The two numbers are users and sessions. They are the least useful figures available, because they tell you how many people arrived and nothing about whether arriving accomplished anything.
Everything worth knowing is two clicks further in, and almost nobody makes those clicks. Here is what is sitting there.
Entry pages, not top pages
The report everybody looks at is which pages got the most views. The useful one is which pages people arrived on.
Those are different lists, and the second one is your actual front door. A page you consider secondary may be where most first impressions happen, in which case it deserves the attention you have been giving your home page.
Check it once. It is frequently a surprise, and it changes what you work on next.
Your home page is where you think people start. Your landing page report is where they actually do. When those disagree, the report is right.
Traffic sources, and the lie inside them
Direct traffic is not a source. It is a bucket for everything the system could not identify.
That includes people who typed your address, but also people arriving from a mobile app, from an email client, from a text message, from a QR code, and from any link where the tagging was lost. Treating direct as brand awareness is a common and expensive misreading.
Two things worth doing. Tag anything you send, so it stops landing in that bucket. And ask new customers how they found you, because that answer will disagree with your analytics and the customer is more reliable.
The signals people misread
Four that regularly get read backwards.
Time on page, on a page that answers a question. Low time can mean the answer was found immediately. A pricing page with a short average visit and a high inquiry rate is working perfectly.
Bounce, on a single page visit that converts. Somebody who arrives, calls you, and leaves may register as a bounce. Bounce is only meaningful where you needed a second click.
Mobile numbers looking worse. They usually are worse, and it is usually the site rather than the visitor. Segment your conversion rate by device. If mobile is half of desktop, that is a fixable problem sitting in plain sight.
A traffic drop that is not a drop. Compare against the same period last year rather than last month. Seasonal businesses panic every year at the same time.
The four questions your data can answer today
You do not need new tracking for any of these.
- 01Where do people arrive, and does that match what you promote
- 02What do people do immediately after arriving, and where do they stop
- 03Which sources produce the visitors who take an action, rather than the most visitors
- 04Does mobile perform differently from desktop, and by how much
Question three is the one that changes budgets. A source sending a hundred visitors of whom four inquire beats a source sending a thousand of whom two do, and the second one looks better on every summary screen you have been checking.
A contractor sees one channel sending sixty percent of his traffic and assumes it is his best. Segmented by conversion, that channel produces two inquiries a month. A smaller source sending eight percent of traffic produces nine. He had been spending his effort on the larger number for a year. Nothing was wrong with the data. He had been reading the volume column and there is a column further right.
What to set up if you have not
Two things, and they take under an hour.
Conversion tracking. Until the thing you actually want is being counted, every other number is decoration. Form submissions, phone taps, and any download or booking.
Excluding yourself. Filter out your own visits. On a small site your daily checking can be a meaningful share of the total, and it makes everything look better than it is.
Without the first, you are measuring attention. With it, you are measuring outcomes, and those turn out to be uncorrelated more often than anyone expects.
What analytics cannot tell you
Worth being clear about the boundary, because people ask it questions it cannot answer.
It tells you what happened. It never tells you why. It cannot tell you that your pricing felt high, that the page confused somebody, or that a competitor was recommended by a friend that morning.
For why, you have to ask people. Three conversations with customers who nearly did not buy will teach you more than a quarter of dashboards, and analytics is best used to decide which three people to ask.
A review that takes fifteen minutes
Monthly, not daily. Daily numbers on a small site are noise.
- 01Compare this month against the same month last year
- 02Look at entry pages and note anything that moved
- 03Check conversions by source, not sessions by source
- 04Check conversion rate by device
- 05Look at what people did immediately before converting
- 06Write down one thing you will change, and check it next month
That last step is what turns a report into a decision. Most people look at analytics, feel informed, and change nothing, which is the same outcome as not looking.
The uncomfortable part
If you install conversion tracking and find that your busiest pages produce nothing while an obscure one produces most of your inquiries, that is not a measurement problem to be explained away.
It is the actual answer, and it usually means the work you were proudest of was aimed at the wrong reader. That is worth knowing early, and it is the same finding that shows up whenever anybody measures content by inquiries instead of traffic.