins and nothing to install.
Built by growth marketers who have run digital marketing for businesses
The people who build Auditaar have run digital marketing for businesses, where a finding counts only once it becomes a task someone can schedule. The engine applies rules drawn from that work, and each finding in the report shows its evidence.
Rules learned running campaigns, applied to your site
Campaign tags followed through your redirects
Auditaar sends test requests tagged with campaign parameters and ad click IDs through your own redirects, and records which survive each hop. A dropped click ID breaks conversion import in the ad platform itself, so it ranks above a lost UTM.
Double-counted sessions flagged on direct evidence
A measurement ID that loads twice counts one visit as several, so sessions inflate and the conversion rate reads low. Auditaar raises it only on direct evidence: the loader fetched twice, two pageview hits on one load, nothing clicked.
Your tag container read against what actually fires
Auditaar reads your published tag manager container, with no account access, and compares it with the tools that actually fire in the browser. It flags tools configured but never seen running, and tools firing from outside the container.
No invented revenue figures
No finding carries an invented loss figure, because revenue cannot be estimated without your own traffic data. Money appears only where a provider measures it: estimated traffic value and keyword opportunity value, each with its method shown.
Every capability Auditaar sells is recorded with its status, tier and limits, and the website describes only what that record backs. Ideas that fall short, such as maturity scores without a published rubric, are recorded as ruled out, with the reason.
The method behind every finding
On every run, Auditaar weighs what a site and its data providers report against what it can observe directly, and scores only what it measured.
Sample lines are demonstration data from the public sample report, a full Auditaar report on an anonymized site.
Every page read twice: as sent, and as a browser builds it
Auditaar loads each sampled page in a real browser and compares it with the HTML the server sent. The gap is the share of the page that exists only after JavaScript runs, which the AI crawlers that read for citation never see.
A real browser, then a real phone
Pages are sampled one per template, so each kind of page is judged on its own, not through the homepage. Each then loads again as a real phone, at 375 by 812 with touch input, to check tap-target size against WCAG 2.2, text size, zoom blocking and sideways scroll.
Consent checked against what lands in the browser
A consent banner can look compliant while the setup underneath ignores it. Auditaar visits as a first-time visitor who has clicked nothing, and catches cookies and tags that fire before consent, storage written early, and cookies missing from the site's own policy.
Index records confirmed before they become work
Backlink indexes lag the live site, so Auditaar rechecks them: it requests up to 25 of the pages an index lists as broken and records today's answer. Pages that already work are marked fixed and left out of the plan, and upper-bound totals are labeled as a ceiling.
Two data series compared before a trend is called
A single series can make a decline look like growth, so Auditaar reads link flow against the running total and calls a disagreement a mixed result. For online stores, it compares the price shown on the page, the price in the markup, and the price in the platform's product feed.
A correct setup is reported as correct
Some results that look like failures are the right configuration, and Auditaar tells them apart: blocking AI training crawlers is valid, and a robots-blocked URL is a fault only if your sitemap also lists it. Correct setups never become work, so engineering time goes to real defects.
The A-Score, point by point
The A-Score condenses your site's scored checks into a single score out of 100. Every point a pillar loses belongs to one check, so it traces check by check, and Auditaar shows each pillar's highest-leverage fix.
Rounded to 67. Each measured pillar counts equally.
Built from the 153 checks that applied and were measured on this site.
Measurement & Consent
100 points, split by weight across the checks that applied and were measured on this site.
- Campaign tags and ad click IDs dropped by the site's own redirects
- −10
- Tracking cookies set before any consent was possible
- −8
- Tracking runs with no consent platform
- −8
- Eight further consent and measurement checks
- −35
- Points lost
- 61
- Pillar score
- 100 − 61 = 39
A pillar point is worth one-sixth of an A-Score point when all six pillars are measured: 10 ÷ 6, rounded, is 2.
Computed once across the plan's scored fixes, then rounded, so per-fix gains are never double-counted.
Reported beside the 67, never added to it.
Demonstration data from the public sample report, a full Auditaar report on an anonymized site. The pillar ledger is drawn from the report's own scoring data; the report itself highlights each pillar's top fix.
The calculation
Each pillar splits 100 points across the checks that apply
Every check carries a weight from 1 to 10. For each pillar, Auditaar splits 100 whole points across the checks that applied and could be measured, in proportion to those weights, so the same check is worth more where fewer apply.
A pillar score is 100 minus what its failing checks cost
A failed pass-or-fail check costs its full share; a check measured across pages costs in proportion to the gap. The pillar score is 100 minus those costs, in whole points that add up exactly.
The A-Score is the average of the measured pillars
The measured pillars are averaged with equal weight and rounded: 66.7 becomes 67 in the sample. AI Visibility leads the display order but carries the same weight as the other five.
