Denys

2025

Drawer AI — Takeoff workflow redesign

Cut time-to-first-takeoff for new estimators and lifted weekly active projects.

Role
Design Lead — research, UX/UI, prototyping, product copy
Team
1 PM, 4 engineers, 1 electrical domain expert
Timeline
12 weeks · shipped [month] 2025

Context

[ What Drawer AI is in two sentences, for a reader who has never heard of construction takeoff. Who the users are (estimators at electrical contractors), what a takeoff is, and why it's painful today (paper plans, Bluebeam, spreadsheets). ]

[ Where the product was when you started: stage, number of customers, what the takeoff canvas looked like, and what the business needed from it (activation, retention, expansion?). ]

The takeoff canvas before the redesign
The canvas as it was: every tool visible at once, no sense of what to do first.

Problem

[ The specific, observed problem — not "the UI was cluttered". E.g. new estimators stalled at step X; support tickets clustered around Y; PostHog showed Z% never placed a second symbol. ]

[ How you framed the problem for the team. One sentence that everyone agreed on. ]

My role

[ What you personally did vs. what the team did. Be specific: interviews, journey mapping, the information architecture, the prototypes, the copy, the analytics events you defined. Name the decisions you owned. ]

Process & key decisions

Research

[ How many sessions, with whom, how you observed real takeoffs. One or two concrete quotes or moments that changed your view. ]

Replace with a real quote from research that captures the core tension.

Senior estimator, [customer], during a shadowing session

Decision 1 — [name the decision]

[ The option space, what you chose, why, and what you traded away. E.g. task-first stepper vs. free canvas; auto-count review pattern; where symbol libraries live. ]

Decision 2 — [name the decision]

[ Same structure. Include one decision that engineering pushed back on and how you resolved it. ]

Decision 3 — [name the decision]

[ Ideally one decision made on data, one on craft, one on constraints. ]

After
Before
BeforeAfter
Drag to compare the symbol-placement flow before and after.

Solution

[ Walk through the shipped flow in the order a user meets it. Three to five moments, each with an image. Keep the prose short; let the screens carry it. ]

Step one of the new takeoff flow
Step 1 — [what the user does here and why it matters].
Auto-count review
Step 2 — [caption].

Impact

−38%
Time to first completed takeoff
Median, new accounts, 6 weeks post-launch
+22%
Weekly active projects
Cohort vs. previous quarter
−41%
Support tickets about the canvas
Intercom, 8 weeks

[ Replace the placeholder numbers with real ones. Say how each was measured (PostHog funnel, Intercom tags, sales feedback) and over what period. If a number is directional rather than proven, say so — it reads as honest, not weak. ]

What I'd do differently

[ One or two genuine regrets or open questions. E.g. instrumented too late; should have shipped the stepper behind a flag first; underestimated migration for power users. ]