Project Homepage

[rcmarron]

Is your feature request related to a problem?

When a user opens a PostHog project, they're brought to the insights list. This page doesn't do a great job of introducing a user to PostHog or helping them leap into other parts of the product. We could do a much better job of using this real estate to quickly get users to what they're looking for and help them understand:

• Important metrics/insights for their project
• What PostHog is capable of
• How their coworkers are using PostHog
• What the project is

It feels like there's a clear opportunity to create a project homepage that addresses some of these concepts, and allows us to improve some of our core metrics (discoveries, number of organizations using multiple features, number of users viewing other user's insights)

Describe the solution you'd like

What should be on this project homepage? I imagine the project homepage is an evolving piece of real-estate. But a few ideas that seem like good options to explore are:

• Dashboard/Insights: Quick launching points to the most valuable information
  • Pinned dashboards/insights (I imagine a company could pin their core metrics right to the home page)
  • Personalized views with your favorite, recently viewed, or recently created insights/dashboards
  • Simple "universal" insights (WAU, % of users on browsers or devices, etc.)
• Lists of recent/incoming events, persons + recording: Make the project "feel alive" + expose the features
• Misc info: to help new users get their bearings
  • Project description
  • Who else is in the project
  • Info about PostHog (new releases, features etc.)
• Search across dashboards/insights and maybe other objects
• "Social" feeds: Show all the ways your coworkers are using the product [Seems a bit further out - post audit work]
  • What are your coworkers doing in PostHog?

Questions

• What else could live on the project home page?
• What could the day-0, day-1 and day-100 experience look like?
• Is there a pattern we could develop that would allow us to have a project homepage that could flexibly meet the needs of the various uses above?

cc @clarkus @paolodamico @Twixes

Thank you for your feature request – we love each and every one!

[paolodamico]

Great context and agree on the huge potential value of this. As we tackle a V1 for this, we should focus on a single problem. Based on recent conversations, I suggest focusing on enabling teammates to answer questions they have (search) and discover potentially useful insights (interesting).

This would mean excluding for now: learning to use PostHog, additional project information, social feeds, aliveness/activity, … This also means the experience of the home page today can be the same on Day 0 and on Day 100. Thoughts?

[timgl]

We've done this sort of thing before, here's some context: https://github.com/PostHog/product-internal/issues/50#issuecomment-944470182

[rcmarron]

We've done this sort of thing before, here's some context: https://github.com/PostHog/product-internal/issues/50#issuecomment-944470182

Got it, thanks for sharing. Reading that, it sounds like the experiment was a success - drove up discoveries. Curious if there's a reason we didn't pursue it further outside of limited resources?

As we tackle a V1 for this, we should focus on a single problem. Based on recent conversations, I suggest focusing on enabling teammates to answer questions they have (search) and discover potentially useful insights (interesting)

• I think search is interesting. I want to raise the question: does search belong on the project home or in the navbar? Big advantage of the navbar is that it's accesible anywhere. And search seems like a weird thing to have to navigate to find.
• Agreed that insights are going to be a huge part of this. I think it's important to think about how they're related to dashboards too.
• I could see pulling in some other misc info to "round out" the page too. But would definitely defer to design on that.

This also means the experience of the home page today can be the same on Day 0 and on Day 100

• I think this is up for discussion. For example, if we show 'pinned insights' or 'recently viewed' what does that look like when the user first shows up?

[clarkus]

I put together an updated concept to help facilitate discussion. The idea is to promote fresh insights and recordings. Pinned dashboards are shown in a secondary placement. If we think dashboards will be created or changing frequently, we could also include a list of recent dashboard updates. Other lists to consider are users (or other actors) seen for the first time, recordings of a significant duration, new incoming events, etc.

These are all just ideas, so feel free to leave feedback on anything you see.

[rcmarron]

This looks great @clarkus! Thanks for pulling it together! I think it really helps make the concept feel more concrete.

Seems like the next big decision is what exactly should be on the home page. I imagine this is a constantly evolving page, so we don't have to nail it the first time, but we should try our best 😀.

So, what should go on the homepage?

