Piwik analytics for communities

Continuing the discussion from Analytics:

I would like to know how flexible would be right now to provide a piwik instance we can use for all community sites so people don’t feel obligated to use Google Analytics.

Some communities (like Mozilla Hispano) are already hosting their own piwik instance for their sites, but it would be great to have a more reliable central one.

This sounds really interesting, but we could really use more hands. Is someone from Mozilla Hispano interested in leading this with our help?

I guess that would depend on what “leading” means :stuck_out_tongue:

@safwan volunteered to help on IRC, he’s been looking for a contribution opportunity.

@tanner @yousef - thoughts?

Maybe @pochy is also interested in helping with this? :slight_smile:

sure, what do you need?, I can work on this after 1 november! :sunglasses:

Yap. I can help to set it up and do rest other things.

well, I can already start to work on this, how we started? @majken @safwan

So at first we need a VM or server where we can host our Own Piwik. We can also have a subdomain like piwik.mozilla-community.org. Then need to know which website we should impliment. Discuss with the Webmaster!

First thing would be to make a PRD (Product Requirements Doc). We need to do a better job of justifying why we’re taking on a project. Confluence has a really good template for writing PRDs.

I’ve copied the template into a Google doc so it’s easier to collaborate on.

We’ve started filling in the PRD so that it’s a bit more obvious what should go in it.

However, we really need to get an idea of what the goals are for communities. Why do communities need analytics, what will they actually use them to do?

@bernimel @plaurino can you help here?


In my opinion, it’s going to be easier to create several scenarios where we could define the different metrics a community could be interested in. Every community is going to be in its own moment, in which it would have different needs, depending on its goals, size & areas, age, resources, contents, …

In general, I think that we could say that communities have 2 goals in their websites. 1) getting contributors and 2) promote their activities and their contents. Depending on the status of the community, one of these goals can be priorized.

For example, a just born community possibly needs to focus on knowing how their website perform in order to get people, how many users land in their contribute page and how many of these are really interested in contributing. It would be also interesting to know what’s the navigation flow, or the time users are spending in different pages or contents related to participation. This would help the community to know what tasks/projects/areas are more interesting and engaging to their audiences.

In a more developed community, with more contributors, structure and more mature, I think that focus is different. In the case of Mozilla Hispano, the questions we try to answer with Piwik are:

General: goal here is to know how is the impact we’re having. Reach.

  • What’s the global amount of traffic in our website? And its
    evolution through time?
  • In which hours are we receiving more visitors? And its evolution through time?
  • Which are the sources of our visits?
  • Which words people are searching to make them land in our website?
  • Who are the references that are sending us users?

Contents: goal here is to offer what users are interested in. Loyalty & Engagement.

  • Which areas are most visited? And its evolution through time?
  • How much time are users spending in our website? And its evolution
    through time?
  • What documentation is more consulted? And its evolution through time?
  • What articles are the most read, shared and commented?

SM atributtion: goal here is to know how our social media comm is working. Promotion.

  • How many users come from our social networks messages? And its evolution through time?
  • How much does every social network contribute? And its evolution through time?

I hope this helps as a starting point to complete that PRD. Ideally, I would encourage you to think about 2, 3, 4 community possible moments/goals. I just listed two, but I can guess that there’s more. From that, to create a list of questions and metrics will be easy. Then, the product (Piwik) will serve as a tool to answer those questions and guide the community in its road.


1 Like

I was thinking about this last night, and I think that perhaps a first good goal for using analytics would be to see what community sites are actually doing well, and what they’re not.

As part of our efforts to improve Community Sites, it’s on our radar to better define the roles of community sites, and also to establish an MVP (minimum viable product) either for new communities, or communities who are struggling with resources to keep a site going.

Putting analytics on as many community sites as possible just to understand how sites are performing would be very valuable to understanding what the goals of a community site should be.

1 Like

@majken Any update in this?

Not really, we need someone to drive it, or it has to wait until the
current contributors have time. You interested?

Yap. I am interested. Can you guide what to do next?

Sure, so first of all, “driving” something doesn’t necessarily mean having the answers. It means making sure the questions get asked, and prodding for responses if the discussion stalls. It’s also not necessarily doing the work but finding out what work needs to be done and matching it to someone who can do it.

So the step we’re on is making sure the PRD looks good. Part of that is figuring out the hardware requirements. It’s also figuring out the MVP (minimum viable product) or “pilot”, what’s the least thing that we can do that will be useful, that we should do first? What data should we gather, for what purpose?

I think most of us on the team would be responsive to an @mention with ideas towards specific questions. We just don’t have the time to remember to ask them, and to work the answers into a plan. @yousef and @tanner can help you with technical implementation questions.

I recommend adding a CDN for serving the piwik installation javascript.

Coming into this discussion late but with a dissenting opinion!

I think that for the moment, Google Analytics makes more sense for measuring information from community sites for the following reasons:

  1. The majority of community sites are already using GA very effectively (Slovenia, Czech, German, Italian, Japan, Turkey and Taiwan)
  2. In GA you have the opportunity to see all of the accounts you have view-rights on in one place which makes it much easier to run cross-site reports and identify long-term trends.
  3. Mozilla internally uses GA on all of it’s sites so as we push to get the immense value of these community sites more recognized, having it in a consistent format only makes it easier and lets us have metrics teams get involved and support data analysis.
  4. If people are concerned that we use GA because Google is seen as untrustworthy, all of our emails, docs, spreadsheets being google would seem like a more pressing issue. Non?

Let me know what you think!