Reading guide · The fine print

What the audit is telling you,and where we step in.

A short guide to reading the scores, the Web Vitals chart and the AI-generated brief — plus the specific work our analytics-engineering team takes off your plate.

How the audit is laid out

Three tabs, one verdict. Each tab answers a different question — pick the one that maps to your problem.

Tab 01

Summary

The fastest read. Overall grade, every score side-by-side, the highest-impact quick wins ranked across the whole audit.

Tab 02

Tracking

What GTM, GA4, PostHog, Shopify pixels and your other instrumentation actually do — coverage, depth, container internals, page-level signals.

Tab 03

SEO & Performance

Lighthouse-derived speed, SEO and accessibility scores plus the Core Web Vitals Google grades you on.

What the scores mean

Every score is 0–100. Higher is better. The bands below are how we read them when we’re briefing a client.

Configuration health

Configuration score

How sound the existing setup is — fires events in the right order, gates on consent, doesn’t double-load tags, references real GA4 properties. Deductions come from concrete misconfigurations.

Tracking depth

Tracking depth

How ambitious the instrumentation is — events captured, custom dimensions, server-side tracking, ecommerce coverage, engagement signals. A site can be configured cleanly and still have shallow tracking.

Performance, SEO, accessibility

Discovery Scores

Pulled from a real Chrome run on both mobile and desktop. Mobile is weighted more heavily by Google for ranking. The breakdown lists every audit that subtracted points so you can see exactly what to fix.

The bands
  • 80–100Good

    Healthy. Worth a sanity check, not a project.

  • 50–79Needs work

    Real upside. Usually the cheapest project to scope — known issues, known fixes.

  • 0–49Poor

    Talk to us. Almost always blocking attribution, ranking or both.

Core Web Vitals

The five real-world metrics Google grades page experience on. Each card shows your value, Google’s bar, and the human-readable description.

LCPLargest Contentful Paint

Time until the biggest visible element finishes loading — the “page is ready” moment.

Good when 2.5 s or less

INPInteraction to Next Paint

How quickly the page responds when the visitor clicks, taps or types. Real responsiveness, not first-load.

Good when 200 ms or less

CLSCumulative Layout Shift

How much the page jumps around as it loads. Low is good, zero is great.

Good when 0.1 or less

FCPFirst Contentful Paint

Time until the browser renders the first piece of content — the “something is happening” moment.

Good when 1.8 s or less

TTFBTime to First Byte

How long until the server returns the first byte. Server + network speed before any rendering happens.

Good when 800 ms or less

AI insights

Each report can run a Sonnet-grounded analysis pass. Strict rules apply — these aren’t rephrased Lighthouse findings.

Grounded

No invention

The model receives only the audit’s own data: scores, breakdowns, container parses, signals. It cannot cite events that aren’t there.

Prioritised

By horizon

Items are sorted into next-7-days, next-month, next-quarter buckets, with effort pills and impact lines so you can plan a sprint, not just a wishlist.

Specific

Cite, don’t hand-wave

Every observation references the exact rule, tag, or metric that triggered it. If it can’t cite, it says “unknown”.

What we do

The work that lands on our desk most weeks. Hover any card for the typical project shape.

01

Server-side GTM

Routing first-party data through your own subdomain. Reclaims signal from ad-blockers, raises Meta/Google match rates, gives consent-mode something real to gate on.

02

Consent-mode v2 / CMP

Right banner, right region, right consent strings — and tags that genuinely respect them. We replace surface-deep cookie banners with audit-ready setups.

03

Meta Conversions API

Pixel + CAPI side-by-side, deduplication tuned, event_id propagated. Closes the 10–25% gap iOS & ad-blockers carve out of pixel-only setups.

04

Ecommerce instrumentation

The full GA4 ecommerce funnel — view_item to purchase — wired up in GTM, in your platform, and in any sGTM you’re running. With QA dashboards.

05

GA4 migration & rebuilds

Old UA mental models → GA4 reality. Custom dimensions, event taxonomy, BigQuery export, attribution windows, audiences that actually port to ads.

06

Core Web Vitals & SEO

LCP, INP, CLS — measured, attributed, fixed. Schema, canonicals, sitemap and accessibility cleanups in the same engagement when it makes sense.

Need help?

Tell us what's most pressing and we'll come back to you with a scope.

Talk to a human

Tell us what’s most pressing and we’ll come back with a scope. Server-side GTM, Conversions API, consent-mode v2 and ecommerce-event instrumentation are our day-job.

+44 7378 538 432
What you’d like to discuss(click to unselect)

We aim to reply within one working day.

← Back to the audit