The overarching goal of the homepage is to try to get users to valuable content as quickly as possible. It's essentially a "launch pad" for the product.

There seems to be 2 distinct types of users to consider here:

• 🔎 The searcher: A user is coming to PostHog for a specific thing, and they want to quickly find it. This user is probably best aided by things like:
  • Quick access to the most common actions in PostHog (mainly: make a new insight)
  • Their most recently viewed insights/dashboards
  • A search bar that works across objects in PostHog (insights, FFs, experiments, dashboards)
  • Their favorited stuff
• 🧭 The explorer: A user is coming to PostHog to "understand their product", but doesn't know what exactly they're looking for. This user is probably best aided by things like:
  • The insights/dashboards that other users are recently creating, editing etc. (Note: this is probably most useful when it's the actions of specific trusted colleague, but generic is also interesting)
  • Pinned dashboards/insights for the project - what have others explicitly decided is valuable
  • Teasers into more exploratory features (recordings, persons, events)
  • Quick look at generic metrics (WAU, retention etc.)
  • Feed of other's actions in the project

I imagine a great V1 homepage is one that mainly spans these two "launchpad" use cases, and maybe addresses a couple secondary goals: explain what the project is + share what PostHog is capable of

With these things in mind, I think the following is a good starting point:

V1 Homepage proposal:

• 🔎 New insight button
• 🔎 List of the user's recently viewed content (insights, recordings etc.)
• 🧭 The most viewed insights this week and which colleagues are viewing them.
• 🧭 Recent recordings (maybe show recent events if recordings is disabled?)
• Project description + collaborators list
• Invite collaborator
• 🧭 Pinned dashboards for the project (update from @clarkus's comment below)

Note: each of these items above needs a layer of deeper thought, but let's agree on general direction first.

Notable omissions for V1:

• 🔎 Search: likely that the navbar is the right spot for this - not the homepage (also, it's probably hard...)
• 🔎 Favorites: very similar content to "your recently viewed", but it requires a user to do something...
• 🧭 Social feed: seems very valuable for the explorer, but need the data first (parallel work in progress).
• 🧭 Generic metrics: questionable "real value" on this one when compared to "pinned metrics" or things like that.
• 🧭 Pinned insights: we already have pinned dashboards, let's see how that goes.

Big questions on my mind:

• What else are we missing above?
• How do we make this feel not "dormant" if people aren't using the project?

Would love thoughts on this: @clarkus @paolodamico @mariusandra @alexkim205 @Twixes @pauldambra

[clarkus]

I think your proposal is on target. Are there any other page-level actions we should consider other than creating an insight? Inviting another collaborator?

The more I think about it the more I don't think pinned dashboards are good fit here. We already have them prominently placed in the navigation. I'd rather try something else that's new and either time-based or views-based to promote the content that is being viewed by others. Thoughts?

How do we make this feel not "dormant" if people aren't using the project?

I actually think this is a good thing. It's transparent, unbiased, and in the current world of algorithms, dare I say refreshing? A lack of activity is a sign that maybe users aren't doing much directly in the product. I think that could be a valid use case for projects with low collaborator counts.

One thing to track with our planning is accounting for projects that don't have collaboration enabled. I think it would be a pretty harsh upsell if we maintained this as the landing page for those users. We just need to think through what their experience should be and identify places we can upsell collaboration.

[rcmarron]

Inviting another collaborator

This makes a ton of sense to me. Especially if we're showing the collaborator list.

The more I think about it the more I don't think pinned dashboards are good fit here.

Sold. I'm all for trying new things on the home page, and I agree that pinned dashboards already have a prominent location. We could always add it down the road if needed.

Also, the more I think about it, the more I like the idea of the homepage showing insights over dashboards. A dashboard is just a container of insights, and so it's weird for the homepage to be a container of containers of insights.

One thing to track with our planning is accounting for projects that don't have collaboration enabled

Agreed, once we have the list of items here, we should dig into which are behind a pay wall and which aren't.

(p.s. updating the list above to reflect these thoughts)

[mariusandra]

This looks good, though I'd propose a tactical change to the plan and split out v0.1 from v1.

