PIPELINE TAKEOFF · ALPHA TO BETA · SIMPSON STRONG-TIE
ABOUT THIS PROJECT
Pipeline Takeoff is the first step in Simpson Strong-Tie’s estimating software for lumber yards: import the plans, measure, and the quantities flow into a sellable estimate. A takeoff is unpaid work to win a bid, so every wasted step costs money. The alpha worked, but estimators found it slow and easy to break. I led the redesign from alpha through beta, fixing the everyday workflow before adding tools, and task completion went from 52% to 86%.
MY ROLE
Lead Designer. Research, strategy and prioritization, design through beta, with the PM and engineering lead.
PRODUCT
B2B · Construction estimating · Web app
TIMELINE
April 2024 – January 2025
RESULT
52% → 86% task completion · Shipped to beta
The Challenge
Estimating shouldn’t be the bottleneck. For lumber yards, it was.
GOAL
Give Pipeline LBM its own takeoff tool, so measured quantities flow straight into a job’s materials list instead of being exported from a separate tool and re-keyed by hand, and so the platform is easier to sell as one product. Ship it for the 2025 launch, scoped by what estimators fought daily rather than by the feature list.
PROBLEM
The alpha could do the work, but every repeated step fought the estimator: a missed point meant redrawing the whole measurement, 70-page plan sets had to be set up one sheet at a time, and a roof couldn’t be measured at all. It was functional, not fluid, and no one would switch tools for that.
TARGET USER
Estimators at lumber yards working through 50 to 70-page plan sets to win bids they usually lose. Most have used takeoff tools for years and judge a new one quickly against what they already know.
The Solution
From plan sheet to sellable estimate without leaving the platform.
I focused the redesign on the parts of a takeoff that an estimator repeats on every job. Measuring and correcting got a clear Record and Edit model. Setting up plan sheets got bulk actions and manual scale calibration. Working at speed got hotkeys and a toolbar that changes with the mode you’re in. Once that foundation held up in testing, I added the tools a takeoff couldn’t be finished without: pitch for roofs, framing, and the measurement types estimators expected from the tools they’d used before. And because the takeoff now lives inside the platform, quantities sync straight into the job’s materials list. The export-and-import step is gone.
Research
Where was everyday takeoff work creating unnecessary effort?
Takeoff is complex, repetitive work that’s hard to evaluate in a single moderated session. I ran a one-to-two week diary study in our staging environment with ten participants, six internal SMEs and four external customers, structured around a set of core takeoff tasks so they worked on their own projects and documented as they went.
MEASURING WASN’T SLOW. IT WAS UNFORGIVING.
People couldn’t tell whether they were drawing, editing, or finished, so they hesitated before every click. And once a measurement was done, a missed point or a stray click meant deleting it and starting over.
SHEET MANAGEMENT WASN’T TEDIOUS. AT 70 PAGES, IT DIDN’T SCALE.
Renaming sheets one at a time is fine on a ten-page set. On the 50 to 70-page sets estimators actually receive, it was the first thing everyone complained about.
HOTKEYS AND CALIBRATION WEREN’T PREFERENCES. THEY WERE HOW ESTIMATORS TRUST THEIR WORK.
Experienced users expected the shortcuts they already knew. Estimators working from scanned plans wanted to set the scale themselves, because if the scale is wrong, every number on the sheet is wrong.
Product Principles
Every interaction has a cost when the work is repeated all day
The backlog I inherited was a list of tools. Research showed the loop underneath them was what estimators fought, so I set one question as the bar and made the case for it with my PM and engineering lead: does this make a frequent task faster, clearer, or easier to recover from, without compromising accuracy? Every candidate feature went through it. The four words it produced became the headings in Final Solution.
Naming 70 pages shouldn’t take longer than measuring one
A plan set runs fifty to seventy pages, and renaming, rotating, or duplicating them one at a time was the first thing every estimator fought. Bulk actions let them select several sheets and act once.
A context menu for every sheet, one right-click away
A right-click menu brings each sheet’s actions to the sheet itself: rename, set scale, rotate, duplicate, print, and delete. No hunting through a panel for the thing you’re looking at.
Dynamic toolbar
Hotkeys and the dynamic toolbar
Experienced estimators work from the keyboard. The dynamic toolbar puts the right controls on screen for the current mode, and hotkeys for Record, Undo, Redo, Edit, and Properties let them work at speed without hunting for buttons.
Record mode
A clear signal that you’re measuring
In the alpha you couldn’t tell whether you were drawing, editing, or done, so people hesitated before every click. Record mode is now a visible state: the toolbar changes, the recording indicator stays on, and every click adds a point until you stop.
Edit mode
The same measurement, now adjustable
Select a finished measurement and it opens in Edit mode with its own toolbar. Nothing is redrawn; the points you placed are the points you adjust, so it’s always clear what a click will do.
Edit mode: add, remove, subtract
Faster only counts if it doesn’t create rework
A missed point used to mean deleting the whole measurement and starting over. In Edit mode an existing measurement can be adjusted directly: add, move, or remove a point, or subtract an area for stairs, windows, or islands.
Scale calibration
If the scale is off, every quantity on the sheet is off
Estimators working from scanned plans don’t trust a preset scale, and they shouldn’t: if the scale is off, every quantity on the sheet is off. Manual calibration lets them check it against a dimension they know before they measure.
Pitch tool
A roof can’t be measured flat
A roof can’t be measured from a flat plan without its pitch, so the alpha couldn’t complete a roof takeoff at all. The Pitch tool adds the slope to area and linear measurements and works out roof rake from the top-down view estimators already read.
Joist/Rafter tool
Completing the frame: Joist/Rafter and Beam/Header
Estimators correct a generated framing layout on nearly every job, because plans rarely match how a house is actually built. In the alpha, the only way to correct it was to redraw it. I made the generated layout a starting point instead: move individual joists, split members, and fix the edges in place. Beam/Header does the same for beams and headers, so a full frame can be taken off in PTO.
The Outcome
From a usable MVP to a workflow estimators could rely on
52% → 86%
task completion across comparable core workflows
2.7 → 4.0
average task satisfaction, out of 5
STAKEHOLDER IMPACT
The second testing round, same people and same tasks as the first, showed most of the friction gone: estimators finished more often and rated the work easier. For the business, it meant a new customer could run a takeoff end to end inside the platform, with the quantities already in their materials list. That’s the bar for selling takeoff as part of the product rather than as an add-on. Real use then showed us the next set of gaps, around review and handoff, and that became the next phase: beta to GA.
DESIGN SYSTEM VALUE
The redesign shipped alongside a UX team move from legacy components to React with shadcn. PTO had a lot of components the shared system didn’t cover, so I wrote the parity playbook: a screen-by-screen spec of what each legacy screen had to keep, change, or drop, so the migration wouldn’t quietly undo the fixes. Its impact can’t be measured on its own, since it shipped with the features.
