The path to a purchase

People don't buy everything the same way. Two kinds of purchase sit at the ends of a scale.

• A considered choice has high stakes, a large payment and often several people involved. An exam tutor for a teen is one: the parent pays for a large package, the teen studies, a school teacher's word counts, and a wrong choice costs months before an exam. People spend weeks comparing.
• A habit is small, repeated and low-risk. Klasna Club is closer to one: a modest monthly payment, an adult who already learns at Klasna, easy to stop. People decide in minutes.

That explains Leo's puzzle. Exam families take three weeks from the first message to paying, adults join the Club in a few minutes, through the same site and the same ads. They are not faster or slower customers; they are making different kinds of decision.

So one landing page with one "book a trial" button serves neither well. Start every channel plan by asking which kind of purchase you are dealing with.

Most paths to a purchase run through the same steps, whatever the product:

1. Notice the need.
2. Look for options.
3. Compare and check.
4. Try.
5. Buy.
6. Use.
7. Buy again or recommend.

An exam family's path, concretely: a mock exam in January shows a weak maths score; the parent asks the school teacher and an AI assistant about options; the parent compares Klasna with the teacher's paid after-class lessons and reads a sample progress note; the teen takes a trial lesson; the parent pays for the first package; weekly progress notes arrive and the parent buys a second package before April.

A considered choice spends weeks in the middle steps, looking and comparing. A habit almost skips them. And the path doesn't end at "buy": for Klasna, the second package is where a family starts to pay back what it cost to win, so "use" and "buy again" belong on the map too.

A common habit is to make everything talk to the "buy" step: every ad says "book a trial". Ask instead what each step needs.

• Look for options: be easy to find where parents look, including through teachers.
• Compare and check: proof, such as a real sample of the weekly progress note.
• Try: a trial lesson the teen agrees to attend.
• Buy: clear package and refund terms.
• Buy again: proof that the first package worked.

Also write who stands on each step. In a considered choice the people change along the path. For exam families, the school teacher sits at "look for options", the parent compares and pays, and the teen decides at "try": a trial the teen refuses is a lost family, however convinced the parent was.

Pushing "book a trial" at every step skips what people need to get there. A parent who hasn't found you, or hasn't seen proof, won't book anything.

A path becomes useful when you count it. Klasna's CRM for last spring shows, of every 100 enquiries from exam parents: 52 booked a trial, 37 came to it, 17 bought a first package.

The headline number, 17 of 100 enquiries, says 17% bought. That hides where families are lost. Measure each step from the one before it instead:

• booked a trial: 52 of 100 = 52%;
• came: 37 of 52 = 71.15%;
• bought after coming: 17 of 37 = 45.95%.

The step view tells a different story. More than half of the families whose teen actually tried a lesson still walked away. That is not a traffic problem, and more enquiries won't fix it; something happens at or after the trial.

The rule: the denominator of each step is the number who reached the step before. Divide everything by the first step and every leak looks like the same vague "low conversion".

Between booking a trial and taking it sits a quiet step that teams often skip: showing up. Of the 52 exam families who booked a trial last spring, 37 came: 37 ÷ 52 = 71.15%. Fifteen families who had already said yes never reached the lesson.

This step has its own needs, and they are different from the steps around it. The parent booked, but the teen has to come, at a time that works, remembering the lesson exists. What helps: a time the teen chose, a reminder the family agreed to receive, and a short note on what the trial will cover so it feels useful rather than like an exam.

Counting the step separately matters because it changes the fix. If you merge "booked" and "came", the 15 lost families look like a weak trial, and the team might rewrite the lesson when the real problem is a forgotten appointment.

Every handover between people on the path, here from parent to teen, is a place where a step can leak.

Klasna's exam path has two big leaks: 48 of 100 enquiries never book a trial, and 20 of the 37 families who came to a trial don't buy. They need different fixes, because they sit at different steps.

• The first leak is a compare-and-check problem. A parent who enquires but doesn't book often hasn't seen enough proof to choose Klasna over the teacher's after-class lessons. The fix belongs at that step: reply to the enquiry with a real sample of the weekly progress note and reviews from real families, so the parent can check before committing time.
• The second leak is a trial problem. It depends on the lesson and on what the parent hears afterwards: what the tutor noticed, what a plan would look like.

A bigger discount, more ads or a louder "book now" act on none of these steps. Before choosing a fix, name the step where people leave and what that step was missing.

For a habit, the hard step is not buying but using. Klasna Club's path is short: an adult joins in a minute. But many leave after the first month. Someone who joins in a minute can forget in a week.

So the fixes that work on a considered choice don't work here. A longer sales page, more benefits, more ads or a bigger first-month discount all act on the steps before joining, which are not where members are lost. More ads only replace the people who leave, at full cost each time.

What the "use" step needs:

• a fixed weekly slot, so the club becomes part of the week;
• a reminder the member agreed to receive, not messages they never asked for;
• a strong first three sessions, so the member meets the part worth paying for early.

The principle is the same as for the exam path: find the step where people leave, then give that step what it lacks.

More parents now start "look for options" by asking an AI assistant: "compare online maths tutors for my son". The assistant answers from what is written about you on the web. So your own site matters for this step even if the parent never visits it first.

What helps: say plainly whom Klasna is for, how lessons work and what proof you have; publish a real sample of the weekly progress note; ask real families for reviews after a finished package, without paying for praise. What doesn't: invented five-star reviews, which mislead people and break consumer law, or hidden refund terms, which do the same at the "buy" step.

An AI tool can also help you draft a path from interview notes, with names and contacts removed first. Then check every step against the CRM numbers and two real conversations: an invented path sounds right and can still be wrong.

How it fits together: know the kind of purchase, map its steps, count each from the one before, and fix the step where people leave. The next lesson picks the channels, each judged by the step it serves.