Evidence beats opinion

"I already know my customers"

You probably do — partly. You talk to them. You've been in the trade for years. The objection deserves respect, so let's state it properly: nobody is asking you to pretend you know nothing.

But knowledge earned by proximity has a specific blind spot. It is strongest about the customers you see most, and silent about the ones you never see: the people who considered you and chose a competitor, chose nothing, or never found you at all. Those people are your growth, and by definition you have never spoken to any of them.

There's a second problem. Beliefs about customers harden into facts through repetition, not through checking. "Our buyers find us on Instagram" gets said in enough meetings and becomes furniture. Nobody remembers it was once a guess.

This module doesn't ask you to distrust everything you know. It asks you to write down what you know, mark how you know it, and then check the pieces that carry the most weight. Some will hold. At least one, in our experience, will not.

Say, do, decided

Three grades of information about customers, in ascending order of reliability:

What people say. Opinions, predictions, answers to "would you…?" Cheap to collect and least reliable, because people are polite, forgetful, and generalise about themselves. A person who says "I'd definitely book online" is describing the person they'd like to be.

What people do. Behaviour you can observe: searches, visits, add-to-carts, orders, repeat orders. Far more trustworthy — behaviour doesn't flatter you — but mute. A number tells you that something happened, never why.

What they actually decided on. The criteria in play at the moment of a specific, real choice: what triggered it, what alternatives were on the table, what nearly stopped them. This is the gold, and it's recoverable in only two ways — reading what people wrote at or near the moment of decision (Lesson 2.3), and asking them to walk you through one specific past decision (Lesson 2.4).

Most SME "customer knowledge" is a fourth category the list politely omits: what the founder said once, remembered as data.

Tomas is right — and it doesn't help

Tomas at Herzog Physio believes his patients come by GP referral and word of mouth. The pack numbers back him up: of roughly 45 new patients a month, perhaps 30 arrive via GP referral and 12 by word of mouth.

So the belief survives contact with the evidence. And it still isn't the knowledge he needs — because "arrived via GP referral" describes how they arrived, not how they chose. A GP in Graz who says "you need physiotherapy" has started a decision, not finished it. The patient still has to pick a clinic, and Tomas has no idea what happens in that gap. His own words: "I don't know where new patients would even look for us, or whether they search at all before choosing a clinic."

Except his data contains a clue he's never looked at. His website gets about 800 visitors a month, mostly on branded searches — people typing something close to "Herzog Physio" into Google. Somebody gives them the name, and they still go and check. What do they find? A site built in 2019 with no online booking, and an unclaimed Google Business Profile with 23 reviews he has never answered.

Be careful with what this proves. Some of those 800 are existing patients checking opening hours; the number alone can't say who they are or what they concluded. That's exactly the point. The behaviour is real, his belief can't account for it, and only research can close the gap. A confident, broadly correct assumption was still hiding the question that matters.

Ana's belief describes her commenters

Ana at Ostara Ceramics believes her buyers are women 28–45 in Portugal, Spain and France who care about craft, buy for their own homes or as gifts, and find them on Instagram.

Ask where that belief comes from and the answer is: Instagram. The people who comment, DM and like are visible every day, so they became "the customer". But Ostara has 14,200 followers and about 91 orders in a typical month. The buyers are overwhelmingly the silent ones, and the engagement data describes the audience of the content, not the buyers of the plates.

Write your beliefs down now, because this module will test Ana's — and at least one part of it will need revising by Lesson 2.6. Watching a belief you wrote down get corrected is the entire pedagogy of Module 2. It only works if you commit before you look.

Assumption, evidence, and the confidence label

Working definitions for the rest of the Programme:

An assumption is a belief you'd struggle to source. "I just know." "Everyone in our industry knows that."

Evidence is something a customer said, did, paid for, or that data shows — with a source you could point to. "14 of our last 20 reviews mention response speed" is evidence. "Our customers value speed" is an assumption wearing its clothes.

From this lesson on, every customer claim in your Marketing Plan carries a confidence label — High, Medium or Low — and an evidence note, where "none" is an allowed and genuinely useful answer. An honestly unsourced belief tells you exactly where research pays first. A false citation tells you nothing and costs you later.

Start from a decision, not from curiosity

Good research is not "let's learn about our users". It starts from a decision the research will inform: Which pain do I lead with on the homepage? Which of my two shortlisted channels deserves the first six months? Is the second location's emptiness a visibility problem or a demand problem?

A research question anchored to a decision tells you when you're done — when you can make the call. Curiosity-driven research is never done, which is why it usually never starts.

What this module does with your list

Six lessons. Next lesson gives you the sharpest thinking tool in the module — Jobs to be Done. Then desk research you can do today without talking to anyone, interviews that don't lie to you, a short honest survey, and finally synthesis into an Ideal Customer Profile where every line is source-tagged. Your five beliefs get a verdict at the end: Confirmed, Revised, or Killed.

Write them down.

---