Some of my insights about (BI) insights

I’ve been privileged to spend some of my time working with insights, analytics, decision intelligence / BI for many years now, all with the mission of helping people make better decisions. Here are some light reflections/insights about some of my observations to date. All of them hold true independent of how much you already leverage AI in your work. Hope helps ?

1) What problem/decisions we try to solve for are often not sufficiently specific

2) Call to actions are often missing

3) Data quality and availability remains a major challenge

4) Many people don’t care enough about the truth

5) Diversity of data is important

1) What problem/decisions we try to solve for are often not sufficiently specific

What are your organization’s three biggest problems? Three most important decisions? Three most important things to learn? Three most important questions to answer?

Many people, organizations and not least leaders tend to jump headfirst into exploring data. That’s not bad, as sometimes you need to explore in an expansive way. That said, every organization, team and individual would benefit from first defining what specific problem you try to solve for, what specific decision you want to enable, what you are specifically looking to learn or what specific question you want to answer.

So instead of asking an analyst? to “pull some data about this or that”, ask them to help you solve a specific problem, make a specific decision, learn something specific or answer a specific question. The insights you will get will most often be much more relevant and strong.

The strongest analysts and leaders don’t rush ahead, but they have the courage to first invest sufficiently in shaping the best possible brief for what they want to accomplish. This is not something new. A Yale Professor stated already in 1966 “If I had only one hour to solve a problem, I would spend up to two-thirds of that hour in attempting to define what the problem is.” (many people think Einstein said something like ‘if I had 1 hour to solve a problem I would spend 55 minutes defining the problem’ or similar - see point 4 below). I think many people understand the importance of defining what to solve for, but I see the actual practice of this linked to insights and analytics being under-invested in.

Next step: Next time you engage an analyst, invest more time than you think you need in getting really specific about the brief.

2) Call to actions are often missing

What defines the quality of an insight? For me the quality is measured by to what extent it changed something. For example, triggered an action that wouldn’t have happened otherwise.

Have you ever seen a table, chart or data point where there’s been a box next to it stating “Potential next steps”? If you have, in what percentage of cases has that clarity of next steps been there?

My observation is that analysts and BI teams are often a bit careful to recommend how to best action the data and insights they provide. I argue they shouldn’t be. They often have a very strong understanding of where the data comes from, what it signals, and what would be good to do next as a result of what the data indicates.

If you are an analyst, always (in 100% of cases) add “Potential next steps” or “Call to action” or similar to any tables, charts or data points you provide or enable. Don’t worry about whether they are the exact next steps you will ultimately agree on or not. Just the fact that they are there will trigger more of the right discussions. If you are a leader receiving the gift of data points and insights from an analyst, start by high-fiving the analyst (always), then make it abundantly clear how we will act on them.

Being very clear on actions / next steps from insights puts you in the top of people and organizations leveraging BI. Also, if you struggle to formulate next steps based on the data you are reviewing, then perhaps you are engaging in non-productive analytics.

Next step: Always ensure there is a “Potential next step” next to any table, chart or data point you see.

3) Data quality and availability remains a major challenge

Sometimes the work of making high quality data available for BI seems like a herculean and insurmountable task. That said, we must never give up. Even in the age of AI-enablers, making high quality data available is still a lot of heavy-lifting work. If it takes a week for an analyst to produce two insightful data points, it’s not because they are dumb, lazy or enjoy playing board games when you don’t look over their shoulder. It’s because the data infrastructure you provide them with is of very poor quality. I don’t see this challenge going away any time soon for most organizations.

Beyond looking at the quality of available data, there is so much data that is not yet readily available. How easy is it for example to conclude how well a municipality in Sweden is performing? Would you be able to do that in an hour? A week? A year? How easy is it to conclude who the top-5 school platforms are in the world? How easy is it to conclude whether the product you just bought includes any element of slave-like labor? How easy is it for a medical doctor to know who the three people they should learn from in the world are? Most of the really important questions still lack a lot of the data availability to be answered.

Next step: Do a major review of the data availability and quality that would really help your organization better achieve your mission.

4) Many people don’t care enough about the truth

Are you a leader? Have you ever seen the result of an employee study? Or customer satisfaction study? If you have, was the error margin spelled out? Was the sample and universe size spelled out? Do you even care if the numbers are correct as long as they make you look good or support your point?

I believe everyone needs to be a guardian of the truth, and work towards the truth no matter who you are. I see confirmation bias at play every single day. Leaders looking for numbers to make their case, not asking uncomfortable questions about data significance, confirmation bias, publication bias or any other factors that may be at play. This can result in decisions that don’t add any value, or even worse, have negative value.

I know it can be very uncomfortable to raise questions around what the data truly tells us, the validity and significance of it. That said, it’s like many other things in life, short-term discomfort leads to long-term success. The better you and your organization will be in truly leveraging data in a less biased way, the better decisions you will make and the more success you will have over time.

Next step: Read/listen to the books “Think again” and “How to make the world add up”, and always challenge the quality and validity of the data - what it really says and doesn’t say.

5) Diversity of data is important

For some decisions I see organizations often relying too much on just one type of data. It can be quantitative analytics where they look at an organization and decide to make organizational changes primarily based on the quantitative data. Or it can be anecdotal and few one-on-one engagements with a handful of people that leads to a decision of a change in a solution.

Data analysis is very much like a non-perfect game of investigation. By combining many different data sources and methods, you can start to zoom in on a concept of reality/truth. It requires a bit of imagination to go beyond the obvious data sources and methods. It requires a bit of stamina to not be happy with the first simple analysis. It requires a lot of teamwork to learn from others.

Personally, one of the things I love the most is deep interviews with subject matter experts. A lot of them. First, because they know so much, and often bring obvious but overlooked insights to the table. Secondly, because when you interview a lot of them, you start to see what they have in common, and where they differ, which makes for two buckets of very interesting data and insights that you can act on in different ways.

Next step: For your next analysis/investigation, consider what “invisible data” you’re not yet taking into account and challenge yourself to be creative.