Product

Ten checks. One schedule per site.

SiteRefiner watches every client site with ten automated services across site health, search, analytics and spend. Each keeps its own cadence, reports progress as it runs and says plainly when it fails.

SiteRefiner is ten automated services, each on its own schedule per client site: Site Audit, Page Speed Insights, Link Crawl, Search Console, Index Coverage, SEO Opportunity, Google Analytics 4, Conversion Funnel, Tag Manager and Google Ads. The three site health checks need no credential. The rest read only the Google properties a client has shared.

Every service keeps its own clock. Each site and service pair holds a cadence, a next run time and a full run history. Due checks start on their own, never pile up, and fail loudly instead of hanging.

Site health

Three crawlers that read a site the way browsers and search engines do: every sitemap URL, rendered in a real browser, scored page by page.

FIG 1.1

Site Audit

No credential
A real-browser crawl of every sitemap URL that checks what a search engine and a share preview will actually see.Full mode widens the crawl beyond the sitemap
  • HTTP status and on-page SEO: title, meta description, canonical, H1, alt text, noindex, html lang
  • JSON-LD structured data for the schema types that earn rich results
  • Open Graph tags, including a real fetch of the preview image
  • Raw versus rendered HTML, for content invisible without JavaScript
  • Security headers, SSL expiry and robots.txt health
  • AI-crawler access for GPTBot, ClaudeBot and PerplexityBot, and llms.txt presence

Page Speed Insights

No client credential
Google Lighthouse on every sitemap page, on mobile and desktop.Focused on the pages that matter most to the site
  • Performance, Accessibility, Best Practices and SEO scores
  • Core Web Vitals: LCP, CLS, TBT, FCP and INP field data
  • Only the failing audits that actually move a score
  • A slow site never holds up the other checks

Link Crawl

No credential
A navigation-focused crawl that clicks what a visitor can click and records where it breaks.Paced to be gentle on the client's host
  • Dead links across every anchor, form action and safe-to-click button
  • JavaScript console errors and redirect chains
  • Orphan pages and internal pages missing from the sitemap
  • Works on hosts behind bot protection

Analytics & tracking

First whether the measurement is right, then what the numbers say. A missing tag is caught before anyone reports on a quiet month.

FIG 1.3

Google Analytics 4

Google credential
The traffic picture for each site over a rolling 28-day window.Feeds SEO Opportunity and Insights
  • Page views, sessions, engagement and events
  • Landing pages and channels
  • Daily trend across the window

Conversion Funnel

No extra setup
Define a sequence of GA4 events and see exactly where visitors leave.Runs once the funnel has two or more steps
  • Users reaching each step
  • Completion rate and drop-off between steps

Tag Manager

Google credential
Reads the container itself, so tracking problems surface as findings rather than as missing data later.Container and account IDs required
  • Draft versus published drift
  • GA4 tag present and firing on all pages
  • Consent Mode configured
  • Duplicate triggers, paused tags and tags without triggers

Spend & return

What a client pays for that the site could own, and what it already owns but still pays for.

FIG 1.4

Google Ads

No extra setup
Ad performance read through the site's existing Google connection, with nothing new to set up.Quality score and impression share are on the roadmap
  • Spend, clicks, conversions and value per campaign, ad group, keyword and search term
  • Wasted spend: money on terms with zero conversions
  • Paid-organic overlap: paying for keywords the site already ranks top three for
  • Feeds the keyword-strategy narrative

Where it lands

Insights
Ads data joins Search Console positions and keyword demand in the Spend & Return view.Buy, build, both, own it, overlap or skip
  • Organic value, CPA, ROAS and blended CAC per keyword
  • A budget planner that predicts clicks, conversions and return
Spend & Return

The life of a check

From a schedule coming due to a refreshed Insights page, in seven steps. Nothing waits on a person remembering to press a button.

  1. Step 01

    Due

    A site's weekly Site Audit comes due.

  2. Step 02

    Start

    The run starts on its own, as long as nothing is already running for that service on that site.

  3. Step 03

    Crawl

    A real browser opens the site and works through every page in the sitemap.

  4. Step 04

    Progress

    Every finished page moves a counter you can watch from the run page.

  5. Step 05

    Store

    Every finding is saved against its page, ready to join with the other services.

  6. Step 06

    Narrate

    SEO Opportunity and Google Ads runs end in a plain-language report you can question.

  7. Step 07

    Refresh

    Insights, Keywords and Spend & Return pick up the new run.

