January 31, 2023

– Advertisement –

For publicly traded companies, it’s not enough just to be profitable — you must grow your profits. And growth is often driven by acquiring new customers while retaining existing ones.

– Advertisement –

Much of the attraction of the Net Promoter Score is its potential to be a leading indicator of growth or a harbinger of declining growth. This is done by measuring people’s intention to recommend companies and using that likelihood-to-recommend rating to classify them as promoters, passives or cynics. Word of mouth is a powerful driver of buying behavior. How people choose a realtor, accountant, or movie to see is often influenced by recommendations from friends and coworkers.

– Advertisement –

But one of the early criticisms of the validation of the Net Promoter Score’s ability to predict company growth was that its author, Fred Reichheld, correlated the NPS data with historical data and not future growth data. Since then, we have replicated some of their findings and found at least some future predictive ability of NPS in 11 out of 14 industries (average correlation of .35).

In our analysis of the published NPS literature, we found a paucity of data at the individual level—that is, few researchers were tracking recommendations and follow-up with the same people (a longitudinal analysis). We needed a longitudinal dataset – one that had individual behavior and behavioral intention data prior to individual behavior data.

– Advertisement –

To measure the predictive validity of likelihood-to-recommend (LTR) items and Net Promoter Score (NPS), we analyzed data collected over a two-month period in the online grocery industry in the US.

Longitudinal Study of Online Grocery Shopping Behavior

Between December 2021 and January 2022, we recruited 390 existing users from eight US-based online grocery shopping brands (Food Lion, HEB, Kroger, Meijer, Publix, Safeway, Walmart and Whole Foods).

We asked participants to rate the overall quality of the user experience of the online grocery service they used most with SUPR-Q® and UX-Lite® (see “UX and Net Experience of Grocery Websites” for analytical details and results) Promoter Benchmark”. Participants also completed the eleven-point LTR item used to calculate NPS (Figure 1).

Figure 1: LTR item.

After this we followed up with all the participants after about 30-60 days (February 2022). We asked whether they had recommended the service they rated to someone in the past month and their purchase behavior (whether they bought, how often, and how much) with the service in the past month.

See also  MIDI Madness with ChatGPT: AI-Powered Tunes That Will Make You Laugh, Cry & Dance | R bloggers

We received a total of 320 usable responses for the follow-up study, an impressively high 82% follow-up rate. In addition to its simplicity, NPS is popular because of the expectation that a high LTR should be positively correlated with actual recommendations.

Prediction of LTR and NPS future recommendations

We analyzed LTR and NPS relationships with reported follow-up recommendations, first at the overall service level (eight brands) and then at the individual level (320 respondents).

Overall Service Level Analysis

With only eight services, our analysis options are limited to correlation. Only very large correlations (greater than .708) would pass the standard criteria for statistical significance (p

As expected, since NPS is derived from LTR, the correlation between LTR and NPS was very high (r = .95, p

The correlations between self-reported recommendations for the previous month, between LTR and NPS were .491 and .327, respectively, neither of which was statistically significant, but both were positive, hence trending in the hypothesized direction. Figure 2 shows a scatterplot for LTR by reported recommendation rates.

Figure 2: Scatterplot of LTR by self-reported recommendation rates. individual level analysis

Relative to the analysis at the service level, we expected correlations to be lower at the individual level (as there is less variability when the data are collected), but because the sample size is much larger (n = 320), the statistical significance of the relationships cannot be established. with smaller correlations (eg, when r = .092, p = .10; when r = .11, p = .05; when r = .144, p = .01). At the individual level, we can also examine non-linear relationships by comparing recommendation percentages for each NPS category (detractors, inactives and promoters).

Correlations for LTR and NPS with self-reported recommendation rates were .331 and .352 (both significant with p

Analysis by NPS Category

Figure 3 shows that more than half of promoters (58%) reported recommending a grocery service to someone in the past month – just over four times the recommendation rate of detractors (14%).

Figure 3: Recommendation rate for each NPS category with 90% confidence interval.

As we saw in the previous analysis of recommendation behavior, not all originators recommend, but originators dominate recommendations. Across all NPS categories, about 44% of participants reported making a recommendation. Promoters accounted for the majority (58%) of all recommendations. This is consistent with our earlier findings that between 51% and 77% of all recommendations come from promoters.

See also  End-to-End Guide Using Trader Joe’s: A Decentralized Application

LTR and NPS Reported Prediction of Future Buying Behavior

The results from the previous section demonstrate a significant relationship between the likelihood-to-recommend and follow-through rates of self-reported recommending behaviors.

