I’ve had the opportunity to analyze many net promoter scores, from internal employee surveys, to standard product ones, as well as some focusing on very narrow aspects of usage, onboarding, or support. I shared some of my experiences in a recent post on how to improve your net promoter score, but I thought I’d take a step back and share some tips on how to analyze an NPS.
1. Getting Demographic Data
Once of the best things about using the NPS as a survey is that you can easily segment it by demographics and provide nice summaries. For example, you could report things like:
- Employee NPS by department or tenure
- Top 10 reasons users with a specific role like your software
- Customer NPS by industry, company size, etc.
- Customer satisfaction over time by cohort
By collecting this information you can segment the responses in meaningful ways and answer questions like “How do I improve selling into industry X” or “What features should I promote to our B2C customers” and so on.
The trick is collecting the data! The easiest way is to integrate your survey software with a CRM. With this approach, you can merge the data and have access to all the demographics you usually track. However, this isn’t always an option as you may want to promote the survey as being anonymous. In that case, think about the three or four most important ways you want to segment the data and add those questions to the survey. Just make sure you put the questions at the end or on a second page. That way, respondents will not be influenced by the questions before they answer the NPS. Also, make sure you don’t ask for information that can be used to ID someone if the survey is supposed to be anonymous.
2. Tagging Comments
Now you need to take all the text responses and tag them. This is one of the most important and time-consuming steps. I’ve literally spent an entire day or two tagging comments from just 300 responses and I firmly believe it is time well spent.
The issue is that with a text response, you usually can’t just bucket them. For example, you might get:
“I love how easy your software is to set up and use. Your support team is always fast to respond. Couldn’t give you a 9 or 10 because the product is a little slow.”
If you get a response like that, how do you bucket it? Does it go in the “Little slow,” “loves support,” or “Easy to use” bucket? The correct answer is all 3! But you can’t do that if you rely on buckets. Instead, you would simply tag this comment with 3 tags (little slow, loves support, ease of use).
Seems simple right? Well, keep in mind you could have hundreds of these responses! So, as you read through them, if you’re not careful, you could have dozens or more tags. Therefore, I recommend you through the comments once and just jot down the tags that come to mind. Also, keep a rough count of how many comments per tag. When done, you’ll have a big list of tags and a rough idea of how popular they are.
Now take that list and try to combine tags. Look for similar tags and keep one but get rid of the duplicates. You can also modify tags some, for example, if you see “Ease of use” and “easy to set-up” then create one “Easy to use and set-up” tag as the comments are related. Lastly, consider dropping some tags altogether if they don’t appear very popular.
I encourage you to then use a program like Excel to store the comments and tags, as well as the survey response and demographics. For example, you could have something that looks like:
In the image, “Good Support,” “Poor Docs,” and “Slow” are the tags. A “1” means that comment is tagged with that tag, a “0” means it’s not. The reason you’ll want to store it that way is for easy analysis later. For example, I could write:
That would sum up how many B2B companies (B2:B5 would be looking at the company type column) believe this product has good support (E2:E5 is the good support column). That is just a simple example, but a program like Excel makes it very easy to quickly slice the data. As you can imagine, once you start to analyze comments like this combined with demographic data you can now understand top concerns by demographic data, compare demographics with each other, prioritize what you want to work on, and so on.