By Max Sherman ยท July 3, 2024
I'm hiring
Grade: F
The failure rate stayed flat around 20%. There's still more work to do here. I made a change mid-way through the month to stop a "banzai" effect where all parts of a multipart upload would be uploaded at once. This is counter-productive because it saturates the network's bandwidth, and it was also unintentional. Changing it to upload parts in serial didn't have an effect on failures but I think it's better still because it paves the way for building a more sophisticated upload manager which can implement two improvements.
First, it can progressively test the bandwidth and upload more parts in parallel if the bandwidth supports it. This means better networks can upload faster and not be penalized in a "least common denominator" way because we are trying to accomodate poor networks.
Second, (though less related to serial uploads) we can start uploads in the foreground initially, rather than waiting for the service worker to start he upload.
Grade: D
This remained pretty flat with only a slight improvement around 25-35%. I'll credit the improvement to two things.
First, I implemented a sticky sign up form at the top of the mobile page as you scroll down. Second, I changed the headline to be a bit more punchy. I think a lot more work can be done on the hero section to increase conversions.
Grade: C
Throughout June there was a slight uptick here, from an average of 40% to 50%. Potentially from our email drip campaign. I'll call this a C because apparently I wasn't measuring this correctly at first, and thought it was below 40% somehow.
Grade: F
This has been pretty consistently around 15-20% the whole month.
Grade: F
Still need to set up acquisition channel.
Grade: F
Didn't get to this. I'll probably remove this goal
Grade: D
This didn't happen but there are updates in this direction, see the next section.
Last month I wrote about hiring an intern, but after thinking about it (and interviewing one person) I realized that this is going to result in hiring someone who doesn't know much and requires a lot of mentorship. I don't mind coaching, but the cost to value tradeoff is pretty bad. In some cases, you can hire an unpaid intern but for software I don't believe it's common, and the laws about this in NYC are very strict around the idea that basically you have to lose money to train them. For me it likely doesn't make much sense.
Instead, I'm going a different direction. I'm currently interviewing Phillipines-based software engineers because you can find engineers who are much more senior for lest cost than a US-based engineer. So far I have interviewed a few dozen applicants, and there are some strong candidates.
I think my offer to these candidates can be a lot stronger, because I can set performance based compensation goals like "double revenue and you'll get a bonus" and still have that be close to break-even for the business. Versus an American engineer, I would need to make the goal "100x revenue" to maintain the same economics.
I'm optimistic that having more help on the engineering is going to accelerate the growth of the business dramatically.
This goes back to the question "what is the bottleneck of the business?". I think it is product quality, and under this umbrella falls the tasks of improving conversion rates at every point in the funnel. Critically though, I want to start nurturing a new customer acquisition channel: word of mouth.
Word of mouth isn't going to happen unless the product is so good that people are willing to talk about it. This means that in addition to optimizing conversion rates, the work output needs to be completely stellar.
I want to start investigating how we can improve the quality of the generated minutes by looking at what we're generating already. I'm also going to build out a feature where you can regenerate minute with feedback to try and have a low-friction way to gather feedback from users.
Eventually for product-led growth, I think we'll need to add a referral system as well. Something like, "invite two people via email to the product and we'll give them 90% off their first month."