These two points:

• 🔎 List of the user's recently viewed content (insights, recordings etc.)
• 🧭 The most viewed insights this week and which colleagues are viewing them.

... require building something that doesn't exist yet. Since building these features is exploratory in nature ("how to store the data?"), it can take anything from 1 to 10 days.

Hence I suggest splitting out a v0.1 that'll be built without significant changes to our data model. Get that in front of real users ASAP to hear what they would actually benefit from. It might turn out that there are even better things to do and we only think we know what v1 should look like. Or not, but at least we'll be certain we're spending those 10 days wisely.

[marcushyett-ph]

Inviting another collaborator

I think this is key - from what I've observed a lot of people are using shared accounts, so they won't get the benefits of collaboration. Getting more team members in, is just as important as showing them interesting insights. Also larger teams correlate with a more downstream success.

This looks good, though I'd propose a tactical change to the plan and split out v0.1 from v1.

I agree strongly with @mariusandra point about keeping the MVP scope tight so we can get it out fast and learn faster from our users, but my gut feel is that a feature like "The most viewed insights this week and which colleagues are viewing them" is probably more valuable to collaboration than many of the other features put together - so would advocate for tackling this sooner rather than leaving it a lot later.

Tactical question:

• Are we going to put this behind an experiment - so we can measure incremental impact on the metric(s) we care about?

[marcushyett-ph]

One thing to track with our planning is accounting for projects that don't have collaboration enabled

Personally, I think these proposed collaboration features should be available to everyone (future ones perhaps not), since organizations are significantly more likely to subscribe (ref paid up section here) if they add someone else to their organization and probably even more so if that person instantly gets some value from the product (through these features).

@timgl what are your thoughts, since this goes against our start paid, make it free later approach?

[neilkakkar]

Some user context:

As a searcher in a single-person project for a website, I still open google analytics more often than PostHog, simply because it opens up on the page I really care about: trend of users over time + people currently on site. It takes me 3 seconds to open-see-close, assuming numbers are not waay high or low.

And the main point:

The overarching goal of the homepage is to try to get users to valuable content as quickly as possible

Judging by the conversation and things we're choosing to build, this doesn't sound like the overarching goal to me? (For example, how does "invite collaborator" fit in here?).

I might be being too pedantic here, but I think when building something like this, the goal is to learn as much as you can (which informs future iterations). Doing that is a lot more valuable when you have a solid specific hypothesis about what users want, and find specific ways in which it's wrong.

For example, it feels like there's not one, but three mostly-independent goals for a homepage:

1. See & use valuable content (insights / recordings / new insight etc.)
2. Drive collaboration + retention for PostHog usage ( a goal for us, much more than users? Is this also what users want? idk yet). Invite teammates fits in here, imo
3. Setting up new users for success (what this project is about, who's in it, new insight button(?), and here are some cool onboarding things you can do (the last 2 things have some overlap with (1) )

And once we have a better idea of goals, we can slice v0.1 by focusing on specific goals, and testing for those goals (while also gathering feedback for other goals), vs. machine-gun-spray-fire everything.

I think (1) and (3) are wildly different users (the veteran vs newbie split), and in most flexible products, necessarily have different ways of approaching things (existing set patterns built up over time, vs, new person learning new concepts to traverse PostHog), and keeping this in mind during design / implementation / the kind of users you get feedback from - can help us get higher quality feedback.

[rcmarron]

I'm all for shipping a quick version ASAP, and I think the changes @mariusandra proposed make sense. I also agree with @marcushyett-ph that the "Most viewed insights" features seems very valuable.

With that in mind, I think we try to pursue both in parallel (getting multiple people on the team involved).

So we would work on:

1. Basic homepage: A quickly shippable project homepage for us to iterate from

• 🔎 New insight button
• 🧭 Recently edited insights (to be replaced once recently viewed insights is ready)
• 🧭 Recent recordings (maybe show recent events if recordings is disabled?)
• Project description + collaborators list
• Invite collaborator

2. Recently viewed insights: Track and display recently viewed insights

• 🔎 List of the user's recently viewed insights
• 🧭 The most viewed insights this week and which colleagues are viewing them. (Replacing recently edited insights)

@clarkus From a design perspective, I think we work toward the version that encompasses all of this, and we can just pair that down for the basic version. Does that make sense?

