If you’ve ever tried to wrangle your product’s data into something actually useful, you know half the battle is getting a dashboard that answers real questions. This guide is for anyone who wants to turn PostHog from “yet another analytics tool” into something that gives you real answers, fast—whether you’re in product, engineering, or just tired of asking for CSV exports.
We’ll cover exactly how to build custom dashboards in PostHog, what’s worth your time, and what’s just window dressing. No fluff. No hype. Just a step-by-step guide for folks who want to cut through the noise.
Step 1: Get Clear on What You Need (and Ignore the Rest)
Before you even log in, figure out what you actually want to see. Too many dashboards are cluttered graveyards of vanity metrics. Here’s how to keep yours useful:
- Start with questions, not charts. What decisions do you want to make with this dashboard? (e.g., “Is our new signup flow working?”)
- Pick 3–5 metrics max. More than that, and nobody will look at the thing. Trust me.
- Decide who’s using it. Is this for engineers, product managers, the whole team? It changes what you track.
Pro tip: If you can’t explain why a metric is there, cut it. You can always add it later.
Step 2: Set Up PostHog and Check Your Data
You can’t build a dashboard without data. (Well, you can, but it’ll be as useful as a chocolate teapot.)
- Sign up & log in. Pretty basic—make sure you’ve got access to your PostHog project.
- Confirm your events are coming in. Go to “Events” in the sidebar. If you don’t see fresh data, check your implementation. Don’t skip this. Half-built dashboards with missing data are a waste of everyone’s time.
- Check your properties. Are you capturing the right event properties (like plan type, signup source, etc.)? You don’t want to find out you’re missing something after you’ve built the dashboard.
Reality check: PostHog’s setup is usually straightforward, but custom events or self-hosted setups can throw curveballs. Don’t be afraid to bug your devs if you’re not seeing the right data. Better now than later.
Step 3: Create a New Dashboard
Now the fun part. Here’s how to start from scratch:
- Go to “Dashboards” in the sidebar.
- Click “New Dashboard.”
- Give it a name and description. Be specific (“Signup Funnel – June 2024” beats “Dashboard 2” every time).
- Choose privacy settings. You can keep it private, make it team-visible, or shareable via link. If you’re working with sensitive data, keep it private until you’re ready.
Don’t get sidetracked by templates unless one fits exactly what you need. Most folks end up customizing everything anyway.
Step 4: Add Your First Insight (Chart)
Dashboards in PostHog are just collections of “Insights”—basically, charts or tables with answers to specific questions.
How to add an insight:
- Click “Add Insight.”
- Pick your type: Trends (line or bar charts), Funnels, Retention, Stickiness, or Paths. If you’re not sure, start with “Trends.”
- Choose the event or property you want to visualize.
- Tweak the filters. Date range, breakdowns by property (like country, plan, etc.), cohorts—get specific.
- Preview the result. If it looks off, double-check your event names and filters.
- Add to dashboard. Once it looks right, hit “Save & Add to Dashboard.” Name it clearly.
Pro tip: Don’t obsess over chart colors or labels at first. Just get the data right—you can pretty it up later.
Step 5: Rinse and Repeat (But Don’t Overdo It)
Keep adding insights for each key metric or question you want answered. Remember:
- Less is more. If a chart isn’t helping you make a decision, skip it.
- Mix up chart types. Funnels are great for multi-step conversions. Trends work for daily/weekly/monthly activity. Retention shows if people stick around. Use what fits.
- Group related insights together. Drag and drop to keep things tidy.
What to ignore: Don’t waste time with “vanity metrics”—pageviews, raw signups, or anything that looks impressive but doesn’t guide action.
Step 6: Polish and Share (But Don’t Get Precious)
Once you’ve got your main charts set up:
- Give everything a clear title. “Signups by Day” is better than “Insight 3.”
- Add descriptions. Future-you (and your team) will thank you for quick context.
- Set up dashboard filters (optional). These let you filter every insight at once—handy if you want to look at just one country or plan.
Sharing
- Share with your team. Change the dashboard’s visibility or grab a share link.
- Schedule reports (optional). PostHog lets you email dashboards to yourself or others on a schedule. Useful, but don’t spam everyone daily—weekly or monthly is usually enough.
Reality check: Nobody cares about your dashboard unless it helps them do their job better. Get feedback and tweak as needed.
Step 7: Iterate (and Don’t Be Afraid to Delete)
The first version of your dashboard won’t be perfect. That’s normal.
- Ask for feedback. What’s missing? What’s confusing? What’s ignored?
- Cut what doesn’t get used. If nobody ever looks at a chart, delete it.
- Update as your product changes. Launch new features? Add new events and insights. Ship something nobody uses? Remove the clutter.
Don’t fall into “dashboard sprawl.” One or two solid dashboards beat ten half-baked ones.
What Works (and What Doesn’t)
What works:
- Keeping dashboards simple and focused.
- Using clear labels and descriptions.
- Building dashboards around real questions, not just “what data do we have?”
What doesn’t:
- Overloading with every possible metric “just in case.”
- Ignoring data quality issues.
- Expecting dashboards to answer questions you never asked (garbage in, garbage out).
Stuff to ignore:
- Fancy chart types you don’t understand (stick to basics).
- Widgets for the sake of widgets.
- Overcomplicating with endless filters and breakdowns unless you really need them.
Wrap-Up: Keep It Simple, Iterate Often
You don’t need a “perfect” dashboard—you need one that helps you make decisions, spot problems, and move fast. Start with a question, keep it simple, and iterate as you go. If something’s unclear, cut it or fix it. Your future self (and your team) will thank you for a dashboard that’s actually useful—not just pretty.
Now go build something that gives you real answers, not just more data. And remember: it’s always easier to add than to clean up later.