The Ultimate 11-Step Guide to High Quality Customer Insights
Follow this step-by-step guide to create high quality customer insights for your product or brand that will strategically drive your product or organizational change. Detailed instructions on how to perform a 10-week qualitative research project and convert the findings into actionable customer insights.
42% of failed start-ups never made it because they didn’t fulfil a market need. For an innovation to be valuable, it has to fulfil a need – there’s no use in innovating if your customers don’t see the benefit in the innovation. And this doesn’t just apply to short-sighted start-ups. It goes for any business innovation, be it product changes, marketing strategy or organisational restructure.
The world’s most successful products are those that fit into the customer’s lifestyle and make sense to them – think about how scarily easily children understand the Ipad. As a rule, changes to a product or service should ensure it aligns even better with the customer’s lifestyle. Because, if the customers don’t like the change, they won’t buy your product.
So, how do you know your changes will align with your customer’s world?
Making change customer-driven begins with reliable insights into who your customer is, and what they want in your product.
But how do you actually create customer insights? This article provides a comprehensive 11-step guide to building high quality, relevant customer insights. This guide is designed to produce a research project that takes around 10 weeks. That’s 10 weeks to go from a mere idea to all-singing-all-dancing customer insights.
Dirt is Good
The best examples of insight-driven change follow a rigorous methodology. This allows them to paint an accurate picture of the behaviours, emotions and culture of their customers. And changes at the organisational level come directly out of the insights.
Take the example of Persil’s advertising campaign “Dirt is Good”. Their research department carried out ethnographic research across the world to understand the narratives that parents create around washing. They discovered that parents connect the dirt and mess their children produce to freedom and creativity. This is much more positive than the “pristine-clean conformity” that other washing powders offer you.
By tapping into parental desires to nurture free, creative kids, Persil developed a marketing campaign that puts washing powder at the heart of a family story. And the numbers talk for themselves: Persil draws in more than $3bn globally and sales have consistently grown by 2% over the past 5 years.
Not Your Average How-to Guide for Customer Insights
So, back to the question at hand: how do you go about creating customer insights? If you google it, you will get an endless list of “how-to guides” and “5 simple steps” listicles.
Yet, ironically, very few of these actually tell you how to create insights. They tend to be filled with vague, uninformative steps like: “do research” or “get to know your customer”. Well, if you’ve just googled how to create customer insights, odds are you’re looking for a bit more information than “do research”. I’m guessing you would like to know how to actually do said research.
As such, below is not just a step-by-step guide to producing high quality insights from a 10-week research project. It also sheds light on how you actually carry out the different steps. By my count, there are 11 specific steps you need to follow if you want to produce valuable, relevant insights that you can use to improve your customer experience.
Of course, this is only one way of producing insights – namely the Danji way – and there are plenty other methods you could use. But follow this one and you can be sure to produce deep insights about your customers’ behaviours that will help drive innovation in the right direction.
Step 1: Identify the problem
Innovations need to address a problem – if things are going well, nothing should need to change. So, it’s very important to start your research by pinpointing the specific problem at hand. This could be lower sales, market disruption by a competitor, or a positive problem like entering a new market. Ultimately, your insights will bring you back to this problem and offer a solution. So, it is important to keep this in mind throughout – you may produce the most fascinating insights, but if they don’t help improve things, then what’s the point?
Step 2: Identify the Problem Group
Now that you know what you want to improve, you need to think about which customer segment relates to this problem. Identifying the right customer segment is very important because it allows you to build more accurate insights. Researching the right customer segment means your insights will lead to much more relevant product improvements and, in turn, more personalised customer experience.
Which customers do you need to target to make the changes happen most effectively? There are different ways of determining who your target segment are. And your choice of method should follow how your market is segmented. Are your customers divided into paid vs. non-paid users? Do you have different markets for each gender? Are millennials, baby boomers or gen z the problem group? Is it a specific country/culture profile?
If you know your market well, how to segment should be fairly straightforward. As such, I don’t want to go into any more detail on this step here. There has been plenty written about this and you can find useful, concise information on blogs such as this one by Mixpanel.
