Review mining: How to use online reviews for audience research
If you’re not doing audience research, you’re late to the party. In a perfect world, every company should have a comprehensive and ongoing customer research strategy.
But what if you just want to get a lightning-fast pulse on what your audience is thinking?
It is the process of digging into online reviews in search of qualitative data.
If you analyze enough third party reviews, you can walk away with a new competitive advantage.
A review left on a website regarding a customer product experience is, in other words, user feedback.
Whether you set out to analyze reviews of a competing product or your own product, you will be able to synthesize those responses into insight that will grant you a competitive advantage.
Regularly and proactively looking at them is a quick and scrappy way to conduct audience research. And qualitative research is becoming more important than ever.
To actually gain a competitive edge you need to go beyond just what your customers, prospects, and target audiences are doing in the quantitative.
You need to understand their ‘why,’ and there's no better way to do that through different types of audience research.
Audience research in general is not easy; it typically takes a huge amount of time and energy.
It’s no wonder many brands feel overwhelmed and seek outside help.
You often may struggle figuring where to start. Some companies simply don't have enough customers yet because they're in startup phase, or maybe they need to go to market at a faster velocity than interviews or surveys may allow you to do.
While review mining should never be a full substitute for deep audience research, it can certainly be helpful.
The data is always available, and the time needed to reach insights is relatively quick.
Often, they represent the final decision-making factor considered before new customers make a purchase.
Reviews have become more and more popular and are consumers’ go-to resource when it comes to making any kind of purchasing decision, especially within the course of the last year.
Everything has just been surging at an extreme rate, leaving more customers there to reply and look at reviews, testimonials, and recommendations online more than they ever have.
If you think back to your own recent online purchases, you might quickly realize that you are included in that percentage.
Products with no or limited reviews can quickly cause a sense of skepticism about a product.
These days, everything from your next book purchase to your next Airbnb rental includes reading the feedback of others to some degree.
When starting the mining process, your first step is to think through the questions your target audience has.
The site that you choose to analyze is less important.
You want to ask yourself questions like:
Once you've answered those questions, it’s time to set up your framework for gathering data.
Because there is so much data out there, it’s easy to get overwhelmed. To prevent data overload, it’s important to have a framework in place.
I'm a fan of using the Jobs-to-be-Done framework to analyze the data.
The Jobs-to-be-Done methodology is based on the theory that everyone is motivated by purpose within their own lives.
We've all heard this before: people don't buy features, they buy a better version of themselves and that dreamer state they're looking to get to.
I do find the customer job statement a particularly useful framework when you're uncovering voice of customer data, because it really helps give you that guideline to follow.
Following the customer job statement allows you to ignore things like demographics and instead focus on customer motives.
What solution are they looking for, and why?
The goal is to begin to understand when the customer has struggling moments, what motivates them to look for a solution, and what desired outcome they hope to achieve.
Next, let's see how to put this framework into action.
You can download the template that I use to see this process in more detail.
In 2019, I and my friend Georgiana Laudi of Forget The Funnel decided to do an experiment.
We wanted to see how the result of review mining would compare to Sprout Social’s more robust audience research program.
Fascinatingly, there was a 90% correlation between the results that we uncovered.
The results in mining the data for Sprout Social showed us a 90% accuracy in all of the research that we did that took us nine months to complete on our own interviews and surveys.
Again, I would never replace that because there's so much more that we got out of that full audience research process that is continuous and ongoing.
But, this did give us a really good framework to start to move faster and more iteratively.
I chose to look at reviews on G2, Capterra, and Amazon. My goal was to look at how people talked about Sprout Social or its competitors.
If you don't have customers, that's okay. Chances are you do have a competitor, or there's probably a solution in the market that you're looking to solve for.
So look for where your target audience is, and that's where you should start.
When you are browsing reviews, you can move quickly. Thirty seconds or less spent per one should do it.
What you are looking for is information that pops out at you.
It’s important to have a fairly large bank of reviews to go through, and not every one will contain actionable information.
Record whatever you can, and then move on to the next one.
As far as what “recording” information looks like, my process was fairly simple.
I recorded each response on its own line in a spreadsheet. And each line had columns where I could record things like motivators and desired outcomes.
You'll often find more than one motivator or more than one struggle. And you want to make sure to capture those.
Once you've mined through enough insights, you can start digging into how to read those to get actionable next steps.
On average, I aim to gather 50-100 responses before I begin analyzing the data. Too little and you won’t have enough data to be confident about trends you find.
Too many and you’ll get overwhelmed. A general rule is that this takes roughly about two hours, give or take, depending on how familiar you are with the process.
Once you’ve logged insights for those 50-100 reviews, it’s time to parse them for trends.
In this step, you are mentally putting customer statements into buckets, and translating that into the spreadsheet in the form of new columns that are titled appropriately.
In my example, I created columns like “Struggle Situation” and placed underneath it themes I encountered such as: save time, build awareness, and improve social media skills.
Each individual entry is boiled down into a “theme” as described above.
And when you’re done with all of the entries, the dominant themes should clearly emerge.
Some entries might contain more than one theme, and that’s ok. At the end, you can highlight the themes you recorded and turn them into a visual chart to see what stands out.
It's as simple as just highlighting those and creating a chart. And then as you dig into that, you can look at the closer trends.
In my example, the trends that emerged were improving workshop workflows as a clear struggle, and saving time as a desired outcome.
We’ll now follow my example to its conclusion.
Sprout Social’s mining process provided actionable insights that enabled us to optimize their onboarding program.
In addition to allowing the team to pull out key themes, the process also provided marketing copy inspiration.
It’s no secret that the voice of your customer can be translated literally into great copy. Who better to talk about and verbalize your message than the customers who are using it every day?
That research can truly turn into remarkable copy and execution strategies that do convert. I just see this time and time again, bringing incredible impact to our results.
In our mined data, I and my team came across quotes describing social media as frightening and overwhelming.
Others talked about the newfound ease of connecting and engaging using Sprout Social.
We translated those copy sound bites into a new welcome email series.
Compared to the old onboarding email series, it was less design-heavy and tied more securely to the Jobs to be Done research — a desire to save time and improve workflows.
The result in that change was a 6x increase in conversion from customers engaging with just that simple email series.
But we also didn't stop there. We started to weave the same copy and the same qualitative finding into emails coming direct to our success team, to improve customers engaging with us whether that was through human-led or marketing-led onboarding.
With measurable success under their belts, I and the Sprout Social team realized that it wasn’t a one-and-done process.
Now, we conduct it annually as part of their ongoing audience research program.
Anytime we're running a campaign where we just don't have time to do that research, we'll then come back and say okay, can we mine some data to help us get there and continue to cross-compare that against everything else we're doing?
Review mining is a helpful addition to your audience research toolkit.
It can be applied to any channel you can think of — from paid ads to lead-gen website copy to onboarding emails.
It's all about finding those key trends and customer insights that are already out there, waiting to be found.
Out now: Watch our free B2B messaging course and learn all the techniques (from basic to advanced) to create messaging that resonates with your target customers.