When to quit your startup and start over: lessons from WP Pusher to Reform

Executive overview

Building a product people appreciate but won't pay for is not the same as product-market fit. Peter Soom spent three years on Branch CI, a WordPress deployment tool that attracted buzz, partnerships, and investor interest — but never generated meaningful MRR.

Fake signals (acquisition interest, VC attention, developer praise) masked the real problem: the pain wasn't burning enough. Once the largest possible partnership failed to move the needle, Peter faced a clear choice about what to do next.

If you can't manually convert a single customer through direct conversation, no partnership or marketing channel will fix it.

Signs Branch CI wasn't working

• Early prepaying customers never actually used the product
• Repositioned and pivoted multiple times without MRR growth
• Partnership with the largest WordPress host generated almost no signups
• Customers expected a full sales process for a $50/month tool — a fatal unit economics mismatch
• The problem was real but not urgent: developers were already deploying, just not optimally

Why fake signals kept Peter going

• Inbound acquisition interest from WordPress hosts
• VC investment from a top-tier firm
• Enthusiastic developer responses to demos
• Co-promotion interest from major hosting companies

None of these substituted for customers who paid and stayed.

The decision to move on

• Tiny Seed's check-ins flagged MRR growth was too slow
• A final partnership that showed "a glimmer of hope" collapsed within weeks
• Peter was living in a one-room Copenhagen apartment, newborn in the background, conducting six customer interviews a day
• The real blocker: belief that taking investment meant he had to make it work at any cost
• Called every investor with a shaking voice; none wanted their money back
• One investor told him: "Just go have fun"

Criteria for the next idea

• Revisited a written list of business criteria at a cabin retreat
• Criteria overlapped significantly with Derek Reimer's (published on his site) before he built Savvy Cal
• Wanted a product where customers could self-serve without a sales process
• Wanted a proven category with demonstrated willingness to pay
• Wanted something he could brute-force early customers without intermediaries

How Reform was discovered

• Started from a different idea: a smarter investor update tool (typeform-style, one question at a time)
• Three founder friends said they'd pay $50/month — but usage didn't change, so the idea was dropped
• Playing with competing form builders during prototyping surfaced frustrations
• Jokingly suggested to his mastermind (Derek Reimer, Matt Wensing): "Maybe it's time for a new form builder"
• They agreed seriously; Peter started validating

Validation process for Reform

• ~20 conversations with form power users and people who worked at form companies
• First idea: a markdown-based form builder — developers loved it, nobody wanted to pay
• Dropped markdown; refocused on a high-quality visual builder that was simply better than existing options
• Built a landing page in Figma, not a live product — sweated every pixel and line of copy
• Iterated with ~30 rounds of feedback before publishing
• Used April Dunford's positioning framework to write copy
• Prototype form was built on top of Netlify's form API to limit scope

Launch and early traction

• Landing page posted on Twitter: ~100,000 views, 500 signups on day one, ~1,300 by launch
• Hosted a Twitter Spaces event; Andrew Warner prepaid on the spot
• Four customers PayPal'd money before a product existed
• Took ~one month to rebuild the early access signup form in Reform itself (dogfooding milestone)
• Onboarded customers gradually in early access, prioritising those who needed fewer features first
• 60+ customers within a few months, with active usage and feedback

What made Reform easier to sell than Branch

• No convincing a boss — individuals could sign up and try it immediately
• No deployment or integration required to see value
• $20/month starting price with no expectation of a sales process
• Switching cost is real but manageable; many microconf and Tiny Seed properties migrated
• Bjorn (co-founder/contractor from Branch) noted: "It's a lot easier to get a customer, and then two, and then three"

Lessons on entering crowded horizontal markets

• A proven category means proven willingness to pay — not a reason to avoid it
• Differentiation through product quality and positioning can carve out a viable niche
• Landing pages still work as smoke tests; low effort = low signal
• Validation is a probability dial, not a pass/fail gate — each step moves from 10% to 30% to 60% confident
• Even with strong validation, failure is possible; Branch had validation and still didn't work