Stitch Fix Started With a Google Form

Validate the job before the website

I first wrote about Stitch Fix in a December 22, 2020 First 1000 issue. Katrina Lake’s early version looked more like a small operations business than a software startup.

Stitch Fix did not start with a polished shopping app. Katrina Lake started with a Google Form, a credit card, and a willingness to style people by hand. The software came later, after the service had answered a more basic question: would a stranger trust a remote stylist enough to buy the clothes?

That is a useful escape hatch for founders who are about to build the interface before they understand the work behind it.

Katrina Lake started Stitch Fix during her second year at Harvard Business School. After testing other ideas, she focused on a personal shopping experience for apparel.

She did not make a landing page. She made a Google Form and sent surveys to friends of friends about shopping habits. After roughly 20 of those tests, she learned three things: people were willing to be styled by someone they did not know, they were open to buying brands they had not heard of, and the service could work without an in-person stylist.

The tool was deliberately boring. A form was enough to learn whether the job was valuable. Building an account system would have answered a different question.

Use constraints as a test budget

Lake used a credit card with a $6,000 limit to buy clothes. She sourced from boutique stores with 14- to 30-day return windows and returned whatever did not sell. The test did not make money, but the return windows meant she could learn without turning the first experiment into a permanent inventory bet.

The exact numbers matter because they describe the shape of the risk: a $6,000 credit limit, a bounded return period, and a small group of real orders. “We will validate cheaply” is not a plan until the cash exposure and the exit condition are written down.

Stitch Fix passed 100 orders without a website. Customers emailed Lake and filled out the survey. She bought the clothes, assembled the Fix, and shipped it. The service was the product while the product was still being discovered.

Let operations teach you what to build

The historical monthly sequence after the seed round was 29 orders in April, 35 in May, 110 in June, and 235 in July. I care less about the curve than what it implies. The team was learning from completed deliveries before it had a self-serve funnel.

Eventually Stitch Fix launched a landing page where people could complete the survey and join a waitlist. The wait could be two weeks or three months depending on the team’s ability to serve the customer.

That waitlist was connected to operations. It did not promise instant access to a finished app. It managed demand while the company learned how much styling and fulfillment work each new customer created.

That is a much healthier use of a waitlist than collecting a giant number and hoping the team figures out fulfillment later.

Turn customer stories into distribution

Stitch Fix grew through word of mouth and posts from customers on Instagram and Pinterest. The issue also describes the team reaching people through Instagram accounts with audiences of roughly 50,000 to 100,000 followers.

Those channels worked because the product created something visible to share. A box arrived, the customer tried the clothes, and the experience gave her a story. Distribution followed the service instead of replacing it.

I would not copy the channel list mechanically. I would ask what part of the service creates a customer story that can travel without a paid ad.

The founder decision

If you are building a product with a large service component, ask:

1. Can a form, spreadsheet, or email complete the core job? Start there if the answer is yes.
2. What is the maximum cash exposure? Use bounded inventory, returns, deposits, or time boxes to make the test survivable.
3. What operational step repeats across the first 20 customers? That is often the first software worth building.
4. Does demand need a waitlist because supply is limited? Make the wait reflect a real capacity boundary.
5. What part of the delivered experience gives customers a reason to talk? Build distribution into the service.

Sources and original research

• Original First 1000 case: 👗Stitch Fix, December 22, 2020

The Stitch Fix First 1000 issue supplies the historical validation setup, units, order sequence, waitlist range, and distribution examples above; the service-first checklist is my adaptation. Its dates around the later landing-page timeline are not fully consistent, so I do not turn its “seven months” phrasing into a precise duration.