Community access that starts from the product the seller already sells
- Problem
- ablefy sellers had the product, the buyers and the payment on one platform. A new community had to decide who gets in, and the first design answered that from the system's point of view: a paywall in settings, or private access with approval. Sellers who wanted the community as a bonus to an existing product could not find the setting at all.
- Solution
- Four versions, each changed by what sellers did in sessions and what the data showed. Access became reachable from the invite flow, then one setting reachable from both places, then its own page with fewer things on each screen. An assistant that set access for the seller was tested and dropped. When more than one product could be linked, access and invites became two separate flows.
- Result
- No friction found in later user tests or support cases. 68% of communities link at least one product; 36% of those set up a paywall. The version worth launching with was designed in three months, then a beta of 22 onboarding sessions before launch.
Role & team
Sole designer on the product
- Strategic planning, user research, design system, product prioritisation, design iteration, prototyping with code.
- Squad: one project lead, three engineers, one marketing.
- Planned and ran the beta programme. Handed the product over in January 2026 to take on customer insights.
Duration & scope
June 2025 to January 2026
- Vision in June, MVP tested with five sellers in July, the launch version with its own design system by September, open beta on 2 January.
- Two research rounds: five sellers on the MVP prototype, then 22 onboarding sessions in four weeks before launch.
- Part of a wider community research programme: 40+ interviews.
Key metrics
68% of communities link a product
- No friction found in later user tests or support cases; every follow-up interviewee satisfied with the result.
- 36% of linked communities set up a paywall: new revenue for ablefy.
- 41% of invite-only communities saw the linked product's revenue rise, not all of it attributable to the community.
At ablefy, a creator-commerce platform, sellers already had the hard parts in one place: a product, the people who had bought it, and the payments behind that. What they did not have was anywhere for those people to talk to each other.
I was the only designer on the rebuild: a community product from scratch, on a new tech stack, in a new squad. Rooms, posts, members and a join page, under the seller's own name and colour, designed for the phone from the start and shipped as an installable web app. The product also gained an AI setup flow that drafts a community from the seller's product. This case stays on one decision, because it is the one that taught the most: access.
As it ships: one choice, then one product
The answer first. A seller picks how people join: request to join, or pay to join with one of the seller's own ablefy products. Members come in through a separate page: by email, by link, or through an ablefy product, one step at a time. The rest of this case is how it got here.
View full size
View full sizeTwo use cases, designed from the system
Two ways a seller would use a community. One: the community is the product. In settings, under access, the seller picks the ablefy product that is the paywall. Two: the seller adds the community to products they already sell. For them, private access with approval was the temporary answer.
View full size
View full sizeSellers looked under Invite, not under Access
In the first user test, the second group could not find the setting. They were not thinking about access; they were thinking about inviting people. So access became reachable from the invite flow, where they looked.
View full sizeOne product per community, reachable from both places
A technical constraint at the time: only one ablefy product could control access to a community. So each community served one use, either/or, and the same setting was offered in both places so both groups could find it. That matched how most sellers plan a community anyway, from the earlier interviews: one community, one purpose.
The tests and the data then said two things. Two entrances to one setting read as two different settings, while the linked product was synced between them; sellers saw two things that mysteriously agreed. And the access page itself did too much: sellers called it complex and overwhelming, and tracking showed almost nobody used the manual invite section at the bottom.
View full size
View full sizeThe assistant that made it worse
In between, a prototype let an assistant set access from a description of the community. For a setting this important, sellers wanted to click through it themselves. Reviewing a result they had not built confused them more. This was not a problem for AI to solve. It was a design problem.
View full sizeFewer things on each screen, one decision at a time
What shipped: the steps separated so one screen carries one decision, a clearer UI, and the unused manual invite gone. Soon after, more than one product could be linked. The invite page was redesigned on the same idea, and the two flows stopped syncing: access and invites are two use cases in two places, and the invite side takes several products.
That covers the three ways sellers use it. The community as the product, with one paywall. The community as a bonus to the seller's other products, through invites. And both, where existing buyers of a product join free and everyone else pays. Two more flows came with it: customers whose product access expired are marked and removed in one step, and synced buyers wait as pending until the seller sends the invitations, so access can be set up before the community is ready.
View full size
View full sizeSellers in from the first month
Vision and the test plan in June 2025. In July an MVP, tested with five sellers before anything was built out. The version worth launching with was designed in three months, with its own design system, so engineers were never waiting on screens. From September, real sellers were in a closed beta.
I planned and ran the onboarding sessions and sat in 22 of them across four weeks, 21 November to 16 December 2025, each a recorded walkthrough of a seller setting up their own community. Engineers joined the sessions rather than reading a report. The backlog was reprioritised on what those sellers said, and a change made after one session could be watched working or failing in the next.
View full size
View full sizeWhat it did, and what I would do differently
Open beta on 2 January 2026. The ablefy access connection merged in the last week of January, the week I handed the product over to take on customer insights. The product and its research kept going.
Voice of Customer, the work that followed
- No friction found in later user tests or support cases; every follow-up interviewee satisfied with the result.
- 68% of communities link at least one product. 36% of those set up a paywall, new revenue for ablefy.
- 41% of invite-only communities saw the linked product's revenue rise, not all of it attributable to the community.
Find where sellers look first
Before deciding where a setting lives. A week of paper testing would have shown it, and saved a version.
One screen, one decision
The page that did too much lost sellers; the version that asks one thing at a time did not.
Some settings are made by hand
A decision about money and people is one sellers want to click through themselves. An assistant can draft a welcome post, not the access rules.