Tarot for customer research: form hypotheses before interviews, not instead of evidence

Tarot for customers is useful not when it claims to reveal “what the audience wants,” but when it helps a team form several competing hypotheses before customer interviews. Cards cannot prove demand, willingness to pay, market size, the motive of a particular person or the future success of a product. Their practical role is narrower: surface a possible customer problem, current alternative, barrier, language of value or condition of trust, then turn each interpretation into a neutral question, observable evidence and a falsification criterion. Real conversations, behaviour, product data and a small experiment must come next. When the evidence contradicts the reading, discard the hypothesis rather than explaining why the cards were “right in another way.”

Before one interview round, I drew The Moon and quickly decided that prospective users were afraid of risk. It was an attractive explanation for weak activation and sounded psychologically sophisticated. The interviews kept returning to something more ordinary: setup time, data import and the wish to retain control of files. That experience changed one part of my practice. A card may generate a question; my first answer is still only a guess, however elegant it sounds.

What a spread may explore—and what it cannot establish

A spread can widen the field of inquiry when a team has fallen in love with one explanation. It cannot convert an assumption into a fact.

It is especially misleading to use court cards, Major Arcana or suits as audience segments: “Queens of Cups are emotional women,” “Pentacles are wealthy customers,” or “The Hermit represents introverts.” That is projection, not research. A segment becomes useful only when it describes an observable context, task or behaviour and is supported by evidence.

Record what is known before drawing

If the deck comes first, almost any product detail can be made to fit a symbol. Separate facts from assumptions before the spread. Write down:

1. Which behaviours have actually been observed: sign-up, abandonment, support request or repeat use?
2. At which step do people continue, and where do they stop?
3. How do they solve the problem now?
4. Which words or questions came directly from them rather than the team’s summary?
5. What evidence is missing?
6. Which explanation feels most convenient to the team?
7. What finding would show that explanation is wrong?

A two-column note is useful.

The right-hand column is research material. A plausible hypothesis remains a hypothesis.

An eight-position customer-hypothesis spread

A useful question is: “Which hypotheses about prospective customers should we test in the next interviews without assigning motives in advance?”

1. The customer problem. Which job or tension might the person be trying to address?
2. The current alternative. What do they use instead of, or alongside, the product?
3. The barrier. What may prevent starting, switching or returning?
4. The language of value. Which outcome needs human rather than promotional wording?
5. The condition of trust. What might the customer need to see, understand or keep under their control?
6. The team’s weak assumption. Which familiar explanation may be too convenient?
7. The interview question. How can you ask about past behaviour without suggesting an answer?
8. The small test. Which action, criterion and date will test the hypothesis?

For every position, write at least two possible interpretations. Then state what observation would make each interpretation less likely.

Save the hypotheses and review date in Reflecta

Worked example: a service for preparing client proposals

Composite case: a small web service helps independent professionals and small studios collect a client brief, organise materials and prepare a commercial proposal. Sign-ups are healthy, but many people do not complete the first setup.

Before the spread, the team records:

• users frequently open the pricing page and template library;
• many begin setup and leave at the client-data import step;
• support receives questions about export and taking materials out of the service;
• some users return to spreadsheets and email after registering;
• the team has not yet run structured interviews about the user’s last real project;
• there is not enough evidence about willingness to pay after a successful first use.

Imagine the cards are:

1. customer problem — Five of Pentacles;
2. current alternative — Eight of Pentacles;
3. barrier — The Moon;
4. language of value — Six of Swords;
5. condition of trust — The Hierophant;
6. weak team assumption — Seven of Cups;
7. interview question — Page of Swords;
8. small test — Ace of Pentacles.

Interpretation one: the barrier concerns trust and control of data

Five of Pentacles may suggest a shortage not of money, but of time, confidence or access to a clear process. Eight of Pentacles describes a familiar manual system. A spreadsheet is inconvenient, yet people know how it behaves and how to correct a mistake. The Moon in the barrier position allows a question about opacity: what happens after import, where files go and whether the user can reverse the action.

The Hierophant in the trust position may point to clear rules, a predictable path and a verifiable explanation. It does not mean that the customer is “conservative.” The hypothesis is more precise: switching is hindered by a lack of visible control. Six of Swords then suggests a value statement closer to “move your current process without losing your materials” than “revolutionise your workflow.”

Test it by asking someone to recall the last service into which they moved client data. What did they check before importing? Where did they pause? What form of export or rollback did they need? The hypothesis weakens when people do not independently mention control, reversibility, access or data handling.

Interpretation two: the problem is switching cost, not fear

The same cards support a more operational account. Eight of Pentacles shows accumulated skill in templates, folders, spreadsheets and correspondence. The Moon may represent an unknown amount of setup rather than an emotional fear. The user cannot estimate how long migration will take or what must be rebuilt. The Hierophant then relates to a standard, legible route, while Six of Swords suggests a gradual transition.

