In a recent article Cedric described the ability to use data as “sense”, alongside tacit knowledge and the qualitative information you get from talking to customers. But if using data is a sense, how can you improve it so you can reach that mythical goal of becoming “data driven”?

Having Data Sense is Not About Mathematical Skill

When someone mentions improving their “data sense” or “data understanding” one of the first places they look is to upskill themselves in more complex data methods. They start looking on Coursera for courses on regressions, Machine Learning, Artificial Intelligence, SQL, and spreadsheet functions/macros. They believe being able to do more advanced techniques is what will unlock their “data sense”.

The belief that mastering ever-more complex methodologies is the path to being data-driven is wrong. Being data-driven means using data to make accurate predictions about your business, and using those predictions to take actions that enhance performance and drive growth.

Being data driven is not about your knowledge of advanced statistics or complex analytical techniques; these may help in certain circumstances but are not the first place you should look.

Instead, the first step in developing your “data sense” is to treat it like a sense. You need to continually expose it to the input that matters and give it feedback to let it know when something relevant is occurring.

Imagine you were training to become a sommelier (wine taster) and wanted to improve your sense of taste to different kinds of wine. Some actions you could take to improve your wine understanding would be:

• Taste different kinds of wine regularly, reading the labels as you did so to understand how different regions and vineyards tasted.
• Visit different vineyards, learning how different climates, grapes, and growing methods come together to create the final taste.
• Experiment with blind tastings, to test your ability to distinguish different grape varieties and growing methods.
• Notice how different serving methods can change the flavour, all the way from temperature to letting the wine breathe.

Developing your data sense is the same, you need to be continually exposing yourself to your businesses’ data, seeing the relationships between different metrics, and learning which metrics matter and which do not.

Developing a Fingertip Feel for Your Business Metrics

In a previous article Bennet discussed the simplest method for getting started with analysing your metrics; taking your most important metrics and plotting them every week. This is what Amazon did and still does to understand its business, and is what we do at Xmrit and Commoncog.

To help you build your data sense, as you are doing these reviews, you should ask yourself 4 questions each time you look at a metric:

1. What is routine variation for this metric and what would be exceptional variation? XmR charts are a great way to do this, but eventually you will gain a natural intuition for your most important metrics.
2. What is the trend or seasonality for this metric? This is important as otherwise you may be fooled by a rise or drop in your data. It may reverse, or it may continue, without knowing seasonality and trend you can’t make sense of the variations.
3. Is there anything happening that I predict will or won’t impact this metric? This can be actions you have taken, e.g. launching a new campaign or product, or something outside of your control, e.g. a new competitor entered one of your markets.
4. What are the common relationships for this metric to other metrics? This is the hardest question to answer, and if done well eventually helps you build a causal model of your business. Causal models are covered in the Metrics Masterclass as one of the most important goals of the Process Control worldview.

Being able to answer these four questions about your key metrics is the heart of being “data driven”. You can look at the key metrics that determine the fate of your business and make sense of them. You are not overwhelmed with data, as you know what routine variation looks like, what data points are exceptional, and which metrics influence one another.

What Having “Data Sense” Looks Like

Signs of someone having a strong “data sense” for their business are when they can make statements like:

Quotes have been trending upward, but I haven’t seen any increase in the dollars of closed deals, what is going on? Are we quoting a new customer segment that is not being analysed on our metrics dashboard?

Why are our inventory levels for raw materials the same as last month? Shouldn’t we have been building a stockpile since lead times from that new supplier have been so variable?

We used more power per tonne and saw more rejections on line A than usual, without any change in the product mix. Do we have a new operator or miss a maintenance check?

Visitors to the website are up, but that doesn’t tell us much as they usually spike the first few months in September as university students come back to campus. The spike is at the same level as the last few years.

Lifetime value has been declining for our newest cohorts, that could be linked to that new marketing campaign we launched last quarter. We better go and validate if that is really the case.

You will also notice that none of these require advanced mathematical skills. They are the result of having a deep understanding of the most important numbers in the business, looking at them regularly, and learning to notice when something exceptional requires further investigation.

As you get better you may even start to recognise patterns in your data, and be able to predict why something unusual is happening just from the metrics alone.

Let’s look at a case study of putting data sense into practice with a recent problem we noticed at Commoncog.

Case Study: Commoncog’s Free Subscription Signup Mistake

In an October 2024 Commoncog Weekly Business Review (WBR) we were looking at the Free New Subscribers and New Unsubscribers metrics. In newsletter businesses these two metrics behave differently depending on if you have published an article in the previous week.

• Publishing Weeks: In publishing weeks, you see higher subscriptions, as people read and share the new article. You also see higher unsubscribes, as people who were not that engaged in your content receive an email and decide to click “unsubscribe”.
• Non-Publishing Weeks: In non-publishing weeks both of metrics are lower than publishing weeks. Fewer people read and visit the site, as there is no new content to draw them in. You also see fewer unsubscribes, as people are not reminded to click the unsubscribe button.

With this behaviour in mind over the last few weeks we noticed an odd behaviour in these metrics. Cedric was sending out emails each week, but we were not seeing the expected elevated subscription numbers. We were however seeing expected numbers of web visitors.

Maybe these were just bad articles and there was nothing to investigate, but it felt off. We asked ourselves “have we changed anything in our emails over the last few weeks that could explain a change in subscriptions”. The only thing we could think of was Cedric had started to send out pure case articles, which linked back to the case library.

When we opened the case articles that were sent out, we realised that we had forgotten to add email signup forms to both the article page and the case library page. Therefore, when the articles were being shared there was no way to signup, even if you clicked the header to go to the case library home page.

This is a simple example, but even to do this required us to understand the following about our business data:

1. What was routine vs exceptional variation (so nuanced that this didn’t even show up on an XmR chart.)
2. What actions we had taken in the last few weeks, and what impact we predicted it should have had.
3. What we had changed to our normal process in the last few weeks, even if we didn’t think it would impact anything.
4. Understanding the customer journey, and what they would see when they read an article.

Building your own data sense will take time and energy. However, the payoff is that you will be able to go beyond your gut instinct and customer conversations when you need to make decisions about your business.