A product analytics stack that answers real questions

Early-stage founder who has shipped a product and now needs to know what users actually do: where they drop off, which feature sticks, whether the people who pay behave differently. Right now the answer lives in three dashboards nobody trusts, a Stripe export, and a gut feeling.

Wired & tested by Sameer + Ankit

Free · the verdict

You do not need four analytics tools to learn three things. Most founders stack an event tool, a separate session-replay tool, a BI seat for charts, and a CDP to glue it together, then pay four bills to answer one question badly. One free platform with events, replays, feature flags, and a real SQL layer covers everything an early-stage product needs, and it can pull your Stripe data in so paid and free users sit in the same query. Wire the snippet in an afternoon, ask real questions in SQL, pipe the one metric that matters to Slack. Pay nothing until you are big enough that paying is a good problem.

The stack

PostHog

PostHog SQL (HogQL)

Stripe (via PostHog data warehouse)

Slack

Metabase (optional, only if non-technical teammates need charts)

PostHogThe whole analytics core in one place: autocaptured events, funnels, retention, session replays, feature flags, and a SQL editor over your own data. This is the tool that actually answers 'where do people drop off' and 'do payers behave differently' without a second login.
$ Free plan is generous: 1M product analytics events, 5K session recordings, 1M feature flag requests, and 1M data warehouse rows per month, all resetting monthly. After that it is usage-based: product analytics starts around $0.00005 per event past the first free million, and volume discounts kick in automatically. PostHog says more than 90% of companies stay on the free tier.
PostHog SQL (HogQL)The part that answers the questions a dashboard cannot. HogQL is a SQL layer over your events, so you can write 'show me 30-day retention for users who hit the aha moment in week one' instead of clicking through a chart builder and hoping. It is the reason you do not need a separate BI tool yet.
$ Included in PostHog. No extra seat, no extra bill. The SQL editor and the data warehouse share the same free tier as the rest of the platform.
Stripe (via PostHog data warehouse)The revenue truth source. Linking Stripe into PostHog's warehouse pulls charges, customers, and subscriptions in, so you can join 'who paid' to 'what they did' in one query. Now 'do paying users behave differently' is a SQL answer, not a spreadsheet afternoon.
$ Stripe itself is free to connect (you already pay per transaction). The import rides PostHog's first 1M free warehouse rows per month. Webhook syncs are the cheapest way to keep it fresh, and most early-stage row counts sit well inside free.
SlackThe 'so you actually look at it' layer. PostHog subscriptions and alerts push your one north-star insight to a channel on a schedule, or ping you when a number jumps or tanks. Analytics you never open is a tax, not a tool.
$ Free plan is enough: unlimited messages on a rolling 90-day history. You will not pay for Slack to run this recipe. The PostHog Slack app installs from your project settings in a couple of clicks.
Metabase (optional, only if non-technical teammates need charts)The dashboard for people who will not write SQL. Point it at the same data, and a co-founder or an investor gets a clean chart without touching PostHog. Skip it entirely if you are a solo or all-technical team; PostHog's own dashboards already cover you.
$ Open-source edition is free to self-host under AGPL, but 'free' has a footnote: a small production setup runs roughly $100 to $200/mo in cloud infra plus your time to patch and upgrade it. Metabase Cloud Starter is $100/mo base plus $6/user/mo if you would rather not babysit a server.

🔌 The wiring

1
Drop the PostHog snippet in your app's <head> and turn on autocapture. That alone starts logging pageviews, clicks, and form submits with zero custom code, so you have data flowing before lunch.
2
Add three or four custom events for the actions that actually matter and will not change: sign-up, activation (your aha moment), key-feature-used, and purchase. Autocapture covers the noise; these cover the questions.
3
Call posthog.identify on login so anonymous sessions stitch to real users. Without this, retention and 'who did what' are guesswork, so do it early, not later.
4
Build the two charts every founder needs first: an activation funnel (signup to aha moment) and a retention curve. These two answer 'is the product working' faster than any vanity dashboard.
5
Link Stripe into the PostHog data warehouse so customers, charges, and subscriptions land next to your events. Now a single HogQL query can compare payer behavior to free-user behavior.
6
Open the SQL editor and write the real question your dashboard cannot: 30-day retention split by users who did the aha moment in week one versus those who did not. That one query usually rewrites your roadmap.
7
Create a PostHog subscription that posts your north-star insight to a Slack channel every Monday, and set one alert for when activation drops below your line. Analytics you have to remember to check is analytics you will stop checking.
8
Only if a non-technical teammate needs charts: stand up Metabase against the same data so they self-serve. If everyone touching the data can write SQL, skip this step and the bill with it.