[posthog-contributions-bot[bot]]

This issue has 2267 words at 12 comments. Issues this long are hard to read or contribute to, and tend to take very long to reach a conclusion. Instead, why not:

1. Write some code and submit a pull request! Code wins arguments
2. Have a sync meeting to reach a conclusion
3. Create a Request for Comments and submit a PR with it to the meta repo or product internal repo

Is this issue intended to be sprawling? Consider adding label epic or sprint to indicate this.

[rcmarron]

As a searcher in a single-person project for a website, I still open google analytics more often than PostHog, simply because it opens up on the page I really care about: trend of users over time + people currently on site. It takes me 3 seconds to open-see-close, assuming numbers are not waay high or low.

I think this is a really good point. They do a great job of getting users to this valuable content as soon as possible (by having the graph right there immediately). I think we could potentially do something similar - maybe we add a concept of pinning an insight to the homepage? And by default, DAU, page views etc are pinned?

I want to make sure we treat the homepage as a flexible place for us to run experiments and try things out, and seems like an obvious thing that we should try (maybe right away?)

For example, it feels like there's not one, but three mostly-independent goals for a homepage

I think you’re right. Homepages are super valuable real estate, and they inevitably end up serving multiple purposes - which makes them tricky.

With that said, I think having 3 independent goals here isn’t great for driving immediate clarity. And I think your first goal is the most important for the homepage, so as much as possible, I want to try to keep us focused there for now.

I’m almost certain that when we want to tackle “Setting up new users for success”, the homepage will play a role in that. But let’s tackle that down the road. Maybe it’s V2.

[clarkus]

I'll be working on MVP updates at https://www.figma.com/file/gQBj9YnNgD8YW4nBwCVLZf/PostHog-App?node-id=7666%3A41216. Nothing there yet, but should have something ready by the end of the week.

[kpthatsme]

Skimmed the thread briefly - as Tim mentioned we ran this before.

To answer your question @rcmarron - no, resources were the only constraint.

I have lots of ideas on what might be effective here let me know if you wanna discuss. In our previous experiment we ran tutorials, tooltipped insights, and org insights for different personas (mostly covered in the write up Tim linked to).

[clarkus]

I have lots of ideas on what might be effective here let me know if you wanna discuss. In our previous experiment we ran tutorials, tooltipped insights, and org insights for different personas (mostly covered in the write up Tim linked to).

Sign me up for a chat. This is very much just an attempt to see how users respond to this kind of data. I think there's a ton of room to make it more valuable longer term, so I'd love to hear any ideas you have.

[rcmarron]

Same! Just set up some time for us for tomorrow.

[rcmarron]

Capturing some off-line conversations that @clarkus and I have been having:

We had concerns about the proposed scope above not really nailing the goal of "get users to valuable content as quickly as possible".

After exploring what other products do for their homepage, it seems like there's pretty strong consensus around 2 things:

1. A customizable project dashboard
2. Lists of commonly accessed things (e.g. your recently viewed insights)

These items seem much more aligned with getting users to valuable content as quickly as possible. The dashboard allows users to decide what the most valuable content is and have it immediately available. And the lists are our best guess of what you're looking for (e.g. the last insight you looked at).

With this in mind, we're going update our approach here to:

V1:

• Default dashboard for the project that sits on the homepage

V1.1:

• Add lists of commonly accessed things (TBD exactly what this is, but things like recently viewed insights, new persons, new recordings, etc.)

A couple questions that come with this decision are:

• Do we want to revisit our current default insights/dashboards? Do we like them? They're going to be much more prominent
• How do we want to handle "discovery metrics"? Putting insights on the homepage is going to skew it.
• Permissions. Who should have access to update this "homepage dashboard"?

@marcushyett-ph does this sound good to you?

[marcushyett-ph]

Firstly, we should come up with a way to measure our progress against this goal: "Get users to valuable content as quickly as possible", so we can understand if we're iterating in the right direction.