Failure is a first-class state

Runs are long and some hosts fight back. Every run carries one of five states, and a failure gets the same care in the interface as a pass.

Pending

Queued, never stacked

A run waits with its dispatch time. A second run of the same service on the same site is never queued behind it.

Running

Progress you can watch

Crawlers and Page Speed report every completed page. Long crawls keep going for as long as pages keep finishing.

Passed

Stored and joined

Findings are saved, the last run is stamped and every view that depends on it recomputes.

Failed

Caught within minutes

A run that stops making progress is failed within minutes, and the run page says why.

Cancelled

Stopped on request

Anyone with edit rights can stop a run mid-flight. The schedule carries on from the next cadence.

Audit one page, right now

Paste a URL and get Site Audit checks, a link crawl and Page Speed on mobile and desktop, merged into one prioritised list with plain-language fixes. Cancellable between steps.

Page audit harbourdental.example/services/implants
Export Audit again
Site audit checks Link crawl Page Speed mobile Page Speed desktop
Issues 2 critical2 warning1 info
CriticalMissing structured dataSite audit
Why it matters

Without JSON-LD the page cannot earn service or FAQ rich results, so it shows as a plain blue link.

How to fix

Add a Service block with name, provider and area served. The template dropped it in the last release.

CriticalUncaught TypeError in the booking widgetLink crawl
Why it matters

The Book now button never opens for visitors whose browser hits the error.

How to fix

Load the widget script after its config object, or guard the call until it exists.

WarningLCP 4.1s on mobilePage Speed
WarningMeta description is 212 charactersSite audit
InfoOpen Graph image smaller than 1200 × 630Site audit
Health summary just now
Technical
45
Mobile
54
Desktop
88
Checklist 6 of 9 passing
Title tagPass
Meta descriptionToo long
CanonicalPass
Single H1Pass
Image alt text3 missing
Structured dataNone found
Open GraphPass
Security headersPass
AI crawlers allowedPass
hreflangNot used

Built for a client roster

An account is your agency or your business. Clients are optional folders. Sites are what gets checked. Every site, run and draft is scoped to the account, and platform staff can only reach it by impersonating, which is logged.

Account roles are assigned per user.
Capability Account Admin Editor Content Generator Viewer
View sites, runs, Insights and reports
Edit sites, clients and schedules
Cancel and delete runs, audit single pages
Generate content, where the site has it enabled
Manage users and roles
Switch services on or off per site

Quiet by design

The unglamorous rules that keep checks honest on a slow host or a throttled API.

Never stacked

Scheduler

Due checks start on their own. A service never queues a second run on a site while one is in flight.

Minutes

Stalled runs

A run that stops making progress is failed with a reason, so nothing sits on running forever.

Hours

Long crawls

Crawls may run for hours on a slow or protected host, and stop only when pages stop finishing.

Shared

Sitemaps

One sitemap fetch serves every check on a site, so a client's host is never hit ten times over.

180 days

Retention

Run history is pruned after 180 days, always keeping the latest successful run per site and service.

Exactly once

Retries

A re-run replaces its findings instead of adding to them, so a retry never doubles a report.

Questions

What does SiteRefiner check?

Ten services per client site: Site Audit, Page Speed Insights, Link Crawl, Search Console, Index Coverage, SEO Opportunity, Google Analytics 4, Conversion Funnel, Tag Manager and Google Ads. Each covers one question a site owner or an agency gets asked, from "is this page indexed" to "where is the ad money going".

How often do the checks run?

Each service on each site has its own cadence: daily, weekly, monthly or on demand. Runs start on their own, never queue a second copy behind one already running, and can be started or cancelled from the run page.

Which checks need a Google connection?

Search Console, Google Analytics 4 and Tag Manager read the properties a client has shared. Index Coverage, Conversion Funnel and Google Ads work from those same connections. Site Audit, Page Speed Insights and Link Crawl need no credential at all.

Can I audit a single page without a full run?

Yes. Paste a URL into Page audit and get the Site Audit checks, a link crawl and Page Speed on mobile and desktop as one prioritised list with plain-language fixes.

What happens when a run fails?

It is shown as failed with the reason, alongside the passes. A run that stops making progress is ended rather than left spinning, and a retry replaces its findings instead of doubling them.

Does SiteRefiner check AI-crawler access?

Yes. Site Audit reports whether the site allows the main AI crawlers and whether it publishes an llms.txt, alongside the usual title, description, canonical, structured data and Open Graph checks.

Ten checks, every client.One schedule per site.