This is a PostHog customer interview, published 2 April 2025. PostHog wrote it. Zealot did not publish a separate engineering note.
Brandon Jakobson is co-founder and CTO of Zealot, which the page calls an AI customer-activation platform. He installed Amplitude and PostHog side by side. A person had recommended PostHog. Amplitude was the tool Zealot’s enterprise customers already used.
The Amplitude → PostHog playbook is the export, the identity map, and a labeled history import. Zealot’s interview is the overlap, the seat-price moment, and a second removal (BugSnag). It does not describe an Amplitude export.
What the overlap showed
Brandon calls Amplitude clunky. While both collectors ran, Amplitude missed data points that PostHog captured. He says PostHog became the tool people opened, and that it was ahead of Amplitude on the events they cared about.
He then invited the rest of the team. The line PostHog quotes is that he was glad they do not price by seat. Seat pricing on Amplitude was the reason he had held invitations back.
The interview gives no event count, no date range for the dual-run, and no dashboard that shows the missing points. “Missing data points” is his observation from using both. It is not a published reconciliation.
BugSnag left in the same change
Error tracking came next, after a recommendation, while PostHog’s version was still in beta. Zealot was on BugSnag. SmartBear had bought BugSnag. Brandon calls that tooling old, and says they kept it because alerts arrived.
The friction was triage. He and co-founder Srikar spent time in logs, matching timestamps, and still could not tell a user-visible failure from JavaScript noise. He wanted to know whether a person had seen the error.
PostHog’s version let him open an error, see who hit it, and open their session replay in two clicks. He says that was enough to remove BugSnag and Amplitude from the codebase in one pull request. He also mentions PostHog’s own AI chat, which he had liked in Amplitude and then found in a PostHog beta.
The page does not list the SDK calls they deleted, or how long replay retention they turned on.
What they say they gained
One analytics and error tool, with replay attached to the error. A pricing shape that let him add the whole team. Direct contact with PostHog engineers, which he contrasts with a support queue on Amplitude. He names David as an example of that contact.
They were about to ship more AI features and wanted LLM observability to already be in the same product. That is a plan at the time of the interview, not a measured result.
No cost delta, no query-speed number, and no imported Amplitude history appear on the page.
What I would copy
Run both collectors before you delete the old one. Zealot’s reason to drop Amplitude was a gap they could see only because PostHog was already recording. A single-tool trial hides that.
Check the seat line before you invite the company. Brandon’s switch from “I use this” to “the team uses this” was a pricing event.
Treat the BugSnag removal as a separate migration that happened to share a pull request. Session replay answered a question BugSnag’s alert did not. If you only need product analytics, you do not have to change error tracking in the same week.
History does not move because you removed a script tag. The Amplitude → PostHog playbook is the export and the label on imported events. Zealot never published that work. If your charts have to stay comparable to last year, this interview is the why, and the playbook is the copy.