✂ What to cut

Cut the four-tool analytics pile-up. The classic sprawl is an event tool (Mixpanel or Amplitude), a separate session-replay tool, a BI seat for charts, and a CDP like Segment to wire them together. You do not need any of those four as a separate product when one platform does events, replay, flags, and SQL together. Segment alone starts at $120/mo for 10K MTUs once you pass its 1,000-visitor evaluation tier, and it mostly exists to shuttle data between tools you would no longer have. Mixpanel's Growth plan runs about $2,520/mo at 10M events; Amplitude bills on monthly tracked users and overages hit at 1.2x. Also cut: the BI seat (PostHog's SQL editor covers early-stage questions for free), the dashboards nobody opens (one Slack subscription beats ten unloved charts), and the instinct to instrument every click before you have asked a single real question. Start with three questions, not three hundred events.

The receipts

What it costs

$0 to start, and likely $0 for a long time. PostHog's free tier (1M events, 5K replays, 1M warehouse rows) covers most early-stage products, and Slack is free. The only real spend is optional: Metabase Cloud Starter at $100/mo, or a self-hosted server at roughly $100 to $200/mo, and only if non-technical teammates need their own dashboards.

Time to wire

an afternoon

What it replaces

A separate event tool, a session-replay tool, a BI seat, and a CDP. In practice that is Mixpanel or Amplitude plus a replay tool plus a Metabase seat plus Segment, which lands anywhere from a few hundred to a few thousand dollars a month once Segment's $120/mo CDP and Mixpanel's per-event Growth pricing kick in at scale.

Where it breaks

It breaks at the edges, not the core. If you skip posthog.identify, your retention and user-level queries quietly lie, so verify that first. Autocapture drifts when you rename CSS classes or restructure the UI, so anchor your important events as custom events, not captured clicks. The Stripe sync needs webhooks to stay fresh, and note PostHog is reshaping its prebuilt revenue dashboard around mid-2026, so lean on the warehouse and your own SQL rather than one canned view. Maintenance is light: re-check events after a big UI change and keep your key queries saved.

Members · the full blueprint

The wiring, the configs, the templates, the Loom.

The free overview gives you the shape. The gated blueprint hands you the build: the exact PostHog event taxonomy for a SaaS product (the four events worth tracking and the properties to attach), a pack of copy-paste HogQL queries (activation funnel, week-one retention, payer-versus-free behavior, the Stripe revenue join), the Slack subscription and alert config so the one metric that matters finds you, an optional Metabase dashboard JSON pointed at the same data, and a 12-minute Loom wiring the whole thing end to end so you clone it without guessing a single field.

▶

How Sameer + Ankit wired it

Frame 1 of 8

Three dashboards, zero answers

Sameer had shipped, users were trickling in, and he had the analytics to prove it: a Mixpanel tab, a replay tool, and a Stripe export he opened on Fridays. The problem was simple. None of them agreed, and none of them answered the one question he kept asking. 'I have three dashboards,' he said, 'and I still don't know if the product is working.'

The reflex to add a fourth tool

His instinct was to glue it together with a CDP and call it a data stack. Ankit stopped him. 'You're about to pay a hundred and twenty a month for Segment so it can shuttle data between three tools you don't even trust. That's not a data stack. That's four bills for one bad answer.' He pulled up the real question instead: do the people who pay actually behave differently from the ones who don't?

One platform, one snippet

They put it all on PostHog's free tier. One snippet in the head tag, autocapture on, and events were flowing before the coffee went cold. 'Events, replays, feature flags, and a SQL editor, all in one login,' Ankit said. 'A million events a month, free. You were about to buy four tools to get what this gives you for nothing.' Sameer added four custom events that mattered and deleted two browser tabs on the spot.

The identify call nobody mentions

Ankit made one thing non-negotiable. 'Call identify on login. Skip it and your retention numbers are fiction, and you won't even know they're lying.' They wired it in, and suddenly anonymous sessions stitched into real users. The funnel and the retention curve went from a vague shape to a number Sameer could actually defend. 'That's the difference,' Ankit said. 'Real users, not hopeful guesses.'

The query that rewrote the roadmap

Then they linked Stripe into the PostHog warehouse and wrote the question Sameer had been chasing for months, in SQL: 30-day retention for users who hit the aha moment in week one, split by who paid. The answer was blunt. Week-one activators retained at triple the rate, and almost every payer came from that group. 'I was building features for people who already left,' Sameer said. 'This one query told me to fix the first week instead.'

The receipts

Wiring time: one afternoon. Monthly cost: zero, because the whole thing sat inside PostHog's free tier and Slack's. The Segment subscription never got signed, the second and third analytics tools got cancelled, and nobody bought a BI seat. 'It needs a tune-up after a big UI change,' Sameer admitted, 'and I keep an eye on the Stripe sync. Ten minutes now and then. Versus four bills and three dashboards I didn't trust? Easy trade.'

Steal next week's recipe.

One wired stack in your inbox every week. Free.

Get the blueprint More analytics recipes

A product analytics stack that answers real questions

Free · the verdict

Members · the full blueprint

The wiring, the configs, the templates, the Loom.

How Sameer + Ankit wired it

Three dashboards, zero answers

The reflex to add a fourth tool

One platform, one snippet

The identify call nobody mentions

The query that rewrote the roadmap

Make it impossible to ignore

The wiring, step by step

The full blueprint is for members.

The receipts

Steal next week's recipe.