The rules around it
Not measured stays out of the math
A check that could not run, or does not apply, leaves the calculation: never a pass, never a fail. A bot challenge, a block page or an empty shell is reported as not measured with the reason, and a pillar with nothing measured drops out of the average.
Market signals measured beside the score
Nine signals that move with competitors and the market, such as AI share of voice and control of your brand's search results, are measured but kept out of the A-Score. Auditaar reports them in their own panel, so the score moves only with your own site.
Each scored fix shows the points it recovers
Each finding tied to a scored check shows what fixing it adds to its pillar and the A-Score, from the same arithmetic that set the score. The plan's total is summed once and then rounded; the free Diagnostic shows one figure, the A-Score points on the table.
The score is final before the agents start
No model sets the score; the arithmetic is fixed. Auditaar's agents start only after the score is final, and tuning the weights never rescores a report already delivered.
Findings ready to hand over, ordered by dependency
For each finding, Auditaar states the business consequence and names an owner, then schedules the blockers before the work that depends on them.
Sample cards are demonstration data from the public sample report.
- 1
Every figure printed with the population it measured
Each figure states its population on the same line, such as the pages rendered or the URLs your sitemap declares, so a rate keeps its meaning. Auditaar also names each section's sources and whether a figure is a full count, a sample, a floor or a gap.
Populations: Diagnostic and Deep Dive. Sources and coverage by section: Deep Dive. - 2
The business consequence, and who owns the fix
Every finding explains how it costs the business, as a mechanism your developer can confirm in a minute. Auditaar also names an owner (engineering, marketing, content, legal or leadership), so the plan can be split by team.
Consequence: Diagnostic and Deep Dive. Owner: Deep Dive.
Tracking runs with no consent platform detected
- Auditaar's test
- Auditaar loaded your pages as a first-time visitor and recorded which analytics and tag tools ran, then looked for a consent platform gating them.
- 1Measured over
- 25 crawled pages, loaded with no cookies sent and nothing clicked (a sample)
- 2Why it matters
- A tag manager and analytics were detected with no consent platform, so nothing captures or honors a visitor's choice. For EU or UK visitors, that tracking likely runs without a lawful basis.
- 3
Blockers first, then the work that depends on them
Structured data that only appears after JavaScript runs is invisible to an AI crawler. So the plan schedules the fixes that restore access first, and holds the work that depends on them for the later phases.
Deep Dive - 4
Phases grouped by effort
After the blockers, work is ordered by severity against effort and grouped into phases, so a quick fix ships early instead of waiting behind engineering work. Passing checks, correct setups and the audit's own measurement gaps stay out, because none is work.
Deep Dive
18 of the 74 findings in this plan are work an engine cannot see until these ship, so it is scheduled first.
- Weeks 1-2Restore access2 fixes
- Weeks 3-4Configuration and markup fixes11 fixes
- Weeks 5-8Template and content changes55 fixes
- Weeks 9-14Engineering and program work6 fixes
Market and competitor data cut down to what you can act on
Auditaar adds the market around your site: rankings, competitor analysis, keyword gaps and AI citations. Raw market data is noisy, so Auditaar drops off-topic keywords before the market data reaches the report, and after the score one agent per pillar adds evidence of its own.
Sample lines are demonstration data from the public sample report.
The right country's market and the competitors you name
Auditaar reads which country's market your business competes in from the site itself (its address schema, a country-coded domain, the page's language), before pulling any search data. Competitors you name lead the comparison, topped up to three with auto-detected rivals once publishers, marketplaces and your own sister brands are removed.
Keyword-gap demand read from the typical month
In the keyword gap, demand is read as the median of the last twelve months, because a one-month spike in a provider's feed can inflate the average and float dead terms to the top. Any outside statistic is printed only after Auditaar fetches the cited page and finds the exact figure, with the link and date.
Keyword gaps cut to a shortlist of briefs
Keyword-gap terms are filtered for navigational intent, duplicates and demand that contradicts difficulty, then the shortlist passes a relevance check against what you sell. The list is not padded back out after a cut, and terms too hard to win now are set aside for later.
Agents that add evidence after the score is final
Once the score is final, a team of agents, one per pillar, investigates your site's data and adds evidence notes. A verifier drops any figure not in the data the agent saw, and where a fix is a ready artifact (schema, a meta title, a robots line, a header) an agent drafts it and shows it only after an offline validator passes.
Read a complete sample report before you run your own
The public sample report is a full Auditaar report on an anonymized site, with every finding, each pillar's top fix and the phased plan. A free AI visibility check reads your own homepage across all six pillars from its URL alone.