Collaborators: Cory Maxwell-Coghlan
During the Christian season of Lent(1) I created a surprisingly successful chatbot named Len.
Len had excellent engagement, retention, and word-of-mouth growth. I'm writing this to document my learnings and how it was built.
What was Len?
Len was a daily SMS based devotional built using Dexter.
Users registered by texting "Hello" and thereafter received a text at 7am every morning with a daily devotional.
The devotional content was specifically designed for the season of Lent. You can view the landing page for Len here.
How Len Grew
Len had terrific growth. By the end of Lent, Len had 624 subscribers receiving a daily message and ~46% of them were engaging every day by responding via text.
This happened with no paid promotion, growth hacking, etc. Here's what Len's growth looked like over the season of Lent*:
Clearly from the chart there were some spikes in subscribers and as far as I know those came Instagram posts.
One example above, was responsible for a one-day jump from 192 subscribers to 313.
Due to this organic growth and retention, we viewed Len as a successful project. To further validate this assumption, we sent out a survey link via text message at the conclusion of Lent.
The survey was sent to 579 Len users and 148 completed the survey. The results were encouraging!
1. 95% of respondents wanted to continue using Len as a daily devotional service.
2. 27% of respondents would be willing to pay for Len!
3. The majority of people preferred a "Pay what you want" model.
4. Finally, and most surprisingly, 60% of respondents were open to an ad supported version to keep Len free.
Given the engagement and feedback we received, I consider Len to be very successful. Here's my summary of what I believe attributed to the success.
People loved communicating with Len and checking the logs I regularly saw people texting "Len" back as if it was a human. At a basic level personification makes Len more fun. It makes relating to it more natural and compelling.
Here's a great example that showed up in our feedback:
- Tying it to a real world construct
In Christian thought, the concept of Lent is well known. Tying Len to Lent meant the value proposition was immediately accessible. Len helped people do something they already wanted to do (engage in Lent) rather than try to get them to start a new behavior.
- Daily with clear expectations
The expectation of a daily, predictable message was set at the very start. Un-predictable messages quickly feel spammy.
- Using SMS
There is still nothing that beats SMS in terms of simplicity. Len could have been deployed on Facebook messenger to provide a richer multimedia experience but I do not believe those benefits outweigh the ultimate simplicity of SMS.
- Having the landing page.
A simple landing made sharing Len much easier (especially on Instagram). Here is the one I setup. This drove growth.
- Quality content
The actual content was all original and well written. It was done by Cory Maxwell-Coghlan.
- Content format
The typical day started with a verse from the Bible, and then the user could text back to receive a short reflection on the verse with a question. Finally, the user could text back one more time to receive a prayer. This type of interactive, short form content was ideal for SMS.
What didn't work too well:
Len turned out to be surprisingly expensive to run! We spent $1,850 on the SMS costs via Twilio during the duration of Lent. This was offset by a soft donation ask we did (raising $710 from 22 people) but running a daily SMS service can get expensive fast!
- Overloading 10 digit numbers
Due to cost constraints, we operated Len on a standard, 10 digit phone number. Those are rate limited to only send 1 message per second. Since we did all our messages at 7am this caused MASSIVE delays in responses. This seriously degraded the experience.
If we were going to continue doing Len, it would have to be on a short code to accommodate the volume. That means we'd be paying a minimum of $1,000 a month. :/
- Human Error
There was a lot of copy + paste involved in loading the content daily. This led to simple spelling mistakes (including sending the wrong verse on Easter Sunday 😭.
Conclusion & Len beyond Lent?
There is demand for a digital daily devotional service and I'm confident a viable business could be built.
I really wanted to keep working on Len but due to a few factors, we've put things on pause. It became clear we wouldn't be able to operate Len without charging for the service in some form, that meant increasing the technical burden and everything that comes along with it. I'm also not currently in a position to quit my day job to pursue this full time. As such, Len is currently on vacation.
It's my hope by publishing this other people may want to run with some some of the conceptual underpinnings of "Len".
Beyond that, the relationship between spiritual growth and technology is very interesting. Len was primarily a content delivery mechanism. But what happens when Len gets smarter? Where is the line between technology enabled spiritual growth and technology enabled spiritual manipulation? Who will be the first person "converted" while texting with an AI bot? And what does that mean about how we view faith?
Those are interesting questions I hope can be explored alongside the technology.
If you're interested in this or doing anything similar! Please let me know! My email is on the side.
(1). Lent is a season on the Christian liturgical calendar marking the 40 days leading up to Easter. It's traditionally a season of fasting from something to create a greater awareness of our reliance on Christ.
Appendix: Building Len
I built Len with Dexter. Dexter is simple to use but also powerful. For Len, I kept things really basic and it took me ~4 hours to do the initial setup.
Using these tools in combination with about 30 minutes daily work I was able to create a fairly robust experience