Democratization of Insights and Experiments

We’ve been trying to shift focus the last years from the idea of democratizing data towards democratizing insights and experimentation.

That has been a journey of both shifting focus and goals but also organizing us differently.

These are some thoughts…

The Starting Point

Some words have so many positive connotations that they almost lose all their meaning. Democratization is one of these words. It is almost impossible to find someone that openly opposes the idea of democratization (let us disregard some current global players for a while).

In theory, democratization is so easy to accept so it becomes virtually meaningless. Yet many are, like us, fighting hard to change our organizations.

Starting the democratization journey we earlier talked mainly about democratization of data. It included things like:

• We should own our data or at least have control over it
• Everybody should have access, don’t keep barriers and make it easy to understand
• We need to move from a Need to Know-culture to a culture of Trust and Transparency
• …

Common fears included, what if data isn’t correct or is interpreted wrong. What if people lack understanding of different figures floating around.

Skilled and well-meaning analytics departments served as gatekeepers and interpreters of data for the organization. The primary output was dashboards and reports and it was unknown if the reports were used or understood.

Organizing for Insights

Organizing for insights meant some teams chose to change focus.

Moving towards a more insight-driven approach, our DataOps-team, who have been building a data platform for our huge amount of data changes our goals from:

• not only making data accessible and correct
• not only making data easy to use and understand
• not only keeping data secure and well documented

But also: Is data actually driving insights?

Much tougher to accomplish and much tougher to measure. That change shifts focus from reports and dashboards to actual usage and understanding of data.

Sometimes working towards democratization we can step over simple power issues.

If we use the power definition “Power refers to asymmetric control over valued resources, which in turn affords an individual the ability to control others’ outcomes, experiences, or behaviours and regard data as a resource, sharing could initially mean a loss of power for some people.

A solution and positive aspect of that can be that some people’s roles might shift from being report-builders to being advanced data- and insight-facilitators. Which I believe can be much funnier.

Progressive Insights to change Public Service

Our newest team Pi, short for Progressive Insight, works with helping out the planning department and program leads to plan content acquisition, production and publishing.

The team was created from the insight that in order to make SVT Play a much better service the best way is targeting content — both through acquisition, publishing and exposure.

With the goal to increase retention and in the long run create a more interesting public service offer to all users the effect is even more interesting.

Abisko — our AB-test platform

In one of our tech innovation sprints (described by Hilda and Natalie here) an AB-testing tool, Abisko, was created. Still unfunded, it is now developed by a few enthusiasts that meet three days a month from their ordinary teams to build a better platform for our client- and exposure teams.

All in order to democratize experimentation and thereby maximize learning. Still at low levels the number of experiments have tripled the last three years. We’ve realized it is as much a cultural change-issue as a tool support-issue. Viewing opportunities as hypotheses that can be validated as a practical change.

New Goals for Data & Insight

What becomes important then? Moving from making data accessible to making data have an impact means new and much more interesting perspectives.

Focusing on the democratization of insight, some common aspects became something like:

1. How can we make data super-easy to use and understand? Helping users through minimizing friction becomes important. Both getting started and minimizing “time to first aha-moment”. Can we help users share insights and show it to others?
2. How can we make it easy to do things right and avoid mistakes? A relevant stress factor is if data are circulated that are misinterpreted or false. It lowers the trust for all things data
3. How can we make using data a habit?How do we create users for the long game? How do we create a data embracing culture? It might be as much an educational effort as product development.
4. How can we meet super-users with possibilities to build advanced models and do advanced analysis? Some patterns are seen through more advanced data science. How do we cater for these users?
5. Where do we need to integrate to make the highest impact? Data is spread through the organization in various ways through systems and tools. How can we make integration and thereby usage easier?
6. …

And most important of all is maybe that it gives opportunities for much closer collaboration with users and stakeholders, to understand and meet them better which is always nice.

A sneak peek into our challenges now and ahead. Don’t hesitate to get in touch if you want to know more or join us in the funniest mission at SVT…