Beyond the Surface: What SaaS Companies Miss About Net Promoter Score
Primary Venture Partners
Top founders start here. Alloy, Alma, Chief, Electric, K Health, Vestwell, & more. Forbes Midas List; Insider Seed 100.
Net revenue retention (NRR) is widely recognized as an important north star metric for assessing sustainable SaaS growth because retention is always the cheapest form of growth: A company with 125% NRR could never close another new customer again and still grow 25% year-over-year, which is pretty incredible. Successful SaaS companies are obsessed with what I call the “sticky drivers” of NRR; they unpack the leading indicators of NRR and tirelessly drive customers toward those behaviors.
Our industry emphasizes financially-oriented metrics such as NRR, Rule of 40, and CAC payback for good reason, but it admittedly surprises me that net promoter score (NPS) often fails to get its due. Sure, NPS has its limitations—it is a lagging metric and may or may not have an actual statistical correlation to NRR, etc.—but at the end of the day, companies compete on reputation, and NPS thereby has implications for understanding long-term growth potential and durability. Promoters drive referrals, which boast higher (and faster!) win rates and, by extension, improve CAC payback. Remember that detractors are 10x louder in the market than promoters! Editor’s note: There’s a reason why VCs all share energy about backing companies competing against incumbents with low NPS.
A few weeks ago I was a guest on The GTM Podcast to discuss the turnaround playbook we deployed at Sailthru nearly a decade ago now (spoiler alert: aligning our team around NPS was a central theme!). Given the positive feedback I received, I was compelled to refresh some older writing on the importance of NPS.
A refresher on NPS
Net promoter score is derived from one question and one question only: “How likely are you to recommend {Product X} to a friend or colleague?” Respondents are given a scale of 0-10; 9s and 10s are promoters, 7s and 8s are passives, and anyone 6 and below is a detractor. NPS is then calculated by subtracting the % of detractors from the % of promoters, meaning the spectrum of possible NPS scores ranges from -100 to +100.
“The” NPS question is insufficient on its own
There are a few reasons why people argue the merits of NPS. The grading scale is tough: You could receive all 8s and have an NPS of 0 or worse, all 6s and an NPS of -100. More importantly, NPS as an absolute number is frankly not helpful, as it lacks context. Does your negative NPS mask amazing customer support that is stymied by a buggy, hard-to-use product (or vice versa)? On a standalone basis, NPS often masks both positive and negative nuances influencing the score, so it is important to really analyze the why behind the results. Verbatim write-in comments from customers offer helpful additional details, but a strong customer insights program poses questions beyond the NPS question.
In the example of the buggy, hard-to-use product and negative NPS, perhaps customers are actually over the moon about Company X’s customer support; that context is lost unless Company X proactively seeks it out. In addition to the NPS question, businesses should solicit specific customer satisfaction ratings (“CSAT,” often on a 1–5 scale) for dimensions such as:
Customer success is a function of your customers’ ability to achieve their desired outcomes, not yours, so it is also important to ask how you are delivering against their objectives: “Thinking about your experience with Company X’s ability to help you drive results for your organization, how would you rate your satisfaction for each of the following: (and then get specific).” If your software promises QA testing automation, you would ask customers to rate you on dimensions such as, “Makes it easy for me to automate my QA testing” and “Helps decrease total support tickets.” While I was at Sailthru (personalization software), we asked for feedback on outcomes like their ability to personalize and the resulting improvements in revenue lift and churn reduction.
There is also merit in posing bonus questions to tease out feedback from different points of the customer lifecycle as well as general sentiment toward your brand. You can ask customers to rate statements such as “Product X is easy to implement,” “Company X is easy to buy from” and so forth.
Dive beneath surface metrics
Do your ratings vary by stakeholder (e.g. day-to-day users rate you differently from your economic buyers)? By industry? Contract size? When scores lag for economic buyers, consider standardizing or increasing the cadence at which your team is communicating results back to the executive sponsors. If NPS is weaker for smaller contracts (perhaps a proxy for client resourcing), be intellectually honest and ask yourself if the product is too complex for small teams; if so, is there a potential upsell support service to offer as a result?
