Five numbers your website should tell you every Monday
Most analytics dashboards are opened twice and abandoned. Five numbers worth ten minutes a week — and the vanity metrics that quietly waste your time.
Almost every business site has analytics installed. Almost none of them are read. The dashboard gets opened the week it is set up, once more when someone asks how the site is doing, and then never again — because it opens on a screen with forty numbers on it and no opinion about which one matters.
The fix is not a better tool. It is a shorter list. Five numbers, ten minutes, once a week, and you will know more about your website than most companies know about theirs.
1. Visitors this week versus last week
Not the total since launch. Not "hits", which counts every image and stylesheet and inflates a page view into thirty. One number, compared to the same length of time before it.
What you are reading is direction, not magnitude. A jump means something happened — a campaign, a mention, a post that travelled. A slump means something broke or something stopped. Both are questions worth asking within the week, which is precisely why an annual report is useless for this and a weekly glance is not.
2. Which pages people actually landed on
Your homepage is not your front door. Search engines and shared links drop people deep into the site, and the pages they arrive on are rarely the ones you spent the most time designing.
This number changes decisions. If a single service page brings a third of your arrivals, that page deserves the next round of work — a clearer call to action, better photographs, a case study linked from it. And if a page you consider central receives almost nobody, you have learned something specific: either it is not findable, or it is not what people are looking for.
3. Where they came from
Four buckets are enough: search, social, referral and direct.
- Search is the one that compounds. It is slow, it is unglamorous, and it keeps working while you sleep.
- Social spikes and decays. Useful for launches, unreliable as a foundation.
- Referral tells you who is talking about you — a directory, a partner, a client's site.
- Direct means people typed your name, arrived from a saved link, or came from an app that strips the referrer. In our market a large direct share often means WhatsApp: someone shared your link in a chat, which is invisible to attribution but very real as traffic.
If nearly all of your traffic is one bucket, that is a fragility worth knowing about. Businesses that depend entirely on one platform's algorithm find out the hard way when it changes.
4. What people did, not just what they saw
This is the number that turns analytics from decoration into a tool, and the one most sites never configure. Define the two or three actions that mean something for your business and count those specifically:
- A contact form submitted
- A WhatsApp button tapped
- A phone number tapped on mobile
- A price list or catalogue downloaded
- A quote or booking request completed
Ten thousand visitors and four form submissions is a very different business problem than four hundred visitors and forty submissions. Only the second number tells you which one you have. Once you are counting actions, you can also do the honest arithmetic: how many visits does it take to produce one enquiry, and what does that make a visit worth?
5. What is not working
Reserve one of your ten minutes for the unflattering view:
- Pages that error. A 404 on a link you printed on a card or posted last month is a customer who gave up. These are cheap to fix and invisible unless you look.
- Mobile share. In Iraq the majority of traffic is a phone. If your site is being judged on a phone, review it on a phone — not on the desktop where you built it.
- The page people leave from. High exits on a checkout, a contact page or a form is a specific, fixable problem, not a mood.
The metrics that waste your time
Three numbers look important and are not:
- Hits. A relic that counts files, not people. It flatters and means nothing.
- Average time on page. Distorted by the way browsers measure the last page of a visit, and by anyone who leaves a tab open at lunch.
- Bounce rate, in isolation. Someone who reads your phone number and calls you is a bounce. That is a conversion, not a failure.
You are probably losing a third of your data to a banner
If your analytics needs cookies, it needs consent — and consent walls typically cost sites somewhere between a third and a half of their measurements. That loss is not evenly distributed either; it skews toward the mobile users in a hurry, which in this market is most of them. So the numbers you are making decisions from are both smaller and more biased than you think.
Cookieless measurement avoids the entire problem: no banner to click through, no personal profiles, and every visit counted. That is one of the reasons we built PearlFibers Analytics the way we did — one screen in plain language, native Arabic with real right-to-left layout, and the data stored on the same servers that host the site rather than shipped abroad. It is included free with any active hosting plan, so the five numbers above are already waiting on sites we host.
Make it a habit, not a project
Put ten minutes in the calendar on Monday morning. Look at the five numbers, write one sentence about what changed, and pick one thing to fix that week. Four weeks of that beats any quarterly report, because the whole value of website data is that it arrives fast enough to act on.
And if you are moving hosts soon, set up measurement before the move rather than after — otherwise the first thing your new analytics will show you is a mysterious cliff on the day you switched. The order of operations for that is here: moving a site without downtime.