# RGSoC 2016: Visual Forum Analytics Community Discussion

**URL:** https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905
**Category:** Data & reporting
**Created:** [July 6, 2016, 3:50pm UTC](https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905 "2016-07-06T15:50:32Z")
**Posts on this page:** 13
**Page:** 1

<div class="post-metadata">

### Author: ![saintsebastian](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saintsebastian/32/59341_2.png) [@saintsebastian](https://meta.discourse.org/u/saintsebastian)
#### Post date: [July 6, 2016, 3:50pm UTC](https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905/1 "2016-07-06T15:50:33Z")

</div>

Hey everyone!

I am one of two people who is planing to provide you with informative graphs and charts on user, post etc statistics. We are doing this as our [project](https://teams.railsgirlssummerofcode.org/projects/96-discourse-visual-forum-analytics) for [Rails Girls Summer of Code](http://railsgirlssummerofcode.org) and this is our introduction to both open source community and Discourse.

As of now we are familiarising ourselves with Ember, because it is fairly new to us and overall pretty powerful. Boy, Ember is magical indeed!

We’ve already read some threads concerning people’s burning desire for good looking stats (and who can blame them, datavis is the best), but we are open to ideas about how stuff should work and look. Anyways, it is nice meeting you, dev. Hope to have some great time together.

Best,  
team kindr3d

---

<div class="post-metadata">

### Author: ![eviltrout](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/eviltrout/32/5275_2.png) [@eviltrout](https://meta.discourse.org/u/eviltrout)
#### Post date: [July 8, 2016, 8:16pm UTC](https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905/2 "2016-07-08T20:16:43Z")

</div>

Hi Team!

Thanks for posting this! I’m hoping that people in the community can jump in and offer help on various topics.

Visual Forum Analytics is something that a lot of people have wanted for a long time. I’m curious for those in the community what kinds of reports would you like to see? Are there any types of charts you think would be useful?

It’s obviously going to take the team a little while to spin up but I figure now is the best time to ask for ideas 🙂

---

<div class="post-metadata">

### Author: ![Falco](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/falco/32/179432_2.png) [@Falco](https://meta.discourse.org/u/Falco)
#### Post date: [July 8, 2016, 8:30pm UTC](https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905/3 "2016-07-08T20:30:30Z")

</div>

I was waiting for you to ask 😉

We are a big company (100K+ employees) and our categories are big parts of the company.

What people are asking is some visual digest per category. They **really love** the new user summary page, and would like something like that per category.

On my side, I would like that info to be public, since we can’t handle moderation power to the people who handle categories.

---

<div class="post-metadata">

### Author: ![erlend\_sh](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/erlend_sh/32/119475_2.png) [@erlend\_sh](https://meta.discourse.org/u/erlend_sh)
#### Post date: [July 8, 2016, 8:56pm UTC](https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905/4 "2016-07-08T20:56:48Z")

</div>

Yay for visualised stats! 🌈

Some parts of [this old request might](https://meta.discourse.org/t/visual-forum-statistics/34744/9) be defunkt now that we’ve got a a first take on [visualised stats](https://meta.discourse.org/t/visualised-traffic-statistics/44298), but I think many features mentioned there could be a major improvement to what we already have.

Also every metric mentioned here would be great to have visualised:

> [@Plugin: Admin statistics digest](https://meta.discourse.org/t/plugin-admin-statistics-digest/44270):
>
> I’ve never been good at tracking statistics. I’m not disciplined enough to observe patterns and try to discern ongoing trends. Give me a good infographic though (aka Stats for Dummies) and I’ll eat that stuff up like it’s taco Tuesday. Imagine every month, you receive an email from Discourse that tells you: You have x more/less visitors than last month. (show overlapping graphs). You have x more/less topics than last month (show overlapping graphs) X topics posted in the last 30 days…

I just haven’t given much thought yet to the visualisation aspect of these stats. You’re very welcome to drop by that topic and suggest some graph types that you’d think work well for these assorted metrics.

Lastly, does the scope of your project cover adding a new top level /admin/stats page to the backend?

 ![](https://global.discourse-cdn.com/meta/original/3X/2/0/204220687e081686475419bc98a2b135f24205ad.png)

If so then we should eventually have a discussion about the overall design of this “stats index page” as well.

* * *

> [@Falco](#):
>
> On my side, I would like that info to be public, since we can’t handle moderation power to the people who handle categories.

That’s a fair request, and do correct me if I’m wrong @eviltrout, but I think _public_ visualisation of stats is outside the scope of this project. At least, I for one would prefer it if the focus was on figuring out as many cool visualisations of Discourse data as possible. Get the visualisations working and in front of admins, and we can worry about the “how” and “what” of publicly available statistics later.

---

<div class="post-metadata">

### Author: ![Mittineague](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/mittineague/32/114259_2.png) [@Mittineague](https://meta.discourse.org/u/Mittineague)
#### Post date: [July 8, 2016, 9:01pm UTC](https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905/5 "2016-07-08T21:01:30Z")

</div>

I don’t know, but I believe what you’re calling Statistics I’ve called Reports.

