Insights from simple data analysis.

Things you can do to make sense of data without being a data scientist.

In the previous article, we talked about collecting data and using it to make better decisions. Data analysis can be complex, but it does not have to be. This article will show you examples of simple analysis techniques that can give great insights that everyone can do — even if you are bad at math.

Note I’m not saying this is what you should do; it’s rather an example of what you could do. Your context will determine what types of insights would be most valuable for you. What I hope to achieve with this data is to make you think of the possibilities!

Analyzing the data with a specific view of gaining insights can be tricky, but using some of the techniques described previously might prove helpful also, and here we will keep it VERY simple.

The simplest way to start doing data analysis is by comparing two sets of data — either to determine if a correlation exists (ok, that is math, but it's fairly easy to use the Excell CORREL function), OR by plotting data on a 2 x 2 Matrix.

We use the second technique the most and usually start with this technique. It's great to do this in a group on a whiteboard, and it is easy for everyone to pitch in! Other statistical analysis methods are left for later, once we have gained value from the low-hanging fruit!

We believe that the key is that you should start simply and get values quickly.

Many insights that you did not have before are straightforward to discern using the matrix plot technique.

Another type of matrix that you can use is not to have a scale on each axis but fixed data on each axis; this is particularly useful if you want to answer questions like X vs. Y where both X and Y is a definitive statement, like for instance;

Which type of customers uses product X to do what kind of jobs?

Usually, the categories are not easy to ascertain the first time around, and often it’s a progression of insights that led to a more definitive answer.

To make the most of it — especially when combing customer and product insights on the same matrix, look for clusters and try and understand the source of commonality!

We often use affinity mapping to try and find logical clusters in the displayed data. In this technique, data with similar attributes are grouped. Doing so very quickly gives you an indication of the significance of the grouping in whatever measurements you choose to use!

It’s best to show you an example of a progression of analysis that leads to a good decision outcome.

A customer wanted to understand what their product is used to evaluate if the product feature-set was market-related.

QUESTION 1: Which of our product features are used by (which) customers?

The result of asking this question was represented by a percentile of customers using each feature.

It clearly showed that some features were used nearly all the time, and some were rarely used.

But what do you do with this data?

The only logical conclusion was that they might consider dropping some of the features in the next release. But is that a wise decision?

Then they compared their features with their competitor’s features, and they were pretty much the same features.

The sales and marketing team resisted the idea of dropping features because it would make their product ‘inferior’ to the competition’s products and harder to sell.

The first question was not a great question to ask because it failed to yield actionable insights.

We then introduced the concept of JTBD (Jobs-to-be-Done) to the customer, and the question was changed.

QUESTION 2: Which jobs do our products do for customers?

To their surprise, the list of jobs customers used their product for was longer than the list of features that customers say they use. Many of the jobs customers used the product for were not what the product was designed to do. The “feature” was re-interpreted by the customer and used to do jobs that the software vendor did not even know existed!

This feedback provided actionable insights, but not enough; they need to understand better what was happening here.

They then grouped customers by the jobs they use the product for (well, actually by customers with similar jobs to be done — we call them job clusters), using affinity mapping. (Black dots are outliers)

They then looked at what the shared attributes were of the customers that belonged to each job cluster. Defining many attributes, they again used affinity mapping to develop shared attributes between customers.

By doing this, they were able to plot jobs vs. customer attributes. This insight proved both useful and actionable!

The company realized that there were three adjacent markets to the market they initially targeted.

Asking more questions to customers in adjacent markets showed that two of these adjacent markets not only had many unmet needs (jobs to do) but that these markets were also underserved.

QUESTION 3: When doing this job, is there a part of the job you cannot do with our product?

QUESTION 4: If you can’t use our product, do you use another product to do that part of the job?

In each of these markets, customers only used a few of the full set of features, but many aspects of the job that customers had to do, needed lots of manual work or other tools that do not integrate well with the company’s product.