Because behavioral intentions are usually correlated with each other, we also examined the relationship between LTR and NPS across three different self-reported shopping behaviors (purchase rates, number of purchases, and purchase amount).

Overall Service Level Analysis

Table 1 shows the relationship between LTR and NPS with three self-reported purchase metrics over the past month. As expected, the correlations for LTR were similar to those for NPS (same rank order in the table), with LTR slightly higher than NPS.

VariableLTRNPS Purchase Rate.439.300 Number of Purchases.427.391 Purchase Amount.649.549

Table 1: Correlation between LTR/NPS and procurement metrics at the overall service level.

In line with the hypothesized direction, all correlations were positive. None of the correlations were significant at p

For example, consider the scatterplot for NPS and the purchase amount shown in Figure 4. The scatterplot indicates that Whole Foods is the one outlier with the lowest amount spent (it is less of an outlier than when analyzing these relationships with the SUPR-Q data, but an outlier nonetheless). Recalculating the correlation without Whole Foods would yield a statistically significant correlation of .71 (R2 = .51, p

Figure 4: Scatterplot of LTR by self-reported recommendation rates. individual level analysis

Table 2 shows the relationship between LTR and NPS for the previous month’s self-reported recommendation rates and three self-reported purchase metrics (purchase rates, number of purchases, and purchase amount).

Variable LTRNPS Purchase Rate.152.153 Number of Purchases.205.208 Purchase Amount.239.243

Table 2: Correlations between LTR/NPS and purchase metrics at the individual level (all p

All correlations were positive and significant (p

Analysis by NPS Category Purchase Rate

Figure 5 shows that the buyout rate for promoters (95%) was statistically higher than that of opponents (79%) (p

Figure 5: Purchase rate for each NPS category with 90% confidence interval. number of purchases

Figure 6 shows that promoters reported significantly more purchases than detractors (3.2 vs. 2.4, a 33% increase; F(2, 317) = 4.91, p

Figure 6: Number of purchases for each NPS category with 90% confidence intervals. purchase quantity

Figure 7 shows that promoters spent an average of $99 (57%) more than cynics and $46 (20%) more than passives. These differences were statistically significant (F(2, 317) = 9.98, p

See also  U.K. lawmakers weigh idea of jailing Elon Musk or Mark Zuckerberg in online safety debate

Figure 7: Purchase amount for each NPS category with 90% confidence interval.

Here are our key findings, study limitations and key findings.

key findings

A longitudinal analysis of 320 users of eight online grocery services found

NPS was a forecast of future recommendation. Both LTR and NPS were positively correlated with future recommendation at the product level (LTR: r = .491; NPS: r = .327) and individual level (LTR: r = .331; NPS: r = .352).

The relationship with LTR was non-linear. As we observed in similar earlier analyses, there was a non-linear relationship between LTR and future reported behavior, through analysis of those behaviors by each NPS category (condemner, passive, promoter) displayed.

Poor NPS ratings predicted fewer purchases and fewer dollars spent. Behavioral intention to recommend was also predictive of spending behavior. Specifically, detractors reported engaging in significantly less purchasing behavior than promoters. This is consistent with our earlier findings on a similar dataset using SUPR-Q scores to predict purchase behavior.

study limitations

Category restriction may have weakened the correlation. Another feature of the data is that the LTR/NPS metrics tend to be high (average NPS of 19%). These generally high NPS ratings can lead to a restricted range (which can artificially reduce the magnitude of correlations).

We measured self-reported recommendation and purchase rates. We relied on self-reported recommendations and purchase rates from people who use online grocery shopping services. It is possible (and likely) that people who have a more favorable attitude toward a brand may increase their recommendations and purchases to some unknown degree. We think it is prudent to be skeptical about the accuracy of actual recommend rates and actual buy rates. While purchase rates are easy to track (with receipts, for example), recommendations are more challenging to track. Instead, we recommend looking at the relative difference between categories (promoters vs detractors) in their ability to portray high and low buy and recommend rates.

Main takeaway

NPS promoters are not only more likely to recommend but are also more likely to shop more often and spend more. Compared to detractors, promoters were four times more likely to report recommending, had a 21% higher purchase rate, reported making 33% more purchases, and estimated spending was 57% higher. Enterprises should try to keep their promoters happy and generally avoid making any customer unhappy.

Source link

– Advertisement –