If so, IMHO it has long deserved it’s own Admin page.

---

<div class="post-metadata">

### Author: ![erlend\_sh](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/erlend_sh/32/119475_2.png) [@erlend\_sh](https://meta.discourse.org/u/erlend_sh)
#### Post date: [July 8, 2016, 9:22pm UTC](https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905/6 "2016-07-08T21:22:51Z")

</div>

> [@Mittineague](#):
>
> I don’t know, but I believe what you’re calling Statistics I’ve called Reports.

Are we on the same page?

> [@How would you design a "Statistics" index page for the backend?](https://meta.discourse.org/t/how-would-you-design-a-statistics-index-page-for-the-backend/47016):
>
> This is not a sure thing yet, but I sure would love it if Discourse had a dedicated /admin/statistics page on the backend, rich with interesting community insights, visualised for my viewing pleasure. Maybe this will come more naturally once we’ve nailed down [the stats we’ll be focusing on for the RGSoC 2016](https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905), but if someone has some ideas for this already I’d love to hear’em. Here are some statistics pages from other web apps for inspiration:

We can continue the “statistics index page” discussion there.

---

<div class="post-metadata">

### Author: ![Mittineague](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/mittineague/32/114259_2.png) [@Mittineague](https://meta.discourse.org/u/Mittineague)
#### Post date: [July 8, 2016, 10:56pm UTC](https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905/7 "2016-07-08T22:56:00Z")

</div>

Very similar but not exactly what I had in mind.

That looks like variations of a “one stop page full of stats widgets”

I was thinking more like a Reports “index” page instead of Reports being only reachable via the Dashboard links or known URL(s).

Admittedly, I’d have a hard time designing my way out of a wet framework, but I envision a much more “plain” pragmatic type of dedicated “menu” page. (what most would call boring).

Now thinking about it some configurable page of favorite stat “widgets” would be nice to have and it would surely look better than anything I could design. As I said, I think “features” not “looks”

---

<div class="post-metadata">

### Author: ![eviltrout](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/eviltrout/32/5275_2.png) [@eviltrout](https://meta.discourse.org/u/eviltrout)
#### Post date: [July 11, 2016, 2:41pm UTC](https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905/8 "2016-07-11T14:41:59Z")

</div>

> [@erlend\_sh](#):
>
> That’s a fair request, and do correct me if I’m wrong @eviltrout, but I think public visualisation of stats is outside the scope of this project.

Yes, I think that’s a fair assessment. One thing to keep in mind is the team is learning _A LOT_ over this process, and so it’s important to start small and build on it from there.

They are going to start with admin reports, adding more as time allows. But if they find that task too easy and have made good progress, maybe they will want to take a look at public reports?

It’s definitely not a priority though 🙂

> [@erlend\_sh](#):
>
> what you’re calling Statistics I’ve called Reports.

Honestly the distinction is maybe not as rigid as it seems in this discussion. The project is about taking data from our database and displaying it in a beautiful and easy to understand way.

Some will be extensions of what we already have (reports), and I hope the team will start with this as it’s a good way to learn, but I would not be against having the team generate some smaller “widget” like reports if the visualization makes sense.

---

<div class="post-metadata">

### Author: ![mcwumbly](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/mcwumbly/32/103861_2.png) [@mcwumbly](https://meta.discourse.org/u/mcwumbly)
#### Post date: [July 11, 2016, 2:51pm UTC](https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905/9 "2016-07-11T14:51:44Z")

</div>

> [@eviltrout](#):
>
> One thing to keep in mind is the team is learning A LOT over this process, and so it’s important to start small and build on it from there.

When deciding which things to add first, I think it’d be healthy to frame the question in these terms:

> [@How to quantify the success of your community?](https://meta.discourse.org/t/how-to-quantify-the-success-of-your-community/40229):
>
> Following a discussion about books on community building I saw this tip and read part of the book. This book seems to put a strong emphasis on data and its analyses as a means of quantifying the success of your community. Quote from the book: The number of unseen problems is HUGE! When the number of new visitors to the platform dips, that’s an unseen problem. When a number of regulars vanish, that’s an unseen problem. When the volunteers begin slipping away, that’s an unseen problem. When t…

Along those lines, one feature to consider adding is a tooltip or explanation that attempts to give some guidance about what a given metric might mean under what scenarios it might be expected to be misleading.

There may be some hints about what kind of data is “missing” in this topic too:

> [@(Superseded) What cool data explorer queries have you come up with?](https://meta.discourse.org/t/what-cool-data-explorer-queries-have-you-come-up-with/43516):
>
> This topic became too long to be easily searched and has been broken down into separate topics. If you have any questions about how to write a [Data Explorer](https://meta.discourse.org/t/32566?silent=true) query, or have one to share, please start a new topic in the #Data & reporting category and tag it with the #sql-query tag. +1information_source For an alternative searchable list of queries that were in this topic, see @SidV’s [query list on GitHub](https://github.com/SidVal/discourse-data-explorer/blob/queries/querys.md).