Personally I think Discoveries is a good proxy for "finding valuable content" and running an A/B test would be a good way to measure it (I would advocate not to overcomplicate by adding something specific to account for "as quickly as possible"). Thoughts?

V1:

• This makes sense, since it sounds fairly straightforward I would be keen to get this out fast so we can to learn and iterate

V1.1:

• We should prioritize these based on usage data we have today, my proposed priority would be:
• Popular Insights - Today it's very clear looking at the most popular insights for PostHog that these are the ones people really care about - (e.g. company useage report, total paid discoveries), it's also clear that our customers are using the product in the same way with some insights very popular that most users would want to review frequently - My rubric for popularity is (num_insights_analyzed_events*num_unqiue_users_analyzing_it), which seems to return fairly good results
• New Recordings - Users who use recordings analyze them very frequently a new recordings insight will enable people to get straight to one of their most important jobs for PostHog. Most users start on the recordings page today, indicating that a good initial version of this could be the default recordings list - we should iterate on this with feedback.

I would not provide any other shortcuts for now - new persons or events feels like for any non-micro user it would become meaningless and overwhelming very quickly.

Answers to open questions:

• Do we want to revisit our current default insights/dashboards? Do we like them? They're going to be much more prominent
  • I think the current default dashboard is okay, I'd be keen to propose a few small tweaks (based on looking at the most popular types of insights on existing dashboards) - I can take this and come up with a proposal this week. I don't think this should be a blocker for v1 though.
• How do we want to handle "discovery metrics"? Putting insights on the homepage is going to skew it.
  • I'm not sure I understand this question, if people are discovering more (valuable) insights, this is surely a good result (if they're actually not useful, people will surely click away and not count as a discovery)
• Permissions. Who should have access to update this "homepage dashboard"?
  • I would not do anything special in terms of permissions for this dashboard, given the new permissions framework we've shipped should allow for locking down of specific dashboards (and generally keeping things more open will encourage people to collaborate and improve the home dash between themselves)

More questions:

• How do we decide the default dashboard for an existing user?
  • If they have only one pinned dashboard, it makes sense to use this as the home dashboard by default as the user has expressed an intent that this is most important
  • For us, we have 7 pinned dashboards we'd need a user to pick the primary one in this case, we need to make it clear and easy to switch the default dashboard if we automatically pick it wrong (I'm sure @clarkus will have opinions on this experience, but I'd advocate for showing a dashboard initially over a list of possible dashboards to give them a better feel for the purpose of this homepage dashboard)
• What do we do if they don't have any dashboards?
  • I think this is a very small edge case, but we should handle it by getting them to create a dashboard which automatically becomes the home page dashboard

[clarkus]

+1 on discoveries as a driving metric.

[marcushyett-ph]

@rcmarron @clarkus context on having a default dashboard for a project:

The majority of pinned dashboards used in the last month are primary dashboards of some kind (inferred from their name) for example:

• (System Default) My App Dashboard
• Default
• Dashboard
• Overview
• Web App Dashboard
• Main Dashboard
• Website Dashboard
• App Analytics

This strongly supports up the approach of a single primary dashboard for the homepage.

[marcushyett-ph]

Default dashboard Proposal

I'm proposing some small tweaks on the default dashboard we create. Like I said above this shouldn't be a blocker for v1 and feel free to make the final call on this @rcmarron.

Current default dashboard:

• Users by traffic source
• Cumulative DAUs (N.B. I think this name might be slightly paradoxical)
• Users by Browser (last 2 weeks)
• Sample - Purchase conversion funnel
• Repeat users of time
• Daily Active Users

Proposed default dashboard:

• Daily Active Users: Same as now
• Weekly Active Users: Straightforward
• Growth Accounting: Weekly Lifecycle on $pageview
• Acquisition Source: Unique Users by Traffic Source (value breakdown chart)
• Activation: $Pageview to $Autocapture funnel (not ideal - but shouldn't be blank)
• Retention: Weekly Unique user Retention graph on $Pageview

Source of proposed changes:

• Most popular dashboard items which have been modified by users
• Industry guides
• Standard metrics frameworks
• Whats actually possible with initial customer data (today)

[clarkus]

Related - https://github.com/PostHog/posthog/issues/5209