Groove, a help desk start-up with 4.5% churn, found that users who left spent about 35 seconds per session while those who stayed spent over three minutes. It emailed at-risk users offering a quick setup call and cut churn by 71%. In 2026 Groove's site redirects to Helply, an AI support platform with unlimited free seats and credits priced by task complexity.
Key takeaways
- Groove found its 'red flag' metric: churning users spent about 35 seconds per session, keepers spent more than three minutes.
- A timely, personal email offering help cut churn by 71%.
- GrooveHQ.com now redirects to Helply, an AI-native support platform (checked 27 October 2026).
- Helply plans start at $49 a month with unlimited users; AI work is paid in credits: 1 for a draft, 4 for a lookup, 12 for a full investigation.
- Free seats remove a classic churn trigger: customers cutting users to save money.
SaaS start-up Groove had a problem.
It was bringing in plenty of new users. But its churn rate was 4.5%. Customers were leaving almost as fast as they arrived, and the growth wasn't sustainable.
So it got funky with its user analytics.
(I made that joke in the book too. I'm sorry. It hasn't improved with age.)
What Groove found became one of the best-known churn stories around. And what Groove turned into in 2026 shows where that lesson goes next.
What the book said in 2022
In Chapter Nine, this was Tip #4: use your customer data.
Groove combed through its user data looking for "red flag" metrics, patterns that customers who abandoned the platform had in common. Two stood out: login frequency and session length.
The session length gap was huge.
- Users who left spent about 35 seconds per session.
- Users who stayed spent more than three minutes.
Low early session times hinted that a customer hadn't finished setting up. So Groove emailed that segment and offered a short Skype call to help. It sent similar offers to customers who weren't using features that would help them.
The result: churn fell by 71%.
The customer got better service. Groove got a chance to show off its product. Everybody won.
What's happened since
Groove's founder, Alex Turnbull, rebuilt the company for the AI era.
When we checked on 27 October 2026, groovehq.com redirected to Helply, which describes itself as an "AI-native customer support platform" built around AI agents that investigate, resolve and act. Helply's founder's note describes Groove as its original product.
The pricing changed shape too.
| Helply (October 2026) | Detail |
|---|---|
| What you buy | AI agents that investigate, resolve and act on support work |
| Starting price | $49 a month |
| Users | Unlimited ("seats are free") |
| How AI work is paid for | Credits, priced by task complexity |
| Credit ladder | 1 credit for a draft, 4 for a lookup, 12 for a full investigation |
| Trial | 1,000 free credits |
Why does this matter for churn?
Look at what happened to the job.
In the book's version, a person looked at the data, spotted at-risk users and wrote an email. In Helply's version, AI agents handle the support work. The company charges for work done, not for the people logged in.
And unlimited seats removes one of the oldest churn triggers in software. When budgets get tight, customers on per-seat pricing cut users. Fewer users means less usage, less value, and a much easier decision to cancel at renewal. If seats are free, there's nothing to cut.
That's a pricing decision doing the work of a retention program. For the full trade-off, see seat-based vs usage-based pricing.
What is churn really costing you?
Work out customer and revenue churn, and what it costs you each year, with the free Churn Rate Calculator.
Why the credit ladder works
Helply's 1/4/12 ladder is a neat piece of credit-based pricing.
Customers don't have to understand tokens, models or compute. They see three types of job and three prices. A draft is cheap. A lookup costs a bit more. A full investigation costs the most, because it does the most.
That's a value metric customers can predict. And predictable bills churn less than surprise bills.
If you're designing something similar, the credit pack designer turns your cost per action into a rate card.
Why the original email still works
Here's what hasn't changed.
Customers rarely cancel on a whim. They drift. They stop logging in, stop booking, stop opening. The red flag shows up weeks before the cancellation does.
Groove's real insight wasn't the email. It was knowing who to send it to, and when.
Steal this: the day-3 / day-7 rule
Pick the one early action that predicts a customer will stay (finished setup, booked a second session, used the product three times). Anyone who hasn't done it by day 3 gets a friendly offer of help. Anyone who still hasn't by day 7 gets a call or a personal message.
When is this the wrong approach?
If your churn comes from price or product gaps, an email won't fix it. Groove's customers were leaving because they hadn't set up properly, and help fixed that. Ask leavers why they went before assuming it's an onboarding problem. Our exit survey template gives you the questions.
And free seats only work when usage tracks value. If your costs rise with every user, don't give users away.
What you can steal
- Find your red flag. Compare customers who left with customers who stayed. What did the leavers stop doing early on? Our guide on how to reduce churn walks through the process.
- Email the at-risk group with help, not a discount. Offer a call, a setup session or a tip about a feature they're missing.
- Consider "unlimited users, pay for usage" if you sell to teams. It kills the seat-count objection.
- Price heavier jobs higher. A simple three-step ladder like 1/4/12 is easy to understand.
- Measure the result. Track your churn rate monthly before and after.
What is the 35-second session in your business? Which customers are showing you they're about to leave?
Churn Risk ChecklistTwenty questions that reveal where customers will leave.Open the free tool →Frequently asked questions
What is a red-flag metric?
A red-flag metric is a behavior that customers who later cancel tend to show early. Groove used login frequency and session length. Its churning users averaged about 35 seconds per session. For a gym it might be no visits in the first fortnight. For a software product, never completing setup. Find yours by comparing customers who left with customers who stayed.
What did Groove's churn email say?
The book describes the approach rather than the exact wording. Groove emailed users with low early session times, a sign they hadn't finished setup, and offered a short Skype call to help them. It sent similar offers to customers who could get more out of features they weren't using. The tone was helpful, not salesy.
Is Groove still in business?
Groove's founder, Alex Turnbull, rebuilt the company for the AI era. As of 27 October 2026, groovehq.com redirects to helply.com, and Helply's founder's note describes Groove as its original product. Helply calls itself an AI-native customer support platform, with AI agents that investigate, resolve and act on support requests.
Why would a software company give seats away free?
When AI agents do most of the work, the number of human users stops tracking the value delivered. Charging per seat also invites customers to cut users when budgets tighten, which shrinks revenue and usage together. Free seats with paid usage removes that objection and ties price to work done.
Does a churn email still work in 2026?
Yes. The tools have changed, but the principle hasn't: spot the early behavior that predicts cancellation and reach out before the customer decides to leave. Today you can trigger the email automatically from your software or booking system. A personal note from a real person still tends to beat a generic sequence.
Get both books free
The Subscription Playbook (ebook + audiobook) and Feed a Starving Crowd, instantly. In Australia? Claim the free hardcover.
Robert Coorey, MBA
Robert is the author of The Subscription Playbook and Feed a Starving Crowd and a Director at BPAG Accountants in Sydney. More about Robert.
Sources
- Helply homepage and founder's note (accessed 27 Oct 2026)
- Alex Turnbull, LinkedIn profile (Groove now Helply; accessed Oct 2026)
Facts and prices checked 27 October 2026. Spotted something out of date? Tell us.