Step 3: Draft a Research Question (RQ)
This is where the meat of the research starts – it’s time to start asking (yourself) some questions. You need to come up with an initial research question that you want to answer. By research question, I don’t mean the questions you will ask your interviewees (they are interview questions). Rather, this is an overarching question that, once answered, will give you the insights you’ve been looking for.
This is a question you will ask about your target segment, in order to find out what the product or proposed change means to them. Remember, insights are useful knowledge about how your customer relates to your product. So the question should be designed to build a generalising picture of the customer and their relationship to the product. And what you ask will depend on which aspect of this relationship you want to understand – is it behavioural practices? Emotional connection to a product? Cultural significance of a product?
In a recent research project Danji AB did for a Swedish media outlet, our research question ended up being:
“What place do we (the media outlet) hold in people’s lives? How do people perceive our brand?”
They wanted to understand the declining consumption of local media among millennials. So, we set out to understand what local media means to this market segment. This research question isn’t so narrow that it guided us towards a predetermined answer, yet not so wide that it left us with no direction at all.
Step 4: Initial Observations and Interviews
Remember, your research question is currently only a draft. There’s no reason why it can’t change yet. In fact, it may be wise to change it. In the beginning you want do a sort of test research on a small sample of people to gauge if your question is anywhere near the mark. It may turn out that your question isn’t relevant to the customer segment at all and you have to revise it.
For this you may want to spend a day or two doing ethnographic observations on your research group. Spend time with them, get to know them and observe their practices. Depending on your need, you can vary the extent to which you engage with them or observe from a distance. You will start to notice patterns of behaviour that either relate well or not at all to your proposed research question. If you see a relation, then you know you might be on the right track with your question.
To confirm your initial observations, or to help you alter your question (if your observations don’t confirm the relevance of your RQ) you can carry out 2-3 in-depth interviews. Expect to talk to your participants for 20-30 minutes, asking them general questions about the topic and their relationship with it.
Gauge their answers. If they get visibly emotional about the topic or have a lot to say then it’s likely that you’re on to an important topic. If, on the other hand, they seem very disinterested and only give short answers, your topic probably isn’t that meaningful so you need to consider changing your RQ.
By now you should have a good idea of the relevance of your RQ. If your observations and interviews confirm your RQ you can move on to the bulk of your research knowing that this topic is in fact relevant and will lead to meaningful insights. Otherwise, you can change it to another topic that came up in your research. Time permitting, you could always go back and do more observations and interviews to explore this new topic in more detail and confirm its relevance.
Step 5: Interviews
Armed with a relevant RQ, you’re ready to do your interviews. There are plenty of different interview types and techniques out there, but I think the best one for this type of research is known as semi-structured interviewing. I won’t bore you with why it’s so good (check out my previous blog if this sort of stuff tickles your fancy) but they are nice and flexible, allowing you to capture the messiness that is the human mind.
With qualitative research you don’t need to carry out thousands of interviews. You probably won’t even need more than 10 in total. This is a tough one for most people to get their heads around – especially if they’re used to handling Big Data. But you’re not trying to get Big Data – you want the “Thick Data” that exposes the human behind the data. You’ll know when you’ve done enough interviews because you’ll hit what’s called “saturation point”– basically you keep hearing the same points over and over, you’re not getting any new information.
Expect each interview to last anywhere between 20 minutes and 1 hour depending on how big your topic is and how much the interviewee likes to talk! “Semi-structured” means you’ll want to have a few guiding questions (5-6 is a good number). You use these to direct the interviewee to talk about your topic, but it is very important not to be too rigid.
If you like, think of it as a conversation in which you’re trying to get to know the interviewee. You want the interviewee to take the lead and discuss the topic in their own words. But you need to make sure they don’t go too far off track. If, for example, you’re asking about smartphones and the interviewee starts talking about her grandma’s cat, you probably want to veer her back to the topic at hand.
And, of course, don’t forget to record the interviews! You don’t need a fancy voice recorder for this, the recorder app on your phone works perfectly well. But what you do need to do is get consent from your interviewees to use the recording for the project. You can write up a simple consent form and ask them to sign it (or simply adapt one you find on google like this one). As good practice, you should also ask for consent orally before starting the recording.