It is both commonplace and helpful to slice and dice NPS vis-a-vis obvious customer attributes (e.g. industry, onboarding date, number of trainings completed, products utilized, service package, etc.), but it’s equally crucial to run the numbers against some of the net new information you collect in the NPS survey (e.g. the aforementioned CSAT scores).
In the example below from my Sailthru days (with redacted but directionally accurate numbers), we quickly realized that customers who rated us 4-5 CSAT on ROI questions had materially higher NPS, and ultimately operationalized that learning by rolling out a “WoW Report” to all customers each Wednesday; this report showcased week-over-week results and made it clear what incremental revenue Sailthru’s personalization algorithm had yielded. Editor’s note: Customers will invariably argue the merits of your attribution model, but you’re always better off putting something in front of them to react to than nothing at all.
Beyond soliciting CSAT scores, we also posed questions like the one below, where we listed out product changes since the prior survey and asked if 1) customers noticed the changes and 2) if so, if the changes had a significant impact on their businesses. The results were fascinating. I’ve redacted the actual numbers, but the magnitude of the spread is absolutely accurate. Customers who simply noticed changes in the product had materially higher NPS, but the difference in NPS between those who saw a real benefit from the changes vis-a-vis those who simply observed the change was far less striking. We ran a similar analysis for NRR and discovered the same takeaways.
Time and time again, only ~50% of customers reported noticing changes, which implied that we had a big opportunity to improve our product marketing and to merchandise updates and releases in a more impactful way. And here’s where things got really wild: Because we sold an email product, we sent out product release notes from our own system, meaning we could track open rates. Shockingly, more than half of customers who said they allegedly never knew about changes regularly opened the emails about those changes! They say people need to hear information 7x before it truly sticks, and this learning confirmed as much. With this analysis in-hand, we expanded our product marketing to include tactics such as product updates on every customer status call, regular webinars where customers would review how they were implementing new features, and a quarterly “Shipped by Sailthru” newsletter that CSMs forwarded to all stakeholders to make extra certain they actually read it.
Customer perception is your retention reality, so companies need to be maniacal about reiterating ROI, product updates, and everything in between to all customer stakeholders on a regular basis.
Run the regression
Notice that in the second chart above, I conveniently snuck in some data on NRR. This Tactic Talk installment deliberately focuses on NPS, but all of the suggested analysis could/should be applied to NRR as well. The name of the game is understanding which customer attributes have a statistically significant impact on the metrics that matter most to your business, regardless of whether that’s NPS, NRR, or something else. Many of those attributes will be extremely unique to your specific business, e.g. revenue lift for Sailthru, but something more like saved searches of watchlist alerts for a business like AlphaSense. Editor’s note: You can read more about NRR-specific analysis in this post from last year.
The chart above highlights some example metrics and attributes to use in a regression analysis, and I’ll call your attention to the derivative or second-order metrics included in the mix. In this example, Company X services ecommerce customers, and has noticed that the ratio of ARR/site visits has a very meaningful correlation with NRR and NPS, likely because it’s a price/value ratio for the customer. These sorts of second-order customer data can be incredibly illustrative; in the past I’ve seen fascinating insights on the correlation of NPS and NRR to price/value ratios, average customer wait time on support tickets, and much more.
NPS: Show me the money!
NPS data is also powerful for prioritizing upsell and cross-sell opportunities; a 2x2 matrix of product white space vis-a-vis NPS response can dictate a roadmap for expansion and cross-sell opportunities. Companies can also use the NPS survey itself for customer research questions and even for building cross-sell pipeline (e.g. “Are you interested in learning more about new product X [with description]?”).