---

<div class="post-metadata">

### Author: ![erlend\_sh](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/erlend_sh/32/119475_2.png) [@erlend\_sh](https://meta.discourse.org/u/erlend_sh)
#### Post date: [August 16, 2016, 12:03pm UTC](https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905/10 "2016-08-16T12:03:55Z")

</div>

In thinking about our work on calibrating [likes notifications](https://meta.discourse.org/t/notify-on-every-like-should-be-the-default/40485), I’d love to have a nice visual statistic on the frequency of different like counts on my forum.

You could adjust the time period, the default being “Likes this month”, and the stat would be:

- 1 like: 8 posts
- 2 likes: 4 posts
- 3 likes: 24 posts
- 5 likes: 3 posts
- 6 likes: 0 posts
- 7 likes: 1 post

and so forth. This data would be super helpful when calibrating your community badges.

---

<div class="post-metadata">

### Author: ![saintsebastian](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saintsebastian/32/59341_2.png) [@saintsebastian](https://meta.discourse.org/u/saintsebastian)
#### Post date: [August 18, 2016, 3:34pm UTC](https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905/11 "2016-08-18T15:34:01Z")

</div>

Hey everyone!

Here are a couple of sketched ideas for the admin dashboard.

This is just our first version, so things will probably change. This is what we think is important for quick overview of stats. Of course we aren’t admins ourselves, so we wanted to share it with the community (with hopes for valuable feedback).

The dashboard consists of 4 components, the admin can switch between three time units: today (last 24 hours), this week and this month. For each component we provide an overall count of new users, new topics etc with comparison (arrow and per cent) with the last time unit and some additional component-specific info.

We are treating those things as modules so the community can improve them later on if the need arises or maybe even build new ones if this info proves to be irrelevant.

We aim for simplicity and clarity and don’t want to over clutter the dashboard. The current dashboard will still continue to exist for more detailed stats.

 ![](https://global.discourse-cdn.com/meta/original/3X/0/0/00dd2387a4a3ee2f1a04d65c002d34a3e3f4f0e8.jpg)

 ![](https://global.discourse-cdn.com/meta/original/3X/0/6/06261b5cd57d2015748b0f3fe097780f41412300.JPG)

---

<div class="post-metadata">

### Author: ![erlend\_sh](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/erlend_sh/32/119475_2.png) [@erlend\_sh](https://meta.discourse.org/u/erlend_sh)
#### Post date: [August 18, 2016, 4:20pm UTC](https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905/12 "2016-08-18T16:20:05Z")

</div>

Neat! Some feedback:

- Could we add a little ❔ tooltip icon or something next to the Today button? This would tell you what timezone “today” is based in. Especially useful for the Visits stat, since my Norwegian “night” might not be the same as [meta.discourse.org](http://meta.discourse.org)’s.

- Visited vs visited&posted is a cool stat to have available at a glance; looking forward to using it! But what does “not visited” mean?

- The pie chart in the Topics widget is also a bit confusing to me. Maybe it’s just the mix of black/white and actual colour coding that’s throwing me off 😛 Are those 3 sections in the pie chart connected to the top categories, or are those two separate datasets?

* * *

I’ve got tons of different stats I’d like to see, but if I could only pick one, it’d probably have to be this one:

> [@How to quantify the success of your community?](https://meta.discourse.org/t/how-to-quantify-the-success-of-your-community/40229/4):
>
> The one thing that I find to be super important is a regular influx of new users who stick around. Without new blood, your site is going to die eventually. That’s an absolute truth. This is why we have a todo item to send automatic encouragement emails to new users who sign up, post, get likes, and otherwise show good rookie potential.

@DeanMarkTaylor already [made a query for it](https://meta.discourse.org/t/what-cool-data-explorer-queries-have-you-come-up-with/43516/4).

The stat would be:

### Trust Levels Achieved: 34 (up 5%)

- TL1: 20
- TL2: 10
- TL3: 4
- TL4: 0

---

<div class="post-metadata">

### Author: ![saintsebastian](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/saintsebastian/32/59341_2.png) [@saintsebastian](https://meta.discourse.org/u/saintsebastian)
#### Post date: [August 18, 2016, 5:04pm UTC](https://meta.discourse.org/t/rgsoc-2016-visual-forum-analytics-community-discussion/46905/13 "2016-08-18T17:04:42Z")

</div>

We have answers to these (sorta)

- the time zone question is also a concern of ours. We are not sure, that discourse db has info on user’s time zone, but we think the default is given admin’s timezone and we can maybe have some indication of that on the page. The bar chart for visits is supposed to be adjusted for time zones, we can also just show time slots based on admin’s time instead

- we wanted to indicate which users are actually active on the forum and logged in today, but maybe the trust level info is more representative, idk, that’s something to think about

- so the pie chart is basically information on topics’ success (new topics which have no replies, topics with one reply and topics with “many” replies), and this chart is separate from top categories section

And thanks for the feedback. We are really excited to implement this idea.
