quietstats learning centre
The metrics handbook
Every number on a quietstats dashboard, in plain English: what it tells you, how it is counted, how to read it well and when not to trust it. For the exact formulas, see the methodology; for one-line definitions, the glossary.
- Pageviews
- Visitors
- Visits
- Views per visit
- Bounce rate
- Visit duration
- Live visitors
- Sources and channels
- AI referrals
- Entry and exit pages
- Countries, regions and cities
- Devices, browsers and systems
- Goals and conversion rate
- Custom events
- Revenue and products
- Engaged time and scroll depth
- Speed (Core Web Vitals)
- A/B test result
Pageviews
- What it tells you
- How much your pages were looked at in total.
- How it is counted
- One per page load or single-page-app route change; the same address twice in a row counts once. See methodology.
- Read it well
- Use it for the size of your content’s reach and to compare pages with each other.
- Watch out
- A reload loop, a busy bot or one keen reader can inflate it. Pair it with Visitors.
- Where to find it
- Top pages, trend chart
Visitors
- What it tells you
- Roughly how many different people came each day.
- How it is counted
- A daily visitor ID from a salted hash that is reset every day. Over several days it is the sum of each day’s visitors. See methodology.
- Read it well
- Compare like with like: this week against last week, this month against the same month last year.
- Watch out
- Someone who comes back on three days counts three times, by design. Never compare it directly with a cookie-based “users” count.
- Where to find it
- Dashboard headline, compare date ranges
Visits
- What it tells you
- How many separate sittings people had with your site.
- How it is counted
- A run of pageviews from one daily ID with gaps of 30 minutes or less. See methodology.
- Read it well
- A rise in visits with flat visitors means people are coming back more often in the day.
- Watch out
- A visit that runs over midnight UTC is split in two.
- Where to find it
- Dashboard headline
Views per visit
- What it tells you
- How far people go once they arrive.
- How it is counted
- Pageviews ÷ visits.
- Read it well
- Look at it per landing page or source: a source that brings 1.1 views per visit is sending people who leave straight away.
- Watch out
- Single-page sites and long articles naturally score low; that is not a problem on its own.
- Where to find it
- Visit depth
Bounce rate
- What it tells you
- The share of visits that saw only one page.
- How it is counted
- Visits with exactly one pageview ÷ all visits. See methodology.
- Read it well
- Most useful per landing page and per source, not as one site-wide number.
- Watch out
- A reader who spends ten minutes on one article is still a bounce. GA4’s bounce rate is defined differently, so the two do not match.
- Where to find it
- Bounce rate and visit duration
Visit duration
- What it tells you
- How long a visit lasted, from first to last pageview.
- How it is counted
- Time of last pageview − time of first; single-page visits count as 0 seconds. See methodology.
- Read it well
- Treat it as a lower bound and compare it over time on the same site.
- Watch out
- Time on the last page is not measured. For reading time, use the engaged-time add-on.
- Where to find it
- Engaged time
Live visitors
- What it tells you
- Who is on the site right now.
- How it is counted
- Distinct daily IDs with a pageview in the last 5 minutes.
- Read it well
- Handy during a launch, a newsletter send or a mention on social media.
- Watch out
- Small numbers jump about; do not make decisions from a single minute.
- Where to find it
- Live visitors
Sources and channels
- What it tells you
- Where visits came from: search, social, AI, email, paid, referral or direct.
- How it is counted
- The referrer host plus UTM tags put each pageview in one channel, in a fixed order. See methodology.
- Read it well
- Tag every link you control with UTM tags so it lands in the right channel.
- Watch out
- Many apps send no referrer, so their visits show as Direct.
- Where to find it
- Sources and channels, UTM campaigns
AI referrals
- What it tells you
- Visits that came from AI assistants such as ChatGPT, Perplexity, Gemini, Claude and Copilot.
- How it is counted
- The referrer host matches a known AI assistant.