In this version, the value is not abstract safety but the ability to test one project without a complete migration. A useful question is: “Think about the last time you changed a work tool. What had to be transferred, how long did it take, and at what point did you continue or stop?”

The hypothesis gains weight when people describe concrete switching costs: time, manual work, team learning or duplicate systems. It weakens when setup seems easy and the real obstacle is another requirement, such as an unsupported proposal format.

What Seven of Cups may challenge

In the weak-assumption position, it can expose the team’s habit of explaining everything through missing features: more templates, more automation, more integrations. Perhaps there are already too many options and the first useful outcome is unclear. This is still only a hypothesis. In interviews, ask people to show the last real workflow rather than choose attractive features from a list.

Ask without placing the answer in the question

A poor question is, “Were you worried about uploading your data?” It interviews the researcher’s hypothesis. Ask for a specific past action instead.

Useful prompts include:

• “Tell me about the last time you…”
• “What did you do before that?”
• “What happened next?”
• “At which step did you decide to stop?”
• “What did you use instead?”
• “What cost you time, money or additional work?”

Weak prompts include:

• “Trust is important to you, isn’t it?”
• “Would you buy if we added…?”
• “How much do you like this idea from one to ten?”
• “What would your ideal product contain?”

People can sincerely praise an idea and never change behaviour. Compare words with the last real action, current alternative and cost of the problem.

The language of value comes from speech, not the card

A card may surface a theme, but value language should come from repeated customer wording. Six of Swords is not a reason to write, “We will carry your business safely to the other shore.” That metaphor belongs to the reader, not the market.

After interviews, group anonymised phrases by:

• how the person describes the problem;
• what they are trying to avoid;
• which outcome counts as sufficient;
• why they keep the current alternative;
• what creates trust;
• which word the team uses that customers do not.

Use direct quotations only with appropriate permission and without personal data. Otherwise, report the meaning accurately rather than inventing a polished “voice of the customer.”

A small test: action, criterion and date

The example team chooses one hypothesis: “People abandon setup because they do not know whether they will retain control of their materials.”

• Action: run five interviews about the last real attempt to change a work tool, then show an import prototype with preview and explicit export.
• Criterion: at least three participants independently name control, reversibility or data access as a material barrier and complete the revised prototype step.
• Review date: after the fifth interview and first prototype cycle.
• Real event: which words and actions repeated without prompting?
• Rejection condition: if another cause repeats, rewrite the hypothesis and do not build a feature merely to defend the initial reading.

Five is the size of this chosen rapid test, not a claim of representativeness. The cycle does not establish truth for an entire market; it helps decide which uncertainty deserves the next test.

Open a spread and record a testable hypothesis in Reflecta

Trust is not a manipulation problem

When a spread surfaces trust, do not search for a psychological button that will make the customer comply. Trust can be examined through real conditions:

• understandable rules and pricing;
• an opportunity to see the product before commitment;
• control of access and data;
• honest limits;
• functioning support;
• no false urgency;
• a way to decline or return to the previous process;
• promises that match the actual product.

Do not use tarot for covert profiling, exploiting vulnerability, finding “magic words” or inferring a particular customer’s emotional state. It is unreliable and undermines the purpose of research.

Common mistakes

Treating a card as evidence of demand. Symbolic coherence does not show how many people have a problem or will change behaviour.

Interviewing your own idea. Leading questions produce polite agreement, not knowledge.

Asking only about the future. “Would you buy?” is weaker than reconstructing the last comparable decision.

Failing to state falsification. Without it, any response can be called confirmation.

Confusing a segment with an archetype. Customers are grouped by context, task and behaviour, not by suit or Arcana.

Repeating the spread after unwelcome data. When real conversations conflict with the reading, change the interpretation or hypothesis—not the evidence.

Frequently asked questions

Can tarot show what customers want?

No. It can suggest a subject for inquiry. Needs, jobs and barriers are studied through real situations, behaviour, interviews and data.

Can I read on a particular customer?

That is unnecessary for business research. Do not assign motives to someone without their words and consent. Work with an anonymised situation, the process and the team’s own assumptions.

How do I know whether a hypothesis is supported?

Define an observable criterion and an alternative explanation in advance. One matching comment is not enough. Look for repetition together with behaviour and continue testing through other methods.

What if interviews contradict the cards?

Give greater weight to the interview and observed action. Record the mismatch and drop the initial conclusion. The card has done useful work if it prompted a better question, not if it can be defended at any cost.

Can a spread provide marketing copy?

Only as a rough source of themes. Final wording should be grounded in verified audience language, match the real product and avoid promises the product cannot keep.

Tarot for customers becomes practical when a test stands between the image and the decision. Record facts first. Form two interpretations, convert them into neutral questions and define what would make you abandon each one. After real conversations, save not the “correct card” but the changed understanding, next experiment, criterion and review date.

Form and test a customer hypothesis in Reflecta

— Daniel Aster, official Reflecta author