Perch · Perch Partner Program Internal · the operating playbook

How we run the first five.

Three to five partners build the product with us, at our cost, and become the proof that wins the next ninety-five. The Growth Thesis says why this works. This says how it runs.

Start here

The operating manual for the partner program: the terms, who we pick, what we trade, and what "done" looks like. The plan sits in the Growth Thesis; the proof it works is in Role Models.

A

The terms

Cohort
3–5 partners, 1–2 general contractors + 1–2 subcontractors. Small enough to serve deeply.
The engineer
One forward-deployed engineer embedded per partner, inside their workflow, building against what they actually do.
The price
At cost. Partners cover our hard costs only: the FDE's time and token usage. ~80% below list · no margin.
The term
~60–90 days, time-boxed, one shared goal. A deadline forces a decision.
The exit
A paid contract, a reference, and a case study, agreed on day one, not hoped for at the end.
B

Why at cost

The pricing call

DecidedNot free, not full price. Free pilots attract tourists; full price is too much to ask of a partner who's also doing us the favor of building alongside us. At cost is the middle.

It carries enough commitment to filter the serious from the curious (Sierra ran paid pilots and converted 100%), and we take no margin, the rule a16z prescribes for early services: sell them at cost, never as a profit center. Partners pay only what they cost us: the engineer and the tokens. Evidence in Role Models.

C

Who we pick

Pick partners who bleed from the problem, not partners who think it looks neat. Pain over friendliness. The selection is the whole game.

+

Bleeds from the problem.

The pain is real and expensive enough that they'll give us access and an hour a week to fix it.

+

Has the bids to seed the loop.

A GC whose bid volume puts Perch in front of real subs. The partner is also the first turn of the engine.

+

A decision-maker who shows up.

Someone who can say yes, in the room, every week. No proxy.

+

Looks like the next hundred.

Representative, not an edge case. What we build for them has to generalize back to the platform.

Where we are: the credibility ladder
Partner 01 · live

Built Exteriors

Sub · founder embedded · $15k/mo. Proves the model works.

Partner 02 · next

Zwick

GC · proves it works beyond a friend.

Slots · open

1 GC + 1 sub

Fill to the full cohort, one more of each side of the loop.

Accelerant · target

A marquee GC

Bigger logo = bigger viral seed. Land once the program is proven.

D

The trade

We give
  • +A dedicated FDE inside their workflow
  • +Direct founder access
  • +Features shaped to their work that generalize back to the platform
  • +The product at cost
We ask
  • ·Payment at cost
  • ·System + jobsite access
  • ·A standing weekly working session
  • ·A decision-maker in the room

The ask is the filter. A partner who won't commit the access and the hour was never going to convert. Better to learn that in week one than month three.

E

The 60–90 days

Week 0

Select & sign

Pick the partner, set the one shared goal, sign the at-cost terms, and the case study, up front.

Weeks 1–2

Embed & observe

FDE inside their workflow. Watch how they actually bid. Don't ask, see. The first build targets come from what we observe.

Weeks 3–8

Build against what we see

Ship to their real work, every week. The standing session keeps the one goal honest and the scope from sprawling.

Weeks 8–12

Convert

The value is proven on their own jobs. Flip to a paid contract, collect the reference, write up the case study.

F

What "done" looks like

01

A paid contract

At list now. The at-cost window closes when the program does.

02

A reference

A logo we can name and a person who'll take the call.

03

A case study

Drafted from day one, the proof that arms the next pitch.

Three of these and we have a repeatable motion. Five and we have a market.