Website Metrics That Matter: 6 Numbers Worth Your Time
A dashboard full of metrics is not the same as knowing how your site is doing. Here are six numbers worth checking, and three famous ones worth skipping.
On this page
Most analytics dashboards show you around thirty numbers, and roughly five of them will change a decision you make. The rest are there because they can be measured, not because they are useful.
Here are the six worth checking on a small site, in the order they matter, plus the famous ones you can safely stop looking at.
1. Human Pageviews
Not hits, not requests, not sessions. Pageviews from actual people.
The distinction is not pedantic. A meaningful share of raw traffic on any public site is automated: search crawlers, scrapers, uptime monitors, security scanners and AI crawlers. If your reporting counts those as visits, your traffic line goes up in a month when a scraper found you and nothing about your audience changed.
This is the number every other number depends on, so it has to be honest before anything else is worth reading. DevDome Analytics classifies each hit against six independent signals and reports human visitors separately from bot activity, and plans are metered on human pageviews only, so a bad scraping month does not become a bigger invoice.
2. Entry Pages
Which page did each visit start on? That tells you what is actually working, as opposed to what you think should be working.
Almost every content site has a surprise in this list: a post written in an afternoon two years ago that now brings in more traffic than anything you have published since. That page deserves your attention, and most site owners do not know which one it is.
Sort by entry visits, look at the top ten, and ask what those pages have in common. That is your site’s real subject, whatever your homepage claims.
3. Outbound Clicks
For most content sites, this is the closest measurable thing to revenue.
If you make money from affiliate links, sponsorships or sending traffic to your own product, the outbound click is where the value is created. Pageviews upstream of it are just context.
It is also the number most likely to be wrong, because the usual way of measuring it relies on JavaScript that ad blockers strip. Server-side click capture solves that by recording the click as it passes through your own infrastructure rather than asking the visitor’s browser to report it.
4. Where Visitors Come From: Sources and Geography
Two numbers that belong together, because each explains the other.
The source split is search, direct, referral and social. What you are watching is not the exact ratio but changes to it. A steady site that suddenly gains a large referral share has usually been picked up somewhere worth knowing about. A steady site losing search share over two months has an SEO problem that has already started. Both are visible in the ratio weeks before they show up in the total, which is what makes this worth a monthly look.
Country and region is more actionable than it sounds, especially for anyone earning from affiliate links. If a quarter of your readers are outside your primary market and your monetisation only works in one country, that is a quantified problem with a known fix. Geography also explains oddities elsewhere: a page performing unusually badly may simply be reaching an audience it cannot serve.
5. Bot Share
Not a vanity metric, a hygiene one. Knowing that automated traffic is a steady fraction of requests is normal and unremarkable. Watching it jump from a routine share to half your traffic in a week tells you something found you: a scraper, an aggressive AI crawler, or someone probing for vulnerabilities. You only get this number if your analytics measures bots rather than silently discarding or silently counting them.
6. One Conversion You Chose
Every site needs a single action that means the visit mattered. A newsletter signup, an affiliate click, a contact page view, a demo request. Pick one.
Without it, traffic numbers have no reference point and every month is just a bigger or smaller version of the last one. With it, you can ask the only question that improves a site: which pages produce this, and which pages get read and produce nothing.
The Three You Can Safely Ignore
These appear on every dashboard and each answers a different question from the one people ask it.
Bounce rate. It records whether a second event fired, not whether the visitor was happy. Someone who lands on your answer, reads it, gets what they came for and leaves counts as a bounce. On a content site that is success.
Average time on page. It is derived from gaps between events, so the last page of any visit usually contributes nothing. The number systematically understates reading time in a way that varies by page.
Total sessions. A session is an arbitrary window with a timeout, and different tools draw the boundary differently. It is the number most likely to shift when you change tools while your actual audience stays identical.
Make It a Ten Minute Routine
Open your dashboard once a month. Look at human pageviews, entry pages, outbound clicks, source split and your one conversion. Note anything that moved more than you expected, and write one sentence about why.
That is the whole practice, and it beats a daily dashboard habit that generates opinions about noise. The point of analytics is to change what you do next, and five numbers you understand will do that better than thirty you skim.
Disclosure: DevDome publishes this blog and makes the products it mentions. We describe what they do rather than promise results.
Key takeaways
- Human pageviews, not raw hits, is the only traffic number worth reporting.
- Entry pages tell you what is working; exit pages tell you what to fix.
- Outbound clicks are the closest thing to revenue that most content sites can measure.
- Bounce rate is the most misread number in analytics and usually means nothing.
- Pick five numbers, check them monthly, and ignore the rest of the dashboard.
Sources
- Google Analytics Help: bounce rate and engagement — how engagement replaced bounce rate as the default quality signal
- web.dev: Core Web Vitals — the performance metrics Google uses, separate from traffic analytics
Links last checked August 22, 2026.
Frequently asked questions
Why is bounce rate on the ignore list?
Because it measures whether someone triggered a second event, not whether they were satisfied. A reader who lands on your post, reads all of it, gets their answer and leaves is a success story that records as a bounce. On a content site a high bounce rate is often a sign the page is doing its job, which makes the number useless as a quality signal.
How often should I actually look at analytics?
Monthly for most sites, weekly if you are running a campaign. Daily checking mostly measures random variation and encourages reacting to noise. The exception is a live view during a launch or a traffic spike, where the point is to confirm nothing is broken rather than to analyse anything.
Is average time on page useful?
Less than it looks. It is calculated from the gap between recorded events, so a visitor who reads for ten minutes and closes the tab often contributes nothing at all to the average. Treat it as a rough comparison between pages, never as a real reading time.
Should I track conversions if I do not sell anything?
Yes, you just have to decide what counts as one. For a content site it might be an outbound affiliate click, a newsletter signup or a visit to your contact page. Pick the action that means the visit was worth something to you and count that. Without it, every other number lacks a reference point.