We had a waitlist of over 150 companies and no idea whether any of them would pay. Beta access now costs $9, which converts to a full Starter month of 2,000 credits when the account opens, and is refundable until then. The reasoning: a free signup is not evidence, a paying user gives feedback a free one does not, and charging early is the cheapest way to find out you are wrong. The strongest objection is that it shrinks the sample we learn from, and that objection may turn out to be correct.
We had a waitlist of more than 150 companies. It was the number we put in decks. It also told us, on reflection, close to nothing.
Beta access now costs $9. The free plan has come off the site while the beta runs. This is the reasoning, including the parts of it that cut against us, because a post that only argues one side of a pricing decision is marketing rather than reasoning.
What a waitlist actually measures
A free signup measures whether someone liked a sentence. That is not nothing — it means the positioning landed — but it is a measurement of our copy, not of our product, and certainly not of demand.
The gap shows up when you try to use the number. 'Over 150 companies waiting' implies revenue if you multiply it by a price. But nobody on that list evaluated a price. They evaluated a description, at zero cost, with no decision attached. Multiplying it by anything produces a figure with no relationship to the world.
We were carrying a number like that around and quietly treating it as validation. That is a comfortable mistake to make and an expensive one to keep making, because it tells you to build more of what you have.
What the $9 is, precisely
It is not a queue fee, and the distinction matters enough that we redesigned the page around it.
$9 buys 2,000 credits — one credit per action an agent takes — which is a full Starter month. After public launch that plan is $39. So the $9 is a purchase at roughly a quarter of the price, and getting in early is the consequence rather than the product.
It is a single payment, not a subscription. It is refundable in full at any point before we switch your account on, without an explanation. And the credits land the day the account opens, not before, because credits you cannot spend are not credits.
The three arguments for
One: a payment is the only pre-product evidence worth having. Everything else — waitlist size, survey responses, encouraging conversations — is people being polite about a hypothetical. Nine dollars is small, and it still requires someone to decide the problem is real enough to act on today. That decision is the signal.
Two: paying users give better feedback. Free users churn silently, because they have nothing invested and no standing to complain. Paying users email you and tell you what is broken, sometimes sharply. For a product in beta, sharp feedback from ten people is worth more than silence from five hundred.
Three: it is the cheapest way to find out we are wrong. If almost nobody converts from free interest to a $9 decision, that is a finding, and it arrives now rather than after six months of building. We would rather learn it at this price.
The argument against, which is real
The honest cost is sample size.
Conversion from a free list to any paid action is brutal — most of a list does not convert at any price. So we are trading a large population we could have watched for a small one we can learn from properly. For a product still finding its shape, fewer users means fewer situations, fewer edge cases, fewer of the specific breakages that only appear when someone uses the thing in a way you did not imagine.
There is a version of this decision that is simply wrong, and it looks like this: we optimise for a metric that sounds good in an investor conversation — 'paying users before launch' — and pay for it with the learning that would have made the product good. We do not think that is what is happening, and we are aware that everyone who makes this mistake also does not think that.
The mitigation is the refund policy and the batch size, not cleverness. If conversion is bad enough that we are learning nothing, that is legible quickly, and the free plan is a decision we can reverse in an afternoon.
Why the free plan came off
Mostly coherence. A page that advertises a permanently free tier and then asks $9 to get in is making two offers that contradict each other, and a visitor resolves that contradiction by closing the tab.
There was also an accounting problem in our own heads. While a free tier existed, every conversation about whether people would pay could be deferred — they will pay later, once they see the value. Removing it forced the question into the present tense, which is where it belonged.
It may come back at public launch. We think a real free tier is good positioning for this audience and we would like to offer one. It was not compatible with a closed beta that we want to learn from.
The part we will not do
We will not argue about refunds. If someone asks, it goes back the same day, without a retention script.
That is partly principle and mostly arithmetic. A refund costs us a processing fee. A chargeback costs the fee, plus a penalty larger than the sale itself, plus a mark against our standing with the payment processor — and payment processors suspend accounts that accumulate those. Anyone we would have argued with is someone who files a chargeback instead. There is no version where friction is the profitable choice.
If you are making the same decision
Two tests we would apply before charging for early access to anything.
Can you name what the money buys, other than a place in line? If not, you have a toll booth, and this audience — builders, indie hackers, engineers — will name it as one in public, at length.
Can you refund instantly without it hurting? If a wave of refunds would be a problem for you, you are not selling early access, you are borrowing from your customers.
We think we pass both. Ask us again in three months, when we will have the number that settles it.
Key takeaways
- A waitlist measures interest in a description. It does not measure willingness to pay for a product, and treating the two as the same is how companies build the wrong thing confidently.
- The $9 is a purchase, not a queue fee. It converts to 2,000 credits — a full Starter month, about a quarter of the post-launch price — and is refundable until the account opens.
- Paying users give different feedback. Free users churn silently; paying users tell you what is broken because they have standing to complain.
- The real cost is sample size. Charging cuts the number of people we learn from, and for a product still finding its shape that is a genuine loss.
- We refund without argument, because a refund costs us a fee while a chargeback costs the fee plus a penalty plus standing with the payment processor.