The company elected to drop a few of the outlying features in its core product and keep features used somewhat infrequently. They then develop two completely new products for these adjacent markets, still using the familiar core name but adding “construction” and “engineering.” In so doing, they could leverage the features where they had long-standing competence and only needed to develop a few new features, which they did in record time!

SOME PLOTS ARE MANDATORY

For ADapT to work, some data must be plotted against, preferably the product lifecycle (diffusion of innovation) graph!

We use this as a trigger to make innovation decisions in ADapT.

So what exactly do you need to do?

Step 1 — create a list of all your current products.

Step 2 — order the list by volume (units shipped), turnover, and profitability.

Step 3 — do the same for the last two years (either by quarter or by month).

Step 4 — for each of the products, do a breakdown of who the customers are and where they are. Once again — do this for two years by month or by quarter.

Step 5 — look at trends and select one of the metrics in Step 2 above, PLUS profitability.

Step 6 — graph the trends

Step 7 — use one of the matrixes (Ansoff, BCG, or Innovation Ambition and plot your products on the matrix!

Step 8 — what does that tell you?

Step 9 — answer the following questions:

• Which products grew in real terms and profitability?
• Which products show slowing growth but are still growing in profitability or maintaining profitability?
• Which products are showing a decline in growth?
• Which products are showing a decline in growth and profitability?
• How many new products are on the list?
• Are they new or just old products renewed and spiced up?
• Have you in the last two years gone into another customer market?
• Have you in the last two years introduced a new product?
• If you have introduced a new product, was it yours, or did you get it from someone else (partnered with someone)
• If you introduced a new product that was yours, how is it doing?
• Have you been able to break into the mainstream market with your new product?
• Have you killed a new product because you could not get it to work?
• Have you killed any product because it was not selling?
• Have you killed any product because it was no longer profitable?

Step 10 Use your newly gained insight and plot all of your products on a product lifecycle (diffusion of innovation) graph? Use the following convention:

· Use an up-arrow to indicate growth and a down arrow to indicate a decline in sales (choose if you want to use turnover or sales volume)

· Use different sizes of arrows to represent the contribution of the product to your turnover

· Colour arrows using RED, AMBER, and GREEN to indicate if the level of profitability is improving, staying the same, or declining.

Let us show you an example of a plot with eight different products plotted on the graph (and the appropriate plot area for each).

To follow an example:

[A] A mainstream product that shows growth, no pressure on margin.

[B] A new product is showing promise in some mainstream market segments. Preferably where there is little competition and generous margins.

[C] A mainstream product that shows slow or no growth and pressure on margin.

[D] A mainstream product still earning good revenue but with pressure on margin

[E] A product in decline with lower margins but still profitable

[F] A product in decline with significantly low margins or even not profitable. The only reason why we keep it alive is that some customers using our other products also use these products.

[G] A new product showing promise, still not profitable, but one or two customers love it

[H] Something we stopped doing because it just does not make sense!

The analysis and insights from above help us to understand the possible product and innovation strategies better!

At a high level, we can describe these by using the concepts of jobs-to-be-done.

Both Christensen and Ulwick postulate that jobs at their core remain the same for very long periods, the nature of how the job is done may change, but not the nature of the job.

We tend to think about four different kinds of jobs, as these closely relate to how you choose to innovate.

We agree that you can most probably break down a job to a level that you can show that it was done in ancient Egyptian times — but that is not very helpful. Jobs-needs to relate to what our customers are doing and doing now!

1. Jobs that customers are currently doing (and want to do better)
2. Jobs that customers want to do but cannot at the moment (some constraint needs to be removed).
3. Jobs that we do ourselves to get our products ready to be hired to help customers get their jobs done (you can think of it as our internal environment and also as non-functional specifications).
4. Jobs that customers did not even know they have, but once they know it, they would love to do it (revolutionary ideas or products that disrupts a market)

This article is part of a series exploring the use of Agile ADapT™, a Digital Transformation Method for incumbent organizations struggling to compete in the digital age.