Why AssemblyAI switched from Mixpanel to PostHog
Mar 23, 2022
AssemblyAI is an API platform which provides powerful AI models to transcribe and understand speech. As a young company in the cutting-edge field of AI modeling, AssemblyAI requires teams to wear many hats and move fast, making immediate access to reliable data for decision-making a must.
To enable this, AssemblyAI had previously used Mixpanel, but as the company grew and users flooded in, Mixpanel’s pricing model made it increasingly expensive.
“We started throttling events,” said Alberto Santos, AssemblyAI’s Web & Brand Lead. “At the time I was doing A/B experimentation on a different platform, then analyzing the results in Mixpanel. When Mixpanel’s pricing ramped up we looked at other platforms, including things like Segment — but PostHog, which can do both experiments and analytics in one, was clearly the winner.”
Unthrottled ingestion for millions of events each day
When AssemblyAI switched to PostHog, they were able to start tracking a much greater volume of events - millions per day - without the need to throttle ingestion. As a result, analysis in PostHog was much more reliable and the team was able to make decisions and test ideas faster.```
“PostHog helps us debug support issues, because we push errors to PostHog as events. It may not be exactly what PostHog was intended for, but it is really useful and shows how adaptable PostHog is.”
“Throttling meant we only had a very rough idea how much people were using us,” said Alberto. “So, when we actually needed to do proper analysis, we couldn’t rely on Mixpanel — we had to look at the logs and waste engineering resources on that.”
As a result of switching to PostHog, AssemblyAI was able to unthrottle event ingestion and empower the entire team to use tools such as trend insights and path analysis to analyze the entire data set.
“Everyone here uses PostHog,” said Alberto. “Literally every single person, from customer support to content writers; engineers to marketers. We do everything from debugging support issues to measuring site traffic in PostHog.”
“Finally having a full view of what users do has helped us so much. It’s helped us improve conversion, improve our support, and optimize the user journey through the platform.”
Fuelling a culture of constant iteration and growth
Now that the team finally has access to all their data, they’ve been able to discover meaningful enhancements and insights on a regular basis. Often these don’t lead to single, dramatic changes in the product — but they do fuel constant iteration and improvements.
“I feel like, every single week, we discover something new that is making a difference,” said Alberto. “We’re an API company, so a lot of our users just come to the site, grab their token, and leave.”
“PostHog really helps us optimize that journey to help our users — and, for us, having visibility of all the data at scale in PostHog is just magical.”