Questions, answered straight

You are probably here with one specific doubt. The hardest questions are at the top, not buried at the bottom, and the answers include the parts that do not flatter us. Every figure below is computed live and carries its grade.

What does this actually do?

Cerqular connects to where you already get paid, reads your real sales, and finds specific money your own numbers say is available: buyers who are due to return by their own cadence, one-time buyers of something people reliably repurchase, a product priced below what your buyers already spend elsewhere in your catalogue. For each one it drafts the email or post that goes after it and puts that draft in a queue. You approve every one. Nothing sends by itself. When a payment lands, it traces that payment back to the thing that caused it, so what you see is which action actually earned money rather than a guess.

See what it finds on your own numbers, no signup

What can it see and do with my payment account?

Reading the live contract.

Who is behind this, and how established are you?

Cerqular Inc. And the honest answer to the second half is: we are early. We have no customer results to show you. Every figure on this site is computed from seeded sample business books, labelled as samples wherever they appear, because inventing customer outcomes we do not have is the one thing that would make everything else here worthless. If backing a company at this stage is disqualifying for you, that is a reasonable position and better known now than after you have paid us.

Read what we do and do not claim

How much will I actually make?

Reading the live contract.

Will this work for my kind of business?

Reading the live contract.

How fast will I see something?

Two different clocks, and conflating them is how software like this oversells. The work is immediate: the first drafts and the first findings arrive in your queue on day one, and the hours those save you are yours from that day.

Measured revenue is not immediate, and cannot be.It follows the real purchase cycle of your business. On our sample books the median gap between a buyer's purchases runs to months, not days. So if somebody promises you revenue this week, they are describing a business with a different cadence than yours probably has. We will not put a date on your first measured dollar, because we would be guessing at your customers' behaviour.

See your own cadence, from your own sales

What if I have very few customers?

Then several of the mechanics will refuse to produce a number, and they will tell you which and why. Each one has a floor, for example a minimum number of real repeat gaps before it will claim a cadence exists. Below that floor it holds back instead of producing a small number to fill the slot.

That is deliberate, and it is worth understanding before you pay: a thin book will produce a short list of refusals and a smaller figure. You will still get the drafting and the hours back. You will not get a fabricated opportunity, which is what a smaller number would be.

Find out which side of the floors you are on

Is this just a chatbot with a wrapper?

The difference is which part a language model does. It does not compute any number here. The figures come from deterministic arithmetic over your own transaction rows: how many buyers sit past their own median repurchase gap, what your real repeat transaction value is, what the measured difference in lifetime spend is between buyers who entered through one product versus another. That is arithmetic on your data, and it produces the same answer every time.

A model writes the copy, and only the copy. Then a separate mechanism does the part a chat window structurally cannot: it puts a unique reference on the link in that copy, and when a payment arrives carrying that reference, the payment is matched back to the specific action that caused it. That is the difference between advice and measurement.

Watch the arithmetic run on your numbers

Will it sound like me?

Not perfectly on the first draft, and anyone claiming otherwise has not tried it. It starts from what your existing sales copy and your products tell it, and it learns from what you approve: every asset you confirm feeds back into the voice it writes in next time. The drafts that sound wrong are the ones you reject, and rejecting them is how it improves.

The safeguard is that nothing reaches your audience without you approving that exact text, so a draft that sounds wrong costs you the time to read it and nothing else.

Could it embarrass me in front of my audience?

The honest failure mode is a draft that sounds off, and you will see it before anyone else does. There is no autonomous sending in this product. Every asset waits in a queue for your explicit approval, and you are the sender of record on your own email provider, using your own domain and your own unsubscribe list.

Separately, generated copy passes an automated honesty check before it reaches you: it rejects income guarantees, invented statistics, and claims about results from customers we do not have. That check is why some drafts arrive shorter than you might expect.

What does it cost, what happens after the trial, and how do I cancel?

Start your 30 days for $1. After those 30 days it moves to the standard monthly price for your segment, which differs by segment because the businesses differ.

Cancel anytime. Cancelling is a setting in your account, not an email to support and not a retention call.

See the price for your segment

What happens to my data if I leave?

You can export it, and you can delete it. Deleting your account deletes your data; we keep it while your account is active and not after. You can disconnect a payment or email connection at any time without deleting the account, and doing so removes our stored credential for it.

On the anonymous audit, the one you can run without signing up, nothing is stored at all.

Read the deletion and retention detail

Figures on this page are read live from our public contracts each time it is rebuilt, so they cannot drift from the product. Where a figure could not be read, this page says so rather than showing a number typed in by hand.