Advice on NPS program implementation 

Published by Ulf Johansson on

I recently shared the following advice on how to set up a NPS program with a friend at a B2B SaaS startup. I believe this advice is broadly relevant to most companies thinking about implementing your first VOC feedback channel.

i) Overview and implementation guides

Here are some useful articles on NPS implementation:

  • Qualtrics has a good introduction of NPS here.
  • Userpilot has a good overview on NPS for SaaS businesses here with some implementation advice.
  • This InMoment blog has some interesting advice for how to make NPS useful (not that I agree with everything in it). 

ii) Which survey tool to choose 

There are conservatively measured a billion different survey tools on the market, each with slightly different pros and cons. I don’t have a strong recommendation for what system will be optimal for your company, so I’d suggest you spend a bit of time exploring the market and evaluating options:

  1. Before looking at specific tools, decide on what functionality you need out of your NPS system. For example:
    1. Do you want a bespoke NPS tool or a general survey tool? A bespoke NPS tool can be simpler and cheaper than a general survey tool, but has much narrower use cases.
    2. How do you want to distribute the NPS survey? Email? In a mobile or web app? In a Unity app (yeah right, good luck!)? Via Intercom? On your website? All of the above?
    3. What type of NPS program is it? Is it a transactional NPS (i.e. survey customers directly after they make a purchase?) or a relationship NPS (survey customers on a regular basis about their overall impression of the product)? Different functionality required.
    4. What aspects of the NPS cycle should the tool manage? The full NPS process is a lot more than just the survey itself – it includes survey scheduling, user enrichment and segmentation, fielding, analysis of results (including open text analysis), communicating and sharing results with stakeholders, creating change tickets, and closing the loop with respondents. Each of these steps can be done automated or manually, and different tools offer different degrees of coverage for the full NPS cycle. 
    5. What other systems do you need to integrate with? Zendesk, Salesforce, JIRA, Slack? What are the need-to-haves, what are nice-to-haves? 
    6. What’s the process for enriching NPS data with customer attributes? You’ll want to add data such as industry, usage behavior, etc to the feedback to be able to segment your NPS responses. Understand how easy it is to append attributes to the responses based on what data sources you need the tool to pull from. 
    7. Are you doing a MVP test or a long-term investment? Are you okay to rip out and replace the first solution in six months time? Do you want a startup-focused tool that you can grow with for the next 1-3 years? Or do you want to invest in an enterprise tool that is overkill to start but lets you grow with it for the long run?
  2. Create an evaluation spreadsheet with your requirements, and the importance of each requirement (Required, High value, Low value)
  3. Look through some options in G2 and blogs.
    1. Here, here and here are some blog posts outlining some common tools and what to look for (they all have agendas, so take specific recommendations with a grain of salt).
    2. My thoughts on tools:
      1. Your existing CRM system might suffice if you plan to just distribute surveys via email. We used a hard-coded email template for NPS that we distributed through Eloqua in Unity – it worked okay but the lack of full NPS lifecycle management meant we had to do a lot of manual work. If I did it again, I would have used something like InMoment for NPS. 
      2. I’ve used Alchemer a lot lately for general ad-hoc survey work. I really like the tool and price point. I haven’t used it specifically for NPS or scaled it to a full company implementation, so not sure how well it works for that.
      3. We selected InMoment as our CSAT tool at Unity, primarily due to their integration possibilities allowing for fielding in the Unity Hub. They’re not the cheapest, and are narrowly focused on CSAT/NPS questions, but they are a solid option. 
      4. I see SurveySparrow pop up a lot in lists – I haven’t used it myself but they seem like a potential option.
      5. Avoid Qualtrics – expensive enterprise tool that is overkill for you. Delighted by Qualtrics is a cheaper alternative, might be worth looking into, but I haven’t used it myself.
      6. Avoid SurveyMonkey – consumer-grade, not great for B2B use, antiquated.
      7. A startup founder friend really likes Jotform for building forms and simple surveys for her startup. It’s not a dedicated survey tool, but might be a good option if you don’t want to immediately invest in a single-purpose survey tool.
      8. If you have a web testing tool like Hotjar already implemented, check if their survey functionality is enough for your initial needs.

c) Other considerations 

Make sure you choose the right tracking metric

There are three metrics commonly used for customer experience tracking – NPS, CSAT and CES. These offer complementary views of the customer journey, and eventually you might track all three. When starting out, make sure you track the one that’s most valuable to you, rather than just go with NPS because it’s the ‘default’.

  • NPS tracks overall customer brand sentiment and loyalty. It gives you the most holistic view of customer satisfaction, but provides the least actionable product feedback.
  • CSAT (Customer Satisfaction, description) tracks customer experience specifically of your product, stripping out things like price, competitive dynamics, sales and marketing impact, etc. It’s the most useful tool for rapid product improvement, but the narrow focus means CSAT score doesn’t correlate as well with business outcome.
  • CES (Customer Effort Score, description) tracks specifically how challenging it is to use your product. It’s primarily a UX tool and even narrower than CSAT; it can be super useful if your challenge is that your customers fail to use your product because it’s too complex (hello Unity Editor).