Step 6: Transcribe
If you’re anything like me, you will find this part extremely boring (although, I have a colleague who actually enjoys transcribing!) But, sadly, it is extremely useful, so you have to do it. Besides the obvious advantage of having accessible quotes to read, the act of transcribing itself is invaluable. You hear the interview from a different perspective so you actually pick up different data points that you may have missed when immersed in the process of interviewing.
Essentially, you need to write out the interview as you hear it. You stick in a pair of headphones, play back to the recording and write down what you hear. You’ll need to give yourself plenty of time for this task: expect 1 hour of recording to take 4-5 hours to transcribe! Glad you’re only doing 10 interviews now, and not thousands!?
There are free web tools that can help you transcribe more easily. I like to use otranscribe.com because it lets you slow down the speed of the recording. So, once you’re in a flow you can almost type as quickly as you hear the recording.
I strongly recommend that you avoid outsourcing this task – however tempting that may be. Nobody can like every part of their job all the time. Just get yourself comfortable and prepare to be sat in front of your screen for a good chunk of time.
Step 7: Categorise the Data
Now that you have your raw data written down and easily accessible, it is time to transform those pages and pages of disorganised data into useful, concise summaries. This means systematically going through the data to identify all the relevant patterns of behaviour, ideas, emotions and concepts.
For something to be considered a pattern, it should come up numerous times in the data i.e. discussed by several of your interviewees. And to be relevant, it should relate to your original research question. For instance, if 7/10 interviewees discuss gender-related issues, you can be sure this is a relevant pattern. In contrast, if 7/10 mention how nice the weather is today, it’s probably not so relevant – they’re just engaging in a bit of small talk!
Everyone has their own technique for categorising, and how you do it should be based on your own preference – you can use sticky notes, colour-coded pens, “meaning-unit” flows, codable unit spreadsheets and so on. I’m quite old fashioned and opt for a pen and pad.
First, I read through the data and jot down sections of text/quotes that relate to themes I have noticed already. By this point you will know the data quite well, so you’ll already be aware of some patterns. And I make sure to accompany it with a note of where I found it (e.g. interview 3, page 5).
Then I re-read the data a few more times. This allows me to get a sense of the texts as a whole. And verify that I understand the meaning of the individual quotes in the wider context.
Also, as you re-read your data, further new themes will emerge. Re-reading allows you to go back through the entire text and see if you’ve missed any more key quotes relevant to these new themes.
Gradually, you will build up a solid base of quotes – or manifest content, as they’re technically called – that all fit into certain themes. Although, if you’re following my pen and paper method, they will still be all jumbled up and disorganised. You need to organise them into some sort of order. If you used a sticky note or spreadsheet system, you wouldn’t have this problem, of course.
Putting them into a spreadsheet is quite good for this because it’s easy to move them around and edit them as you go. You could also use sticky notes, a whiteboard, or list them in your notebook. It doesn’t really matter how you do it, so long as the themes are clearly defined and make sense.
Step 8: Interpret the Themes (Intuition and theory)
With your data all collated into thematic categories, you can start to explore the relationship between the interviewees (i.e. your customers) and the product. It should start to become clear what the customers think and feel about the product, how it fits into their cultural context and how it makes them behave.
Some of these themes will appear to you fairly clearly and feel like common sense. This is normally fine, it shows that you understand the customer-product relationship well. For many researchers, it is enough to stop the analysis here and go no deeper. However, if you want to produce genuinely high-quality insights, I argue that you need to theorise your themes.
By this I mean you need to understand the very essence of the customer-product relationship and generalise how it works in every context. This is what theory does: it provides a framework for generalising every (or, at least most) manifestation of a relationship. This is invaluable if you want to predict how changes in a product would impact the customer or pinpoint the significant factor in a chain of effects that lead to, say, a reduction in customer satisfaction.
Because we’re dealing with human behaviour, you will want to adopt a social science theory. But which one? Well, your choice of theory depends on your field of expertise, your research needs and the data itself.
I’m a trained anthropologist and I tend to rely on anthropological and sociological theories because they are useful for explaining patterns of group behaviour and human-product relations. But at Danji we also have psychologists, socionomists and literary critics working for us who each look at a problem in a slightly different way.
