The First Yes

Design the smallest believable step that produces real value before asking for the demo, migration, contract, or major behavior change.

The Field Guide

Methods and tools to design AI products people trust and keep using.

Read Me Because

The first yes isn’t a conversion event.

Explore this concept

Create an account. Book a demo. Connect the system of record. Import the entire workspace. Invite five teammates. Commit to an annual plan.

Every one of those asks arrives before the user has a reason to say yes. What could one real case prove? Can the person judge the result before bringing another? What would the product have earned by then? Stacking the asks up front can hide whether any value is there.

Big asks hide weak value

The first yes shouldn’t require the user to believe the whole promise. It should give them a safe and simple way to find out whether the promise deserves another yes.

Yes, this describes my situation.

Yes, I’ll give this one real case.

Yes, I’ll let the product prepare the work.

Yes, I can see myself coming back.

The commitment should be proportional to proof

Before the product has demonstrated value, every request feels expensive.

Data access creates security questions. Team invites create social risk. Setup consumes time. Payment makes the user defend the choice. Autonomy asks them to accept consequences they haven’t seen the system handle.

The product should earn larger commitments by delivering smaller proof.

A diagnostic can earn the right to ask for a sample. A useful result can earn the right to connect a source. Repeated accurate preparation can earn the right to act on routine cases. A good summary doesn’t prove the product should read the entire drive. A correct draft doesn’t justify automatic sending. Let the next commitment stay proportional to what the first result actually demonstrated.

A good first yes has three qualities

Fill-in card

A good first yes

Check a proposed first yes for each of the three qualities.

Specific
the user knows what will happen next.
Reversible
trying doesn’t create a hidden migration or permanent rule.
Meaningful
the result touches a real job, not a toy exercise that proves nothing.

The biggest friction point isn’t the decision to try. It’s the fear that trying commits the user to staying. And a fake success animation after a tutorial click may teach where the button lives, but it doesn’t prove the product can help with anything the user cares about.

“Try our AI” is vague. “Bring one account you’re worried about; we’ll show what changed and where the evidence came from” is a first yes. The second sentence names the situation, scope, value, and evaluation path.

Example

A good first yes: one account the user is worried about

The ask is small, and the user can check the result before agreeing to anything larger. The product shows what changed in one account and where the evidence came from.

  • Situation: the user is already worried about this account.
  • Scope: one account.
  • Value: what changed.
  • Evaluation path: where the evidence came from.
  • Reversible: Not known yet: Can the user undo this step?
  • Meaningful: Not known yet: Does the result matter to the user’s real work?

Against the three qualities, it’s specific: the user knows what will happen next.

The first yes begins before interaction

The user has to recognize the product as relevant.

Precise situational framing does the work. A founder with a positioning problem should see the struggle, not a list of consulting capabilities. A success lead facing an at-risk renewal should see the trigger, not a generic claim about customer intelligence.

A broad AI claim invites the user to compare the product to every other AI tool, including the blank box they already know how to use. A specific trust claim tells the user what kind of work is safe to bring to it.

The opening should help people recognize the situation and show everyone else that the product is not for their current job.

Compliance isn’t conviction

A user can finish onboarding because a manager required it. They can connect a source because the setup wizard blocks progress. They can click through every step and still withhold the real yes: relying on the product when the job appears.

In analytics, compliance looks identical to commitment.

Fill-in card

First voluntary use

Track the first voluntary use of a meaningful result.

Act
did they act on the recommendation?
Use
did they use the draft?
Return
did they return with another object?
Share
did they share the artifact?
Increase
did they increase scope?
Stop
did they stop performing the old step?

Those behaviors show that the product crossed from curiosity into trust.

Find your first yes

Map the commitment ladder:

0 of 8 done

Then ask where the product currently demands a bigger yes than it has earned.