Some forum members have raised questions regarding the relationship between Net Promoter Score (NPS) and growth. This blog entry will address some of these concerns and suggest practical solutions for two of the issues:
Can a one-question survey predict growth as accurately as a long survey?
If you can convince a customer to spend time answering dozens of questions, you can predict that customer’s behavior more accurately than you can with one question. The problem is, most customers in this busy world won’t give you that much time—witness typical survey response rates from 2% to 20%—and you couldn’t afford the surveying and data-processing expense if they did. So most firms take a middle ground and sample customers with surveys of twenty questions or so.
While these surveys provide good forecasts for the behavior of the customers who fill them out, they do not produce good estimates of growth for the business. The culprit is sample bias . Statisticians can never figure out exactly which customers take the time to respond . Are they the most lonely, compulsive, compliant customers in the population—or are they the most loyal? And how should you adjust for this sampling bias to generate results for the entire customer base? In B2B the problem is even thornier, because the senior execs who drive purchase decisions are the least likely to tolerate lengthy surveys.
In contrast, the NPS approach yields slightly less accurate predictions for the behavior of individual customers, but a far more accurate estimate of growth for the entire business. High response rates (60% to 90%) for the short NPS surveys more than compensate for the slight decrease in accuracy for individual customers. (Bain research finds that the ultimate question captures from 70% to 90% of the customer-specific predictive power of multiple-question surveys.) This is why NPS explains differences in growth among competitors so effectively.
Are the correlations between growth and NPS reliable—and does higher NPS cause growth?
A number of perspicacious readers have noted that the statistical evidence provided in my book The Ultimate Question is imperfect. It does not provide proof of a causal connection between NPS and growth. Nor are some of the timeframes ideal. When the manuscript was completed in mid-2005, that data, imperfect as it was, represented the best available information: 8 quarters of NPS and annual organic revenue growth rates for competitors in a dozen or so industries.
Today, this database has grown enormously. Bain has conducted additional research in scores of industries around the world, gathering NPS data and apples-to-apples revenue-growth statistics for individual competitors. Bain teams consistently find that NPS explains the majority of variation in organic growth rates across an industry over many time periods.
More importantly, we have found that individual customer responses to the ultimate question predict those customers’ subsequent behaviors across the multiple drivers of profitable growth: repurchase rates, margins, cross-sell, retention, referrals, etc. Many companies, including GE, Intuit, eBay, American Express, and others have seen similar results. It is not shocking news that getting more customer promoters and fewer detractors accelerates growth. All we did was quantify this common sense in a way that made sense to business leaders—the target audience for my book . These practical leaders have little interest in advanced statistical methods.
Frankly, we see little value in continued debate about cause versus correlation, timeframes, or statistical methods. A more productive use of time would be to start building your own data to determine the NPS-growth relationship in your business. Simply identify a sample of promoters (several hundred should suffice) and a similar sample of detractors. Wait six months or so and gather data (through interviews if necessary) from those customers, including retention, repurchase, cross-sell, referrals, and so on. Then figure out for yourself what the true growth impact is of getting more promoters and fewer detractors in your business.
 For more on the “sampling” versus “census” approach, read Scott Smith’s blog, Easing the Burden, Leveling the Load: Respondent Strategies for Sampling Versus Census.
 For insights on correctly classifying customers, which improves the quality of your Net Promoter program, see Richard Owen’s Identifying Decision-Making Despots.
 Complementing Fred’s point, Scott Smith’s blog, Make the Data Personally Relevant, underscores that this metric is so easily understood that the whole company can rally around the concept of customer recommending customers.