Within the research, the choice of theory should be driven primarily by the data itself. If, for instance, power struggles are a dominant theme in your data, you may want to draw on Foucault’s theory of power to interpret the data. To do this, you obviously need to read his work on power (or at least a summary of his work by another academic) then imagine you’re looking at the data through his eyes. What would he say about it? It’s safe to say that Foucault probably thought differently to you and I. Looking through his eyes will give you a different perspective on the relationship between the customer and the product/brand.
But don’t cast away your own opinion if it doesn’t coincide with the theory – theorists may have a good grasp on how the world works generally but they don’t have the data in front of them. You do.
And the wonderful thing about humans is they often challenge the norm. Just as we think we’ve got a theory to explain them, they change. This is an opportunity to engage even deeper with the data and come up with your own explanation that, whilst supported by your original theory, offers a more comprehensive insight into this specific situation.
Step 9: Validate
Qualitative research is not a simple linear process from A-to-Z. Rather, it is what they call iterative. You should be constantly going back to the text and checking to see if your ideas make sense. Are your themes definitely relevant? Is there really a pattern? Does your explanation make sense in context?
This is an extremely important discipline to build into your research. It’s a fail-safe way of ensuring your analysis is sound and your insights are valid. Ideally, you’ll be doing this constantly as you go. But, for now, let’s make it its own distinct step to make sure it definitely gets done.
It also takes honesty. It’s very easy to convince yourself that your analysis is correct after all this hard work – nobody else will be going through the data with a fine toothcomb so you could probably get away with submitting it as it is, right? Wrong. You need to be brutally honest with yourself and not be afraid to put your hands up and say “this doesn’t quite fit, let me have another look”.
At Danji, we analyse data in small teams of 4-5 researchers. One of the major benefits of this is that everybody reads the data, making it harder to misinterpret and making the validation process more naturally occurring.
Step 10: Create a Customer Storyboard
Confident that you have a water-tight analysis of the data? Great, now you need to package it in a compelling way that clearly explains what you’ve found. Be it an external client, your own boss or the C-suite, people are eager to hear your discoveries.
One of the many things that makes Danji unique is how we package our insights. In short, we tell stories. We create storyboards that narrate the customers’ own stories of their relationship with a product or brand.
Storytelling is a powerful way of presenting your insights because humans are natural storytellers – and listeners. We’re wired to listen to stories and even interpret data into stories in our heads. This makes storyboards a very effective way of packaging your results so your audience will understand them and be able to use them easily.
A storyboard basically represents a sequence of events that your customer goes through. Imagine a film storyboard with your customer as the central character. Each frame represents the key practices and emotions they go through with regards your product. This story extends beyond the standard brand touchpoints to include pre-purchase and post-purchase.
If you know your data well (and you should by now, if you’ve read and validated it enough times) you’ll be able to explain the events in each “frame”. And, more importantly, you’ll be able to map the key emotional responses and the relevant cultural context for each one.
If you or a colleague is a decent illustrator, you could even draw the storyboard just like a film storyboard to capture the emotional experience of your customer. This approach has become a driving force for companies like Airbnb. Alternatively, you could create a more traditional story using words alone, complementing your text with poignant quotes from your data.
Step 11: Answer the Research Question
It would be reasonable to think that your job is done now. You’ve done all the research and you’ve package them into an enlightening customer storyboard. But you haven’t actually answered your question yet. You started with a problem that needed solving, and you finished with a story.
As illuminating as your storyboard is, it won’t directly answer the initial question. You need to explicitly state how your insights relate to the problem at hand. You don’t want all your hard work to be ignored because your client doesn’t see how it answers their question.
To ensure your client gets the full potential out of your insights, they need to answer the original question. This is the primary way in which they will drive strategy and sustainable change. So, make sure you give them a definitive solution to the original problem. Use your customer storyboard to reinforce how you got to the answer and highlight how it implicates the customer.
And if you need to, you can use your theory as a framework for predicting the outcomes of different solutions. But the most important thing to do is actually give them the solution!
And We’re Done!
And there you have it. An 11-step guide to producing high quality, usable customer insights based on qualitative research. All that’s left now is to try it: go explore your customer’s world. And come back with some well-informed, illuminating insights for your company, client or department.
And let us hear about your experience. We would love to know about your research process. What did you find helpful? Or did you, perhaps, disagree with anything in the guide? We want to hear your thoughts.