Have a plan for sampling

I expect you will run into the same issue that plagues most B2B customer feedback programs – a lack of eligible customers to sample. On one hand, you want enough responses to get reliable feedback, and on the other hand, you need to avoid repeatedly surveying the same customers in order to avoid survey exhaustion. 

Your sampling rules (i.e how you determine who you send the NPS survey to when) will become very important to the long-term success of your NPS program. Decide on these rules before you start implementing any tools, to ensure the tool supports your needs.

Considerations:

  • Survey frequency – to avoid survey exhaustion over the long run, I avoid asking the same respondent the NPS survey more than once per 12 months (assuming it’s a relationship NPS; transactional NPS can be asked more frequently). 
  • Sample size – there isn’t a hard-and-fast rule for how big you want your sample, but NPS is a pretty blunt tool so I’d be wary of any reported NPS score based on less than 100 responses, and preferably you’d want 300+. 
  • Segmentation – hand-in hand with sample size. Do you want to segment your NPS responses by e.g. customer size, industry, etc? Each segment then needs sufficient responses for each wave. Make sure you have enough customers in the segment not just for the first time, but for every subsequent survey wave.
  • Sample tracking – you need a system to keep track of who you’ve surveyed when, so you can avoid over-sampling the same users. This can be done manually in a spreadsheet, but an automated tool helps a lot. 
  • Activity-based vs Time-based sampling – you can either link NPS surveys to go out e.g. 12 weeks after a customer signs up, or you can batch send out surveys e.g. the first of each quarter. There are pros and cons of both approaches, but you need to figure out what’s best suited for you. Either way, you’ll probably want to batch your analysis to ensure you have enough responses.  

Open-text analysis is very time consuming but very useful

The most (only?) actionable insights in the NPS survey comes from the open-ended responses, but it consumes 90% of your analysis time. 

I haven’t found any magical shortcuts – I have yet to encounter a survey tool that provides great open text analysis, though I anticipate that will change dramatically in the next 1-2 years with the advent of GTP-based analysis tools – but there’s still value in tools that let you tag and organize open-text feedback in the tool rather than just spit out an Excel sheet with the raw text.

Breadth of integrations is very valuable in a survey tool for future growth

An important factor when evaluating survey tools is integration capabilities, both in terms of ability to field and in terms of integrating with assorted data tools to pipe results to other systems.

You might not immediately plan to send specific customer responses to e.g. your customer success team, but a tool having the functionality to integrate with for example Zendesk (if that’s what you’re using) gives you future room to grow the impact of your NPS program.

The inability to do this was a key failing in how Unity’s NPS program was set up, and something I focused a lot on when rolling out our CSAT program.

NPS is easy to roll out, hard to get right

Rolling out a NPS program is pretty straightforward, but the iceberg metaphor that 90% of the effort is under the surface is entirely appropriate. It’s easy for NPS to just become a metric you CMO glances at for 5 minutes once per quarter, ticking the box of “tracking customers sentiment” without actually delivering real benefits.

Considerations: 

  • Understand that you’re committing a resource for perpetuity – a NPS program is not a set-and-forget thing, it needs to be a key component of someone’s role (say 10% of a role; I usually assigned NPS to the most junior person on my team as a great opportunity for them to directly own a exec-facing project). 
  • Ensure you have product leadership buy-in if you want changes – there needs to be a concrete plan from CPO/CTO to use NPS feedback as input for product changes, and resources allocated to analyze feedback and add change requests to product roadmap. If you don’t have enthusiastic CTO buy-in, expect NPS to just be a tracking metric but not have any real impact.
  • Make results turnkey for stakeholders – make results available where your stakeholders live (e.g. inside product dashboards, quarterly email reports, etc). Don’t expect a standalone NPS dashboard that requires a separate login to be used much at all. Then it just becomes a checkbox ticking exercise.  
  • Closing the loop with respondents is powerful but time consuming – ideally, you want to be able to pipe feedback from individual respondents to their respective customer success managers, so they can follow up 1:1 on what changes you’re making. If you can scale this, it’s an incredibly powerful B2B retention tool and something we started to implement for CSAT at Unity. However, it’s also a big time commitment, so perhaps not something to start with.

Be aware of what’s not captured in NPS

All NPS feedback comes from active, engaged customers. This means NPS offers only a partial perspective, missing out on e.g. churned customer feedback and non-customer or competitor customer input. Plan to capture this feedback through other channels, and make sure any decisions based on NPS take these limitations into account. 

Photo by Towfiqu barbhuiya

Categories: Voice of Customer