Our referral programme failed. Here is why
Eleven customers in four months. People recommend tools in conversation, not with links.
On this page
Last year we launched a referral programme: invite a friend to Tally and you both get a month free. Every growth guide recommends one. Ours brought in eleven customers in four months, and cost more to run than it earned. We closed it. Here is why it failed, and what we learned.
What we built
The programme was simple. Every account had a personal link in Settings. When someone signed up through the link and paid, both accounts got a free month. We announced it by email, in the app and on the blog.
In four months, 3,100 people visited Settings. 410 copied their link. 73 people signed up through a link. Eleven became paying customers.
Why it failed
We interviewed twenty customers, including several who had recommended Tally many times without using the link. Three reasons came up again and again.
- The reward was too small for the effort. A free month is worth a few dollars to a team. Nobody goes looking for a link for that.
- People recommend in conversation. Most recommendations happened in a meeting, a group chat or a conference. Nobody paused to find a referral link.
- It felt awkward. Some people did not want their recommendation to look like they were being paid.
I tell people about Tally every month. I did not know there was a link. I would not have used it anyway.
A customer, in an interview
What works instead
Our customers do recommend Tally. When we asked new customers how they found us, a third said a friend or colleague told them. That is the real referral programme, and it does not need a link. What it needs is a product worth mentioning and a moment where people show it to someone else.
So we are putting our effort there instead. We are building charts that can be shared with a link, and every shared chart will carry a small Tally line at the bottom. A product people show to each other does not need a referral link.
What we learned
- Measure the cost, including the engineering time and support questions.
- Ask your customers how they recommend things before building a programme around it.
- Make the product easy to show, not just easy to recommend.
