We analyzed 22,641 verbatims from a retail NPS survey in 2 minutes. This is what we found.

In the last two years, retailers have been forced to work harder than ever at meeting ever-changing CX requirements to stay ahead of competitors and remain relevant.

So as a retail brand, how do you stay top of mind and ahead of your competitors without the right technology helping to identify what customers really want?

And how do you drive towards the optimal customer strategy without this information?

We decided to put Kapiche to the test with 22,641 retail NPS surveys to see what insights we could uncover in the customer feedback. Importantly, we'll be doing this without any existing domain knowledge of the store.

The depth of insight we're able to uncover would have a significant impact on CX for this store, leading to tangible increases in NPS and more importantly, revenue.

Follow along and see how we go identifying NPS drivers, measuring the impact of NPS on themes and reporting actionable insights!

Understanding 22,641 verbatims from a retail NPS survey (in under 2 minutes)

I don’t have the time to spend weeks manually coding this data, extracting insights and producing a report on NPS drivers.

To save time, we've used Kapiche to code 22,641 verbatims (this was complete in under 2 mins).

We start with a 30,000 ft overview across all survey responses using the storyboard visualization (see below).

Size matters

“Size” and “Fit” of the clothing stood out to me so I started there. If you look closely you can see numbers such as 2,4,6 and 8 spoken about in the context of “lbs”. There’s another relationship between “usually” and “reviews” which also caught my eye. What's going on?

As it turns out there's an interesting story.

The context network above actually changed once "reviews" were added. It appears shoppers are checking the reviews of other customers before making purchase decisions.

When reading individual responses, I noticed an interesting behavior trend. Customer reviews were making sizing suggestions based on their own purchase experience (and shoppers were listening!).

• "As the other reviewers stated.."
• "I agree with the other reviews, this top runs large so size down."
• "Got this in the red, like previous reviewer mentioned - it does fit snug"

This was repeated over and over again. As you can see in the context network visual for "reviews", words such as "listened", "decided" and "glad" are all spoken about in relation to "reviews" and "size and fit". It's here we're able to merge a high level insight that's automatically identified in the data for us and then drill down deep, right to the individual response level for that human connection.

Clothing appearance has an impact

We then looked at overall NPS and noticed there were two clear dips (April and July). Why did NPS drop in these months?

As you can clearly see in the above visualizations, the April drop in NPS was driven by reviews talking about clothing appearance and store inventory levels. Curious!

Where it gets interesting is we have a structured field in this dataset for spend category (High/Medium/Low).

The lowest NPS is found with "Price" (7.48% of verbatims, NPS +7), but "Clothing appearance" is represented more in the dataset (31.35% of verbatims, NPS +19.83). There's something customers clearly don't like about the the appearance of clothing they're buying from this store. Going back to the spend category mentioned previously, it makes sense to see what the story might be for customers in the high spend category. Does anything change?

There are a few movements. High spend customers have lower NPS in reviews talking about "Discount / Deals", "Clothing color" and the "Summer range" but slightly higher NPS for "Dresses" (but that theme still has relatively low NPS and should get looked at).

We know this theme is a negative driver of NPS in April and continued to be a negative driver into May. As we dig even deeper let's look more closely at what customers are specifically saying about the appearance of clothing.

To figure out what's going, I examined a slice of the data for the clothing appearance theme from March 1 to May 31, 2021. The impact on NPS is -6.88. In fact, this theme was having the most impact on overall NPS for this slice in time!

After reading some of the reviews for this theme, we discovered some customers were returning oversized or poorly fitting clothes because it made them look pregnant/frumpy. It's also interesting to see them referencing reviews from previous customers, lamenting not having listened to those reviews.

Oversized, frumpy clothing seems to be an underlying driver here. We then checked to confirm if this was overrepresented in a department or product class. We were fortunate to have structured data to work with that was associated with each review. This allows us to go even deeper with the analysis to tell relevant stories to these department leaders.

In the above visual you can see Dresses, Knitwear and Sweaters are over represented in this time period amongst NPS detractors. The numbers aren't huge but they do suggest there's something going on with these categories.

The story we've uncovered here is that customers return oversized clothes that make them look pregnant/frumpy. We've got a few individual reviews we can highlight to demonstrate the pain experienced by customers.

Refunds and returns are driving potential customers away

With this in mind, we took a step back and examined what else is dragging NPS down.

Here’s the full impact on NPS for this retail store:

As you can see with the above visualization, the elephant in the room are reviews talking about "refunds" and "returns".

There's no need to look at sentiment to empathize with customer pain. These reviews have an NPS of -38.96. It's also noteworthy that this theme is dragging overall NPS down -6.38.

That’s a whopping 3X higher than “price” (-2.14)!

As more organizations seek to measure the ROI of CX by calculating the dollar value of each NPS point, this could potentially have a massive impact on revenue and spend!

For the curious amongst us, I’ve included a rather large context network around “returns”. No doubt there's quite a bit you'd discover if you went digging. At a glance it appears people are returning low quality, cheap items, talking about how their purchases make them look unflattering, large/frumpy and looking to exchange clothing for smaller sizes.

These are the key insights we discovered after a few minutes:

• Of these reviews talking about getting a refund or returning, the Intimates and Tops product categories were negative drivers on NPS in the last month. Dresses, Bottoms and Jackets were positive drivers.
• Customers returning dresses to this store were a major negative driver of NPS between March (-19 NPS) and April (-41 NPS).
• When we look at specific item categories in greater detail, it was also apparent in the last month NPS for both Sweaters and Swimwear nose dived. Sweaters went from +6 to -71 while Swimwear went from -30 to -100. I've included a visualization to show the story.

What's next

Here are the main stories we found after an hour of analysis:

• Clothing sizes are too large which is making customers feel "pregnant" / "frumpy". This is a leading driver behind why items are returned.
• The above story is also resulting in customers leaving advice in their review to other customers to go down one or two sizes to avoid the same problem. Shoppers are paying attention to this advice and modifying their purchase behavior accordingly. Reviews talking about "refunds" and "returns" are driving overall NPS down -6.38.
• NPS went down in April and this was driven by reviews talking about clothing appearance and store inventory levels.
• More recently, two clothing item categories driving negative NPS in the last month were Swimwear and Sweaters. NPS detractor reviews for this slice in time for these item categories were also talking about sizing issues.
• High spend customers have lower NPS in reviews talking about "Discount / Deals", "Clothing color" and the "Summer range" but slightly higher NPS for "Dresses".
  

That’s all for now. Would you like to see Kapiche in action with your own data?