Perhaps most importantly, ask for referrals! This is one area where the NPS response on its own actually is quite useful; promoters are indicating they would recommend your product, so this is a prime opportunity for an executive to follow up with a thank you note and an ask. Ask the customer if there are others in their networks you should engage (and for introductions to those individuals!). Ask your promoters if your sales team can connect with them on LinkedIn and ask for help with account introductions. Ask your promoters to write reviews on G2, help with a case study, speak at a conference, etc.
Operationalize an action plan
Companies can’t manage what they don’t measure, but they also shouldn’t measure what they don’t intend to manage! More important than the data cuts and insights is a commitment to operationalizing a plan fueled by the newfound insights (like the examples we’ve already highlighted). Once you unlock your company’s sticky drivers, you should optimize your implementation efforts to ensure those traits. For several of our more technical portfolio company products, there is a strong correlation between the number of product integrations implemented and both NPS/NRR; in those situations, new customer onboarding is focused on integrations, even if that focus slows time to value. Even though “lift and shift” implementations are often the quickest from a time-to-value perspective, they are rarely the best path for optimizing NPS or retention. Sailthru’s data suggested that our personalization algorithm was the stickiest driver of them all (see below, again with redacted numbers), and as a result we changed our implementation process to focus on personalization setup rather than first email sent/traditional “go live.”
Every NPS playbook needs an action plan. If detractors are mostly concentrated in the 5-6 score range, perhaps you should focus outreach efforts on moving 5-6 respondents into the 7-8 passive zone. You should track the composition of each detractor/passive/promoter bucket over time, as an NPS of -100 with an average response of 6.0 is quite different from an NPS of -100 where the average score is 1.0. (Relatedly, even if very low scores are outliers, each one of them warrants executive outreach. A “0” is a terrible score; even if the customer is a poor fit or a legacy client, it was your decision to sign them or retain them, so it warrants follow-up. Your detractor set not only helps with dictating product and service roadmaps but also with refining your beachhead.)
NPS as a rallying cry—it’s never too early, particularly in the AI era!
NPS is everyone’s job, and it is the CEO’s responsibility to drive alignment around as much. If the customer success leader is bonused on NPS, arguably the entire management team should be bonused on NPS. NPS should be a rallying cry for businesses; raw data should be shared and celebrated, regardless of the feedback (remember: bad news is good news, good news is no news, and no news is bad news!).
As your customer sample size grows, you can go beyond the more straightforward data cuts and build a regression model between the “other” question scores and NPS. But even if you have a small customer base and limited dimensions for analysis, it’s never too early to measure NPS and one or two supporting CSAT questions to help you dig a level deeper. Earlier-stage companies should actually be the most consumed with NPS as they iterate on the product and refine their ideal customer profile, especially since they may not have the luxury of NRR data. And in the era of AI-native startups, I’d argue NPS is more important than ever before. As we navigate through what Jamin Ball has dubbed the “ARR vs. ERR Conundrum” and face the reality that 60% of generative AI spending is coming from innovation/experimental budgets (meaning we should expect rampant churn), customer success is more important than ever before.
–
You can argue the merits of NPS because again, on a standalone basis it’s quite useless. The devil is in the details of the analysis, and I’ll say this: When companies do everything in their power—and within their financial constraints—to make their customers successful with their products, they would have to catastrophically screw something up to not succeed themselves!
It is amazing sometimes that companies do not see the obvious as is so eloquently stated in your last sentence: When companies do everything in their power—and within their financial constraints—to make their customers successful with their products, they would have to catastrophically screw something up to not succeed themselves! Great article.
CEO at UVII | Patent Owner I TEDx Speaker 🎤
4moAbsolutely agree. NPS is the gate keeper for customer feedback, an "ice-breaker", the precursor to initiating customer comfortability for expressing themselves. Where things get interesting is when you pair NPS with qualitative data. Mining short answers, voice notes, and video testimonials. This is layer where AI, tagging, sentiment analysis, and intelligent data aggregation really start to shine. Coupled together you get the numbers and the narrative.
VP of Brand and Marketing at Primary Venture Partners
4mothis is great