Opened edrone to self-serve after ten years of sales-led growth, reaching 5,050 stores in 11 months with 7.8% converting to paid
edrone is marketing automation for ecommerce. Automated messages bring shoppers back to buy: an abandoned cart, a first order, the next one.
I decided what the setup keeps, what it asks, and what it decides on its own, then designed every screen that carries those decisions. Dozens of calls with small stores redrew each pass of the design.
For ten years every customer came through a Salesperson and stayed with a Support team. The product always had people doing its setup, which is exactly what a small store cannot pay for.
Small stores were never worth acquiring: the cost of a person sat on top of every setup, and no path ran without one. Almost nobody had ever run edrone alone.
Open edrone to the stores the old model could not afford to serve. No salesperson, no Support team. The product sells itself, sets itself up and earns its keep.
Ten years of the old model built 2,500 paying stores. Freemium had to reach the ones it never could, and most of those would never pay.
Active meant a store had connected its shop to edrone, had automations running, and was doing around fifty orders a month. Not logins.
A person had set up every paying customer, so the product had never stood on its own. Freemium had to turn everything that person used to do into decisions the product makes itself.
Those customers still wrote their own newsletters, so it was never all or nothing. I had to find the line: what is done before the store arrives, and what it still wants to do itself.
Then a third thing. A newsletter only counts if the next one is coming. Setup that runs once is a demo.
The build was quick; the eleven months were the iterating. Dozens of calls with small stores redrew the design each pass: the step they stalled on, the automation they switched off, what they asked us for.
A trial ends on a date. Freemium ends when the store outgrows it. Which one holds a store long enough to be worth anything?
No two stores want the same messages going out under their name. If the store has to decide, the setup is not finished. If I decide for everyone, some of it is wrong for someone.
Freemium, not a trial. Orders come when shoppers are ready, not when we are. A trial closes the account before that happens; a free one costs us nothing to leave open.
Everything is built from the store's own site in the minutes after signup. It gives us a URL and nothing else.
Account creation in full. Four stages, no question asked at any of them, all of it read off the one address the store typed.
Someone still had to decide what runs, and I took it. edrone knows what works better than a store that never ran marketing automation, so I switched everything on. Seven automations are live before the store logs in.
What the store finds waiting: seven automations, all of them running. Each keeps an off switch, because disagreeing with edrone should cost a click.
I started with two, to see whether a store would leave running something it had no hand in making. Nobody switched them off, so the rest followed.
Nothing is asked before something is shown. Tailoring means questions, and a question comes before anything runs. So the same seven run for a furniture store and a sock shop, each with an off switch. Disagreeing costs a click, not a setup.
Nothing sends until the store connects its shop, so that ask falls last, not first: by then the store has seen what it is getting.
The shipped onboarding, end to end. Four steps of showing what is already done, and one ask at the end of them.
Almost nobody had ever run edrone alone; a year later more than five thousand stores had. The entrance the old model could not afford to serve became the one most stores came through.
Signed up in under a year, without a salesperson or an onboarding team, the two costs that made a small store unprofitable. 393 started paying, adding 16% to a base built over ten years.
| Cohort | ACTIVE STORES |
|---|---|
| May 2025 | 44% |
| June 2025 | 33.6% |
| July 2025 | 40.9% |
| August 2025 | 23.2% |
| September 2025 | 37% |
| October 2025 | 51.5% |
| November 2025 | 56.4% |
| December 2025 | 49.9% |
| January 2026 | 74% |
| February 2026 | 77.6% |
| March 2026 | 80.3% |
Every monthly cohort above. Over the same months, the median time to a first attributed order fell from 35 days to 5. That is a sale from a shopper who came back through an edrone message.
Everything shipped as one preset. Next I would give the store a small set of choices, so it has a say without adding a step to setup.
The model was built to take Support out of the customer, and Support’s workload went up anyway. I never put a number on that, which is the first thing anyone should ask of a model that claims to remove a cost.