Poppy Flowers
thoughtbot · 2019
We built a quiz to take the florist out of it, and learned the florist was the point.
Buying what you cannot specify
A wedding is booked about a year out. By then the person buying the flowers does not know her guest count, her table type, her venue layout, or how many bouquets the day will need. Every florist’s quote depends on exactly those things.
That is the problem, and in October 2019 no one was pricing a wedding’s flowers before those answers existed. Poppy Flowers brought thoughtbot a one-week product design sprint to find out whether the business was worth building at all. Nothing had been built. The week ran on thoughtbot’s published sprint method, six phases from Prepare through Test, with one other designer, a developer, the founder and her team.
I ran it out of a Notion workspace, which is where the whole week lived.
Research came first: the business model, the mission, the market. Then interviews on both sides of it. Potential customers, people who had already bought wedding flowers, and local wedding florists, with a Google Form carrying the same questions out to the wider floral community. Sketching sessions followed, then a storyboard, and the group converged on one critical path narrow enough to prototype in the days remaining.
The path was short on purpose. Arrive at the site, choose color and style, describe the venue and the seating, receive the arrangement. That last step included knowing what to do with the flowers once they were delivered, which is the part a storefront usually drops on the floor.
The prototype answered the specification problem in four places. I don’t know yet was a first-class answer to a quiz question rather than a dead end. The form asked for rough estimates instead of facts. The document it produced called itself an editable proposal. And it promised a call two months before the wedding to confirm what the day actually needed. Pricing was itemized down to the unit, which in this category is itself the argument.
It also carried a florist’s view. So the week tested a two-sided marketplace, not a shop.
Seventeen assumptions, and a column to fill
The sprint’s real instrument was the assumptions board: seventeen assumptions, each with a method for testing it, a result, and a product takeaway. The board’s value is not that we wrote it. It is that eleven of the seventeen results came back before the week ended, in time to reverse the product.
The finding that reversed the product came from the assumption that a human connection builds trust. Every customer we spoke to wanted to talk to a person as their next step. So did the ones on the smallest budgets, which killed the neat theory that a low budget means a customer content to self-serve. The theory was tidy and wrong.
We had built a quiz to take the florist out of the process. The week told us the florist was the point. The quiz’s job changed from replacing that conversation to preparing for it: quiz, then a mood board, then a design specialist, then a proposal you can edit.
Other assumptions fell just as usefully. Customers did not need to see a visual of the final arrangement, so a mood board would do the work imagery was supposed to. Customers wanted color, style and vibe rather than per-item customization, which is a pricing decision as much as a design one.
Sustainability and local sourcing came back as a nice-to-have rather than a reason to buy. That is a narrower result than it sounds, and it is worth keeping narrow. The founder’s conviction was about florists and their future, not about what closes a sale. The business-model finding pointed the same way she did: hand off to the florist one to six months out, get very good at the vetting and the handoff, then let the florist do their job.
The founder heard it firsthand
None of this arrived as a recommendation. Cameron Hardesty was in the room all week and ran some of the research sessions herself, so she heard I would want to talk to someone from customers directly, in sessions she was leading. There was no deck. There was no persuasion step.
That is what a facilitator is actually for. The job is not to gather findings and present them; it is to arrange the week so the client witnesses them. A finding the founder helped produce is not a finding anyone has to sell her afterwards.
What the week could not answer
Six of the seventeen assumptions went unanswered, and every one of them was about florists. The survey did not return enough from the supply side. So a two-sided marketplace was designed for and prototyped, and only one of those sides was actually validated. That is worth stating plainly, because a sprint that reports only what it proved is not reporting.
A filled cell has a result — including the ones that came back no, which is the board working. The empty cells are all on one side: the survey meant to answer them under-returned. Both sides were built anyway.
What the week actually produced
Customers were most drawn to the convenience, the transparency and the quality of the approach. Naming the unknowns out loud, rather than hiding them, is what made someone more confident about a decision they were making a year ahead of the day.
Inc. placed Poppy at No. 1,049 on its 2026 Inc. 5000, on 334 per cent three-year growth. That growth belongs to the company. The sprint’s own output was narrower: a validated critical path, a tested prototype, and an assumptions board that recorded what we got wrong as carefully as what we got right.
“The prototype from our sprint really stood the test of time! We still use so much from that work.”
The durability is the part worth noticing. A prototype is normally the most disposable thing a project produces — it exists to answer a question and is thrown away once answered. This one was still in use years later, which says the week spent arguing about the critical path was the week that mattered.