SAM RUKIN

I never guess where onboarding breaks.

I find the target — and show my work.

Evidence

Case Studies

Attio

Severely Hurt
Howard's onboarding evaluation of Attio

The most severe failure in the pool.

  • Signup blocked before the product was ever reached.
Read the Attio breakdown

PostHog

Hurt
Howard's onboarding evaluation of PostHog

Signup completes, but the account stays empty.

  • Evaluated by Howard from real walkthrough footage.
See the PostHog report

HubSpot

Slightly Hurt
HubSpot checkout stalled on a blank, unlabeled loading spinner

“Buy now” opens a real checkout on a brand-new trial.

  • Drops into a bare spinner with no product, no price, and no payment form.
  • The same unlabeled spinner recurs at login, in a feature preview, and at the account switcher.
  • It's the only loading state HubSpot has anywhere, and it never once says what's happening.
Read the HubSpot findings

Zendesk

Slightly Hurt
Zendesk admin center replaced by a full-page notice that Sales will be retired, with no way back

Everything upstream of this is genuinely well-built: a live demo ticket queue behind the signup modal, a clean AI-agent setup wizard with its own test panel.

  • One click on “Sales” in the product-switcher menu, a normal-looking peer of Support, Chat, and Analytics, strips the entire admin shell.
  • Drops you on a page announcing the product is being retired, with no way back but an unlabeled logo.
See where the crack is

Linear

Not Hurt
Howard's onboarding evaluation of Linear

The strongest onboarding in the pool.

  • A live, populated workspace in under a minute.
See what Linear got right

Airtable

Slightly Help
Airtable's AI cobuilder stating a plan and waiting for confirmation before building anything

The AI cobuilder states its plan and waits for confirmation before building anything.

  • Labeled every step, never a blank or unlabeled wait.
  • Built a working app with realistic sample data on the first try.
See what Airtable got right
Process

Method

Ruled Out

Assumptions without evidence. Gut feelings. Cargo cult UX patterns that don't match the user's actual context.

Evidence Held Up

Observation. Real user behavior. Patterns that repeat across different contexts. The signal, not the noise.

The Target

The specific problem to solve. The moment where friction breaks the funnel. What must change first.

Pricing

Straightforward, either way

Find the Target

$1,500

Flat rate. Diagnostic assessment and evidence-backed recommendation.

Find It and Fix It

$2,200–$2,500

Full scope. Problem diagnosis plus design implementation and testing.