Consulting · 3 teardowns

Teardowns that find where users leak.

Alongside product work, I audit live products for founders and teams. Each teardown below is a real piece of work: what I looked at, what I found, what I recommended, and what I do not know. None of them claims a result I cannot show.

• 2025 · Fintech app · Funnel and usability audit

  Wekeza

  The context

  A multilingual app for the African diaspora that offers U.S. stock trading at $1 a trade. The client shared a funnel snapshot: 64,000 visits to the sign-up form, 5,000+ downloads, 1,500 newsletter sign-ups, and low funding of brokerage accounts.

  What I looked at

  A heuristic review of the live app against Nielsen’s ten heuristics, each issue rated on a 0 to 4 severity scale and tied to a task path, across registration, dashboard, KYC, settings and email, plus a funnel plan from install to first funded trade.

  What I found

  • Sign-up: key requirements and the privacy checkbox sit in low-attention areas, the primary button lacks emphasis, and the form asks for a lot.
  • Dashboard: a news block pulls attention away from the “Create Account” action and the cash-available figure.
  • KYC: the instructions sit where users do not look, and yellow and light-green text on white is likely to fail contrast.
  • Verification should not block the dashboard: let people in first, then ask for it.

  What I recommended

  • Meet 4.5:1 text contrast and enlarge touch targets.
  • Add Google sign-in and relax the password rules.
  • Add a Skip to verification so users reach the dashboard before they are asked to prove who they are.
  • Segment the follow-up messages by stage and language.

  The funnel figures are the client’s. I do not have conversion results from after the audit, so I claim none. The heuristic pass was done with AI as a collaborator, which the report states.

• Aug 2025 · AI meeting assistant · Marketing-site teardown

  Cluely

  The context

  A premium, well-animated site for an AI meeting assistant, with strong brand motion and a clear positioning line.

  What I looked at

  A quick review of the homepage, product, pricing and enterprise pages from the public site.

  What I found

  • In the first three to five seconds, a visitor cannot tell which outcome Cluely delivers and for whom.
  • Pricing leaves per-seat, per-minute and storage costs unclear.
  • Enterprise buyers do not see trust and security signals near the price and the hero.
  • There is no short answer to “why this and not the recorder already in my meeting tool”.

  What I recommended

  • Keep the visuals; sharpen the headline to the outcome and the places it works.
  • Add microcopy to each call to action to lower decision anxiety.
  • Restructure pricing into three tiers with the billing model stated.
  • Put trust badges and a one-line differentiator beside the hero and the price.

  Based only on what is public on the site. I did not have any of Cluely’s analytics, so the report ranks fixes by effort and likely impact, not by measured lift.

• Aug 2025 · Career-tools product · UX strategy summary

  Careerspan

  The context

  A set of AI career tools: Vibe Check, Resume Tailor and Crystal Ball. I wrote the summary after a call with the team.

  What I looked at

  A walkthrough of the product and a short strategy document, with two engagement options attached.

  What I found

  • New users get no “first win”: nothing says where to begin or what success looks like.
  • The tools feel like separate features rather than one system, so users do not know how they work together.
  • No visible reward loop: no progress marker, no confirmation, no moment of win.
  • Small friction adds up: no undo, unclear resume limits, slow loads.

  What I recommended

  • Design a first-win onboarding that ends in one high-value action.
  • Present the tools as one guided path to job readiness.
  • Add micro-feedback after each action and a visual tracker of progress.
  • Break long application forms into steps and pre-fill with AI where possible.

  A strategy summary, not a shipped redesign. I make no claims about results.