Measure calls booked, not vanity traffic

The number that tells you a company site is working is a conversation with the right buyer. For a studio, that is an intro call or a serious email. For a product, it is the signup or the request that matches the person you built for. Sessions, likes, and time on page are context around that number.

Website conversion metrics are the small set that tracks the conversation. Everything else is a clue you use when the conversation is missing, not a score you celebrate on its own.

The action that pays for the site

Name the action before you open an analytics tool. Write it down in one line: "A founder books an intro call" or "A qualified team asks for a demo." If you cannot name it, you will end up optimizing whatever the dashboard puts at the top.

Put that action where it can be counted. A link to a scheduler, a form with a real inbox, a button that creates an account. Give the event a name you will still understand in three months. "generate_lead" is a label. "intro_call_booked" is a record. Count the event, and count it by the page that sent the person, so you know whether the homepage, a project, or an article did the introducing.

Quality belongs in the definition. Ten calls from people who need a different kind of partner are not ten successes. Note the fit after the call, in whatever simple way you will actually keep up: a yes, a no, or a not-yet. The site's job is the right conversation. Volume without fit is a sign the first screen is attracting the wrong reader, which is a copy problem before it is a traffic problem.

Hanabi's commercial action is an intro call. The site should make that path obvious, and the review each month should ask how many of those calls were with founders who needed a site, a brand, or a product interface. That is the metric. The rest explains it.

Sessions are context

Traffic tells you whether anyone arrived. It does not tell you whether the page deserved the arrival. Nielsen Norman Group's 2011 note on page visits is a useful corrective. Users often leave within 10 to 20 seconds. The average visit lasts a little less than a minute. People have time to read only about a quarter of the text on the pages they do open. A "long average session" can mean a confused reader hunting for the offer. A short one can mean they found the scheduler and left to book.

Read sessions next to the action. A page with steady traffic and no calls is a page whose promise, proof, or path is failing. A page with little traffic and a high rate of fitting calls is a page worth showing to more of the right people. Neither number stands alone.

Google's helpful-content guidance frames the same idea from the publishing side. The reason to make a page is to help a person reach a goal, not to attract visits for their own sake. A report that leads with users and pageviews, and mentions the call in a footnote, has the order backward. Lead with the calls. Use the visits to ask why.

Rankings and impressions are context too. They tell you which queries brought people. They do not tell you whether those people were buyers. A query about a definition can bring readers you are glad to teach. It should not be scored as if it were a query from someone ready to hire.

What to instrument

Keep the setup small enough that you will look at it.

• The primary action, counted once per person when you can.
• The page they were on before they took it.
• The query or the campaign, when you know it.
• A simple fit note after the conversation.
• One speed check, so a slow launch does not hide inside a "low conversion" story.

Skip the custom dashboard that takes a week to interpret. A founder should be able to answer, in one sitting, how many of the right people asked to talk, and which URL helped them.

If you use a form, test the inbox. A counted submission that lands nowhere is not a conversion. If you use a scheduler, agree what "booked" means: a hold, a confirmed meeting, or a meeting that happened. Pick one and keep it. Changing the definition mid-quarter makes the chart a story about the chart.

Do not add a metric because a template included it. Bounce rate, pages per session, and scroll depth are easy to misunderstand and easy to game with longer pages and extra clicks. Use them only when you already have a question. "Do people reach the proof?" is a question. "Is the bounce rate healthy?" is not, until you know what action the page was supposed to earn.

How to read a weak week

When the calls dip, look in a fixed order. First, did the action break: the form, the scheduler, the inbox, the button on mobile? Second, did the offer change on the page without a matching change in who you attract? Third, did a new source bring people who were never going to buy? Fourth, only then, ask whether you need more of the right traffic.

A redesign is the last explanation, not the first. If the sentence at the top is unclear, rewrite the sentence. If the proof is a screenshot with no decision, write the decision. If the button is below a long essay, move the path up. Measure again after the change, on the same definition as before. A new metric and a new design in the same week teaches you nothing.

Give the change two or three weeks unless something is actually broken. A single quiet Tuesday is not a trend. A month of the wrong buyers is.

A review that fits a founder

Once a month, sit with three lines: calls booked, calls that fit, and the pages that led to them. Bring the queries if you have them. Decide one change. Ship it. Leave the rest of the dashboard alone until the next review.

That habit keeps the site in service of the business. It also keeps a studio honest about what it promised to improve. "More traffic" is a weak promise. "A clearer path to a conversation with the right founder" is a promise you can design, build, and count.

The weeks after launch are when this habit either starts or gets replaced by nerves. The companion piece is the first 90 days after a site launches. Read it before you treat the first quiet week as a reason to start over.

When you want a partner who will treat the call as the point of the page, book an intro call. Bring the action you want the site to earn. The conversation should decide whether that action is clear enough to design toward.