If you’re getting this email you either subbed to one of my old newsletters or once asked to be kept in the loop about what I’m working on.
I’m trying to get better at telling the world about what I’m building.
So starting today, every week I’ll post a short update here about what I did this week and how I’m thinking about things.
A personal wiki powered by language models.
I’m working on Unriddle, a personal wiki and writing assistant that’s powered by language models. It learns continuously from your reading and notes to provide writing suggestions and display highly related content.
We launched six months ago and today it’s at $206K ARR with 95K users.
The premise is pretty simple.
When we form a new memory or learn anything new, we connect it to existing memories. But these connections fade over time, which is why we only remember a fraction of the information we consume.
Unriddle is designed to do all the remembering for you. It works just like your brain, drawing connections between things you read and write about.
It understands the context behind your writing and then links it together based on meaning. So that when you start writing you see relevant content from your library as you type.
Exactly what you need at the exact moment you need it.
The last two weeks.
Our focus lately has been increasing conversion.
We’re running an influencer marketing campaign and every time a video pops off (250k+ views), we get a ton of traffic but only ~1% of visitors convert to paying users.
1% isn’t terrible but it could be better.
Ideally, I want it at ~3%.
We started by creating a new landing page with much clearer messaging and a design that appeals more to students and researchers, our main users.
Next, we added onboarding emails, the first of which sends on sign-up and then another sends each day for the following three days.
Finally, we built an onboarding survey to get more concrete data on who’s most likely to convert. This helps us be more targeted about which influencers to work with, e.g. PhD students watch v different content to 1st year undergrads.
So far it’s working!
Since making these changes conversion is up to 2%.
And this was only focusing on the first couple steps in the funnel — finding the right people (onboarding survey), convincing them to try Unriddle (landing page) and prompting them to come back (onboarding emails).
Next, we’ll be working on feature discovery for new users and better paywall UX.
On the marketing side, we’ve seen explosive growth since launching our influencer campaign two months ago — 4x’d MRR and 3x’d DAUs.
It’s going well.
For every dollar paid to creators last month, we made $5 in revenue. Here’s one from the last batch.
Since we have a channel that’s working the plan is to double down over the next few weeks with more ad spend and reassess if it shows signs of waning.
Second to conversion and content, we launched a few smaller features like Graph View and document sharing for Teams. But I’ll say more about these another time because I’m trying to keep these posts brief.
That’s all I got for update #1 :)
love this - please keep these updates coming :)
re: "For every dollar paid to creators last month, we made $5 in revenue. Here’s one from the last batch."
How'd you find these creators to make unriddle content?