PostHog Dashboard: Six Tiles Worth Having, and Where Built-in Insights Fall Short

PostHog makes it easy to add insights to a dashboard. It does not decide which ones matter, or stop you adding up numbers that cannot be added.

A useful PostHog dashboard has six tiles: sessions and page views, traffic by channel, signups and signup rate, activation within a fixed window, retention by signup week, and the pages that bring visitors who sign up. The common failures are not visual. They are arithmetic: adding unique people across periods, dividing by cohorts that are too young, and defining signup differently in each tile. This post gives the six tiles, the rules that keep them honest, and where asking questions of the same data in Quiriz helps beyond built-in insights.

The six tiles

1. Sessions and page views, week over week. A health check. Sessions should match the figure on PostHog's own web analytics page, which makes it the trust anchor for everything else.
2. Sessions and signups by channel. Channel is the dimension that explains most movement. Show volume and rate together.
3. Signups and signup rate. Signup events divided by sessions, by day or week.
4. Activation within 7 days, by signup week. The share of each cohort that did the activation event inside a week of signing up.
5. Retention at week 1 and week 4, by signup week. The share of each cohort active in those weeks.
6. Entry pages that lead to signup. Pages with their entry sessions and entry signups, so you can see which landing paths convert.

Six is a ceiling for a team dashboard, not a floor. Someone who needs a dozen tiles usually has several audiences and needs several dashboards. If the people who read it are the marketing team, read the companion Google Analytics report guide as well, since the question about channels is shared.

Four rules that keep the tiles honest

One definition per event. If the signup tile counts account creation and the activation tile starts from email confirmation, the rates will not chain. Choose the signup, activation and paid events once and use them everywhere.

Never add unique counts. Weekly active people for four weeks do not add up to the monthly active figure, because a returning person counts in each week. Show each period as itself. If you need a quarterly figure, compute it as its own distinct count.

Judge cohorts only when they are old enough. A week 4 retention value needs four full weeks after the signup week. Cohorts younger than that have blanks, and the correct handling is to exclude them from the rate, not to treat blank as zero.

Compute rates from totals. Signup rate for a period is total signups over total sessions, not the average of daily rates. An average of rates overweights quiet days.

Where built-in insights stop being enough

PostHog's dashboards are good at what they were built for: exploring behavior and keeping a set of saved insights in view. Three cases are less comfortable.

• A question nobody saved. Someone asks which referring domains brought signups during last Tuesday's launch. Answering means building an insight, or asking the person who knows how.
• Definitions that live in individual insights. When an insight's filter changes, the number changes quietly, and the dashboard still looks the same. A shared definition is easier to audit than a dozen filters.
• Questions across tools. Product data says who signed up; the payment system says who paid. A dashboard in one tool cannot answer a question about both.

Asking the same data in Quiriz

Quiriz is an AI business analyst. The PostHog connector is in beta, read-only, refreshes daily, and is included in paid plans. It reads aggregated data for the last 13 months, with the signup, activation and paid events taken from the Events panel in your connector settings. The datasets are [PostHog] Traffic by day, Pages by month, Audience by month, Active users, Signup cohorts, Events by day and Event users by month, and the metrics are defined once, including PostHog signup rate, 7 day activation rate, week 1 retention and week 4 retention. The cohort age rule is built into those metrics.

Each tile is a question:

• How many sessions and page views did we get this week compared with last week?
• Which channel brings the most signups, and which channel converts best?
• What percent of last month's signups activated within 7 days?
• What is our week 4 retention for each signup week?
• Which pages bring visitors who sign up?

The answer shows the table, the dataset and the filters, so a reader can tell whether a number is the same one they saw in PostHog. For anything the dashboard does not already have, such as signups by country and device from the Audience dataset, or the launch-day breakdown by referring domain, you ask rather than build. Once the answer matters, an admin can pin the definition in Company Context. The PostHog AI post goes through the event setup in more detail.

Limits

• The connector reads aggregates only. It does not replace PostHog's funnels, recordings or experiments, and it copies no person-level data.
• Referring domains and UTM values keep the top 100 per month; entry pages keep the top 5,000 per month by page views.
• Active users are not additive and are not exposed as a summable metric. Read each period as its own row.
• On cookieless projects, user and cohort counts are approximate.
• Signups read zero until a signup event is chosen.
• Quiriz does not draw PostHog's own interactive insights. It gives you tables and charts from the same underlying data.

Frequently asked questions