Getting Solopreneurs to Invoice
In a 50/50 test, my editor redesign raised the share of visits that start an invoice from 74% to 85%
In short
Started an invoice
85%
of editor visits
from 74%A/B test, early result
Issued and sent at once
28%
of editor visits
from 9%A/B test, early result
A/B test, early result: 50/50 split by seller, 28 Oct to 2 Nov 2026, 132 sellers.
- JimdoPay shipped in May. By summer it worked, and almost nobody used invoices.
- Research showed we were pitching the wrong benefit. I moved the message into screens sellers already open and rebuilt the invoice editor around the person you bill.
- In a 50/50 test, more visits started an invoice and three times as many issued and sent it at once. It's an early result.
We sold the wrong thing
Our Q4 2026 key results needed sellers to issue invoices: 500 monthly active customers, 1% of new customers invoicing (from 0.37%) and €100k GMV through the new JimdoPay. By September, 698 of about 1,150 sellers who had finished payment setup had never issued one.
A directional survey of 101 health practitioners on Jimdo found that none used Jimdo invoices, 39% wrote theirs in Word and half didn't know Jimdo could invoice. Medina Vultur from product and I ran interviews to find out why.
For them, getting paid was already solved. Bank transfer and cash cost nothing. The open problem was clean records when the tax deadline comes, and we had been pitching the invoice as a way to get paid.
Nothing new to build
My proposal needed no new feature. It changed where and how we told the story: inside screens sellers already open, in their words, with certainty and time as the benefit. I presented it at the Berlin offsite in September 2026.
FlowStep 2: the invoice waiting in the list.
FlowStep 3: the offer to get paid online, inside the invoice.
FlowStep 5: connecting a provider, with the best fit first.
FlowStep 6: the seller sees the email their customer gets.
The editor, rebuilt around the person you bill
The old editor lost 1 in 4 visits before a single field changed. I reordered it around the customer: who you bill first, then what you sell, dates last. Payment terms and notes moved behind "+ Add", and usage held at 19% of visits, against 20% before.
Each line item became its own numbered card. The draft saves itself, and one live preview sits next to the form.
Before
After
Choices before the handoff
I explored three line item cards and three places for the draft status and the preview before choosing. The handoff board marks every change for engineering, with the copy for each state.
Impact
From 28 October, half the sellers got the new editor and half kept the old one, with a rule to stop the test if errors rose by 5 points or more.
Started an invoice
85%
of editor visits
from 74%A/B test, early result
Issued an invoice
56%
of editor visits
from 52%A/B test, early result
Issued and sent at once
28%
of editor visits
from 9%A/B test, early result
Gave up
30%
of editor visits
from 38%A/B test, early result
Invoices issued a month
780
in October 2026, by 337 sellers
from 11 in AprilBefore vs after
A/B test, early result: 50/50 split by seller, 28 Oct to 2 Nov 2026, 602 visits from 132 sellers. Invoices a month: Mixpanel, April to October 2026. The product was growing too, so I don't count that number as the redesign's result.
Team and role
- Me, product design
- Ran the research with product, wrote the proposal, designed every state of the new editor, set the test metrics and the error guardrail, and pulled the Mixpanel numbers myself.
- Medina Vultur, product
- Ran the adoption research with me.
- Kristina Bardavelidze
- Recruited the research participants.
- Sabrina Schomburg
- Ran the survey.
- Kevin Rademan
- Wired the new editor.
What I learned
Honest read
Still early
132 sellers is a small group, so the gain could still be chance. The decision is planned for 24 November 2026.
Paid is the real goal
No invoice from the test is old enough yet to show whether it gets paid on time.
Next: Issue in the spotlight
Sending tripled although the redesign left the Issue menu alone. Save still competes with Issue, the button that brings in money, so Test 2 gives Issue the spotlight. Guardrails: drafts still issued within 7 days, and no rise in save or send errors.