SAASPOCALYPSEverdict #POSTHOG-9DC4
scanned 2026.04.24 · 17:41
subject of investigation
posthog.com
▸ product analytics & session replay platform
verdict: DON'T
buildability score
18
/100
tier · don't
the blunt take
“PostHog is less a SaaS and more a small analytics empire — event pipelines, session replay, feature flags, A/B testing, and a data warehouse, all baked into one. You're not cloning this on a weekend, you're cloning it in a career.”
The homepage was mostly font data (which, honestly, is very on-brand for an analytics company obsessed with load performance), but PostHog is a well-known open-source product analytics suite. The real moat isn't features — it's the years of ClickHouse tuning, SDKs for every runtime, and the self-hosted trust that indie developers put in it.
cost breakdown.
their price ←→ your price
what they charge●
Scale plan (PostHog Cloud)
usage-based
/ events ingested
※ Generous free tier makes this brutally hard to undercut on price
annual:free up to 1M events/mo, then ~$0.000225/event
what it costs you✦
01 · ClickHouse cluster (the beating heart of this thing)$500
02 · Kafka or Redpanda (event ingestion buffer)$200
03 · Postgres (metadata, users, orgs)$25.00
04 · Session replay blob storage (S3)$150
05 · Worker fleet for pipeline processing$300
06 · SDK maintenance (JS, Python, iOS, Android, Go...)???
07 · On-call engineering when ClickHouse breathes wrong at 3amyour sleep
TOTAL / moyour 20s
▸ break-even:approximately never — PostHog's free tier covers 1M events/mo, making it nearly impossible to out-cheap
or, you know, use one of these.
if building feels spicy
option A
PostHog (self-host)
It's open source. Docker compose up. You get the whole thing free. This IS the alternative.
option B
Plausible Analytics (self-host)
Tiny footprint, Elixir + Postgres, genuinely shippable solo. Not a full PostHog replacement but covers 80% of indie hacker needs.
option C
Umami (self-host)
MIT-licensed, Next.js + Postgres, deployable on Railway in 10 minutes. Dead simple pageview analytics with zero ClickHouse nightmares.
what'll actually be hard.
est. total: ∞
▸ 6 months for event ingestion · another 6 for session replay · a decade for ClickHouse wisdom · therapy budget not included
easy
medium
hard
nightmare
01
medium
Basic event capture SDK
A JS snippet that POSTs JSON. Hard part is making it non-blocking and < 5kb gzipped.
02
hard
High-throughput event ingestion pipeline
Kafka + workers + at-least-once delivery. Dropping events in production will end you.
03
hard
Session replay recording & playback
rrweb handles the recording side, but storage, stitching, and privacy masking are a separate nightmare each.
04
nightmare
ClickHouse at scale
Fast OLAP queries on billions of events require schema design, merge trees, and materialized views that take years to get right.
05
nightmare
Feature flags + A/B testing consistency
Flags must be consistent per-user across SDKs, edge, and server. Distributed systems paper territory.
06
nightmare
Competing with a free open-source version of yourself
PostHog literally is the open-source alternative to PostHog. Your TAM is people who can't Docker.
recommended stack
ClickHouse (OLAP event store — the whole ballgame)Kafka / Redpanda (ingest buffer)Next.js + Django (PostHog's actual stack)rrweb (session replay capture)Postgres (org/user/feature flag metadata)
ready to build?
No build guide for this one. Some things you have to pay for.
▸ generated with love, by a heartless robotverdict v2.1 · saaspocalypse.dev