- Read it well
- Watch which pages AI assistants send people to; they tend to be your clearest explanations.
- Watch out
- Google AI Overviews arrive as google.com and count as Search.
- Where to find it
- AI referrals
Entry and exit pages
- What it tells you
- Where visits start and where they end.
- How it is counted
- The first and last pageview of each visit.
- Read it well
- Improve the entry pages with most traffic first; they are your real front doors.
- Watch out
- An exit is not always bad: a contact or thank-you page is a fine place to finish.
- Where to find it
- Entry and exit pages
Countries, regions and cities
- What it tells you
- Where visitors are, roughly.
- How it is counted
- Country from Cloudflare’s network; region and optional city are daily totals with small numbers held back.
- Read it well
- Use it to decide opening hours, delivery areas, languages or ad targeting.
- Watch out
- VPNs and private relays show the VPN’s location.
- Where to find it
- World map, places and cities
Devices, browsers and systems
- What it tells you
- What people use to read your site.
- How it is counted
- Device, browser and OS families from the user agent and screen width; the user agent is then thrown away.
- Read it well
- If most visits are on phones, test every change on a phone first.
- Watch out
- Families only (Chrome, Safari, iOS), not versions.
- Where to find it
- Devices, browsers and OS
Goals and conversion rate
- What it tells you
- Whether visits did the thing you care about.
- How it is counted
- A goal is a page, click or event; conversion rate = visits that reached it ÷ all visits in the range.
- Read it well
- Set one goal per business outcome (enquiry, order, sign-up) and judge sources by conversion rate, not traffic.
- Watch out
- Tiny numbers swing wildly: 1 of 4 is 25%. Look at a month, not a day.
- Where to find it
- Goals and funnels
Custom events
- What it tells you
- Actions that are not page loads: clicks, sign-ups, downloads, plays.
- How it is counted
- Sent by the optional events add-on or your own code, with an optional short property.
- Read it well
- Name events as verbs people do (“Book a table”), and keep the list short.
- Watch out
- Never put personal data such as an email address in an event name or property.
- Where to find it
- Custom events
Revenue and products
- What it tells you
- Money by source, page and product.
- How it is counted
- Order values from Stripe, your shop or the conversions API, joined to the visit’s source without cookies.
- Read it well
- Compare revenue per visit by source; it is fairer than total revenue.
- Watch out
- Refunds and offline sales are only included if you send them.
- Where to find it
- Revenue by source, product performance
Engaged time and scroll depth
- What it tells you
- Whether people actually read.
- How it is counted
- The optional engagement add-on sends one small message when a page is left or hidden.
- Read it well
- Use it for articles and long pages where bounce rate tells you little.
- Watch out
- Only counted when the add-on is installed; it never changes bounce rate or visit duration.
- Where to find it
- Scroll depth and engaged time
Speed (Core Web Vitals)
- What it tells you
- How fast pages feel to real visitors.
- How it is counted
- Optional vitals add-on: LCP, INP and CLS from real page loads, kept as totals.
- Read it well
- Fix the pages with most traffic and the worst LCP first.
- Watch out
- Lab tools such as Lighthouse measure one load; these are real visitors on real devices.
- Where to find it
- Speed budgets
A/B test result
- What it tells you
- Which version of a page did better, and how sure you can be.
- How it is counted
- Visits and goal completions per version, with a two-proportion test.
- Read it well
- Decide in advance what “better” means and how long to run; do not stop early on a lucky day.
- Watch out
- Small tests can’t prove small differences. “No clear winner” is a real answer.
- Where to find it
- A/B tests
Three habits that beat any single metric
- Compare periods, not days. Week on week and month on month smooth out the noise.
- Segment before you conclude. A site-wide drop is usually one source, one page or one device. Click any row to filter the dashboard.
- Tie traffic to an outcome. Set at least one goal so you can tell good traffic from busy traffic.
Next: put it into practice with a playbook for your kind of site.