A sketch to a "how-to guide" to make your data make a difference!

Data as the new oil, data scientist as the new omniscient oracles, predictions and prescriptions made by fancy machine learning models has been hot stuff for some years and still evolving – management phrases like “snoozers are losers” and “the early bird catches the worm” makes everyone running very fast in the areas of business intelligence and decision intelligence adding AI and ML. Though all this data, all these numbers, all the statistics, all the infographics and various presentations: Does it really matter? Do you make your analysis-work make a real difference?

The difficult move from data to walk the talk!

Often, we - as analysts - have an idea like a classic rational assumption that all knowledge and new insights from data itself motivates any necessary change. As data is an integrated part of a decision-making process:

Data – Discuss – Decide – Do!

However, changes are not created by the data or the discussions and to frank not even by decisions! Changes are mostly carried out by the doing part in the process. It’s the actions that changes systems, makes shift in the culture and can measured as different behavior and new impacts ... so the main task of any analysis should be to nudge the audience for a more strong commitment.

Actions speak louder than words! Then it’s not enough to bring your audience to discussion or decision mode – you need your audience to act after the decision. The good thing is, that data will talk, if your torture them long enough! Let me show you a few examples of effective data storytelling using simple data visualization to engage your audience.

Three simple tricks to get your data narratives effective nudging.

First, go for time serie data. I admit the first trick is not always easy to establish as your default data but time series data are very powerful as communication. Everyone can relate to changes over time, so you engage your audience, when your data narrative shows what happens over time.

Second, always prioritize to make a measurement of an outcome. Remember: It’s always about the impact! Especially a management can easily relate the results to the value chain:

Purpose – Plan – Resources / Activities – Output – Outcome

The logic is simple, if we change something in the value chain, then we suppose to see a change in the outcome. That’s the underlying meaningful story of why we do as we do. Let your data clarify the case, set spotlight on an indicator or a proxy that represents outcomes in terms of money. Then your data have the full attention from your audience!

Third, you do not need a lot of fancy graphs to tell strong stories with data. Actually, I think there is only eight archetype narratives in your time series data, that you need to know to be able to nudge your audience to action. 

The eight archetypes narratives

The eight archetypes are the data stories touching the feelings of hope, glory, remaining status, fear and horror. Presented as simple as a descriptive line chart. Maybe you can spice up the graph with showing statistical confidence intervals around your predictions, because doubt is not the end of wisdom, but the beginning of new insights.

Almost the line chart tells the story itself, nevertheless you should not forget to bring a few lines describing the context and offer an immediate interpretation to strengthen the analytical narrative. 

Story 1: Nothing happens – status quo!

This figure might often - and wrong in my opinion - be passed by as boring, but non-findings can be very important. If nothing happens after investing in new products, or we hired new people, or a changed strategy or the market itself has changed – then a flat curve is an alert! At least it’s an important information for the board of executives. Stability may be a (first) sign of decline. Especially if our predictions for the coming years shows the same level. Always fear a flat curve, if all other things are not equal around you.

Story 2: Going down!

This story of a steady fall is hard to tell. Anyhow, the curve is easy to interpret. We are losing. We are doing something wrong, so this calls for actions. This decline story can be even more effective if you be able to compare with the development for others over the same period.

Regardless of any alternative explanations from critical viewers, surely other are convinced, where there's smoke, there's a fire. The predictions can even be horrifying.

Story 3. Going up!

This is a story of upcoming gains and this is always an easy data story to tell to the audience. Anyhow, that might be wise - as analyst - to do some extra control for other influencing variables. And again, benchmark with others can make this story stronger. If we gain five percent, but the competitors gain ten percent, then the good story turns into a sad story!

 Story 4. The magic point – here is the focus!

When two curves crossing each other, then it is always a fantastic strong story to tell with your data. Curiosity killed the cat, but satisfaction brought it back. This archetype graph automatic makes everyone in your audience data savvy. At least they all have an explanation of the situation. In the finance department this could be the point of break even, as the most classic example.

Story 5. From zero to hero!

From bad to better is a popular data story to tell. And if you are lucky also to have optimistic predictions as well, this is a huge smile and a meeting on cruise control. The only question will be: What is the significant trigger at the turning point? 

Story 6. Over the hill!

Ouch. This kind of data story in your data always hurts. Over hill, but are we also going from rubbish now to worse in the future? You can clearly feel the frustrations and fears in the meeting room. And often there will be a request for more or even better data, and typical there is a need for more investigation to be done.

Story 7. Lesson learned!

This archetype graph is a dramatic curve, ups and downs, but at the same time it’s story about we did managing to muddle through. We won the battle as we defeated the challenges over time. This as close as it gets when data tells a classic fairy tale with a happy ending.

Story 8. Alert: Trouble ahead!

Sadly, this is the story of things going worse again – we must realize going up was not forever. But how do we get back on the right track? You can image Billy Ocean’s “when the going gets tough the tough get going” as the soundtrack for the meeting. Well, responsible leaders always step up – this data story calls for actions and probably more data as well.

As always, at a meeting there are a lot of different opinions and respect to certain matters outside the data, so the eight archetypes above might bring very different suggestions for actions to the table, but feel confident that your data have definitely nudged your audience so they are ready for action.

AI and ML develops - why bother about data storytelling?

Artificial intelligence, machine learning, automation, robots - you name it. Working with automated machine learning, then in the best case the model can make its own decisions – in a digitized production flow as industrial automation this is smart as a cost effective way to improve quality, minimize errors and avoid hacking of systems, so if that’s what you use data for, then it’s a fast run ahead. I think we all see a lot of successful automated ML in the industry all over the world. Still, you should look for the eight archetypes of data stories in your data for the sake of your stakeholders, your sponsors, your board, and at least as documentation to strengthen your reputation for successful automation of business.

For the strategic development of a company or an organization I not sure that truly pure data-driven-decision-making is a clever choice. Or even a legal choice regarding to the European GDPR? I read a lot about automated decision intelligence, and still I haven’t seen the really good and impressive examples in the business area of Strategy or Human Resources. Let me know if you have.

I think there are several good reasons. First of all, do any organization really wants to be purely data-driven? Data might always be a good perspective for the decision making, but also there must be an idea by having the board and executive management – maybe we just want our decision intelligence to be well-informed by data. Then it is not data-driven, but knowledge-based management. Then data storytelling is needed, and the eight archetypes of data stories are valuable to look for in your data.

Second, delegating decision-making to a robot seems difficult as we do never have all the data needed, the issues of unconscious bias also require awareness to the algorithms, and in the end, all in all, it might be an issue about the legitimacy of the leader. We need to keep responsibility very high when making decisions about people, money, society and the nature. In the long run that’s why politics matters more than data in a democracy.

Nevertheless, apart from very busy businesspeople, politicians are probably the easiest group of people to influence with the eight archetypes data stories – so good luck in making your data make a difference!

Please feel free to comment or share your experience below :-)