How to turn raw analysis into actionable insight
You’ve done the research… now what?
In this article:
▶️ How to turn research into summarised, useable knowledge.
▶️ A step-by-step synthesis process for extracting useful data points, cross-referencing and refining them to derive actionable insight.
▶️ Ways to structure and communicate findings for maximum clarity and impact with your stakeholders.
We’ve all been there. User interviews. Competitor analysis. Customer feedback. Industry reports. Site analytics. Financial reports. Lots and lots of data but…so what?
Data is nothing without insight.
Doing the research is the easy part. Translating your findings into something you can share and act on… that’s tricky.
When you have multiple data sources, making sense of the noise is even harder. According to the Harvard Business Review, this is a critical business capability:
“What matters now is not so much the quantity of data a firm can amass but its ability to connect the dots and extract value from the information.” — HBR, Building an Insights Engine
In this article, I’ll give you a step-by-step process to refine your analysis into insight that will power your decision-making.
The Synthesis Process
Deriving insight from data is a craft. It takes skill and practice, but the more you do it the easier it becomes. The technical term for this process is ‘synthesis’.
I’ve spent a lot of time around professional services and consultancy firms and observed how they add value by doing this consistently and effectively. I’m sharing that knowledge with you.
Here’s a step-by-step model that will work for most scenarios.
These steps will help you to extract the most useful data points, combine them and refine them into usable statements that you can shape into an actionable report.
It’s like mining for gold and then making a crown 👑
Can AI do this for me?
The good news is AI can do some of the heavy lifting for you. But like most skills, knowing how to do it yourself is fundamental.
You’ll get mixed results — I’ve tried using three different AI solutions to Cluster some findings and wasn’t happy with any of them. On the flip side, ChatGPT is great at generating ideas for Considering, Prioritising and Summarising.
How to do it
Let’s go through each step and explore the key actions:
1. Highlight
Whatever your raw data is, go through it and pick out key phrases, numbers or insights. Anything that grabs your attention. You don’t know yet what’s going to be useful — just highlight whatever jumps out.
2. Categorise
Review your highlights and label them. Tag them by topic or theme.
This may take a few passes to get right — I often end up with way too many tags the first pass through the data. Consolidate, and aim for around 10–20 tags. More than 20 becomes unworkable.
3. Cluster
Think of Clustering as the smelting process. We’ve extracted the raw ore and are now applying some energy to transform it. These data points need to mix and move around. What emerges will barely resemble what went in.
Another term for Clustering is Affinity Grouping. It’s really simple — just put like with like.
Copy the labelled highlights from the raw data source and drop them somewhere you can manipulate them. If you’re analysing multiple data sources, this is the step where they come together and cross-reference.
Choose the right tool for your data type. You need something that lets you easily manipulate and move things around. I tend to go for an online whiteboard like Miro.
Split the labelled highlights into clustered groups. Draw a circle around those groups and write a header for them.
Be patient with Clustering. I’m often daunted when I first start this stage and it seems like there’s a mountain of data to work on. Take it step-by-step, and don’t rush. Once you get into the flow it’s a pretty quick process because you’re not considering these data points in detail — just sorting for now.
Also, be flexible. You may want to subdivide into separate clusters within those initial groups as insight starts to emerge.
👉It can be tempting to group highlights by the type of impact they have— pains, gains, and jobs to be done are common examples. Don’t do that.
Clustering is about making connections from the data and contextualising it. Understanding the impact of these data points comes later. If you have additional tags already from the original research then keep them noted, but don’t let it affect the groupings.
For example, a user might say that they find the registration process high effort — pain. But, your analytics shows that customers who have registered are more likely to repurchase — not a pain. These two data points need to be considered together to be valuable. They’d get clustered under a theme of ‘Registration’.
I like to colour-code the clusters at this point. It makes Referencing easier later on. Keep the colour coding going through the next steps.
We’re working closely with the data now. Maybe you want to go back and change some of your Categories? That’s fine. This is a flexible process. Insight emerges organically as you follow these steps. It’s the human brain doing what it does best — making connections.
When you’re happy with your Clustering, it’s time to extract the good stuff.
4. State
Write ‘Key Takeaways’ beside every cluster. Then, make yourself find three or more statements that illustrate what that cluster of findings is telling you. Keep it factual, objective and non-judgemental. Statements like:
“Research participants felt that the subscription price of $15 was more than they were prepared to pay, and they were more likely to purchase a subscription if they had made 3 or more purchases in the last year.” Or “60% of users abandoned checkout after 30 seconds, users signed in with single sign-on were 10% less likely to abandon checkout.”
You’ll end up with a set of statements that outline what your combined data sources show.
Copy the statements over into a new workspace. Next, you’re going to figure out the implications of your findings.
5. Consider
Put your statements in a list, with two headings: Statement and So What?
Read each statement and ask ‘So What?’ for each one.
Consider the fact. What are the causes and effects of that fact? These could be in relation to your goals, the competitive environment, your customers, risks or opportunities.
I like to give this stage some space and indirect focus — take a walk or leave it for a couple of days to give your brain room to consider the question from all angles.
You might want to share your findings and consult with some experts. The Futures Wheel is a great tool to help with exploring consequences and implications in a workshop.
👉You might find that this process uncovers more questions to be asked. Either go back around and do more research if you’re missing something fundamental or take it forward into your recommended next steps.
6. Prioritise
Rank by implication. Prioritise the statements which will have the biggest effect on your goals.
Look for the key statements to take forward as takeaways. More than seven is too much, you’ll lose your audience and they won’t remember anything. Three is preferable.
6. Reference
You want to ensure your takeaways are backed up by some references from the raw data.
Go back to it and extract one or two quotes or figures to illustrate your statements.
Remember that I suggested colour coding? That will help to pinpoint the sources you need.
7. Summarise
Now we get to the reason for doing all this work. Write up your research.
I like to use the Pyramid Principle by Barbara Minto. Start with the answer and build out the detail for direct, concise and logical communication.
▶️Begin with the action that needs to be taken
▶️Outline the key takeaways
▶️For each takeaway, outline how you uncovered those findings
▶️Scatter in some references for colour
▶️Leave your audience with a short summary and the next steps
For reports that are going to a wide audience, I refer to this guide from Neilson Norman: Creating Engaging Reports & Asynchronous Presentations which shows how to appeal to different types of readers
And we’re done! Turning analysis into actionable insight.
The more you practice this skill the better and quicker you get at it. Sometimes, I don’t follow all 8 steps, others I have so much data to distil I need to take it slowly and cycle around the process a few times.
This is a basic framework — adapt it to each specific situation. Good luck!
Found this helpful? Check out some of my other Product Management tips:
If you enjoy reading articles like these, consider signing up to become a Medium member. It’s only $5 a month and gives you unlimited access to all stories on Medium.
If you sign up using my link, I’ll earn a small commission.
