Does a survey completed on a Monday have a lower score than another day?
[caption id="attachment_2814" align="alignnone" width="300" caption=""Monday Monday - can't help that day..." This client clearly shows Monday is a bad day."]
[/caption]
Last Friday's Net Promoter News featured a snippet of research from UK analyst John D'Arcy. According to him “People hate Mondays. And they were taking that fact out on our survey. More people were satisfied on Thursdays, Fridays, and Saturdays before the rate started a depressing decline to a Monday low.”
Rainy Days and Mondays
Was it true? We at CustomerGauge decided to take this as an Easter challenge. We fired up our data extraction systems, and over a coffee-and-chocolate fuelled data-binge we cranked the numbers of ten CustomerGauge clients. Using nearly a quarter of a million data points we arranged the results by the Day of the Week of Survey Completion. Our aim: to see if we could spot any patterns to support the premise that Net Promoter® scores would be lower on a Monday because people feel gloomier.
Our first dataset looked promising, and surprised us. The data immediately showed that Mondays were in fact the lowest scoring day, some 5 NPS points below the client average. You can see this in the Client 7 graph above - we show the difference from the client's overall Net Promoter Score, and the total number of points in "Resp" (Resp = response). In this case, Monday is 4.6 NPS points below the average. (so if the overall NPS was +15.6, the Monday score was NPS +11).
[caption id="attachment_2816" align="alignnone" width="300" caption="Happy Mondays. Saturday Night Fever."]
[/caption]
Then our next dataset stumped us. Client 2 showed a 6 point increase on surveys completed on a Monday. If anything, Saturday was the worst day for Net Promoter.
[caption id="attachment_2818" align="alignnone" width="300" caption="TGI Thursday."]
[/caption]
We pressed on. The next dataset showed that Client 10's customer base favoured Thursdays, but hated Tuesdays. And Sundays. Hmmm.
[caption id="attachment_2817" align="alignnone" width="300" caption="Uh-oh. Wednesdays, no, Tuesdays child is full of woe."]
[/caption]
Client 1's data told a similar story on Blue Tuesday, but had a brighter Monday, Wednesday and Friday.
[caption id="attachment_2815" align="alignnone" width="300" caption="I do like Mondays. And can't wait for that Friday Feeling"]
[/caption]
Here's another example where Mondays are favoured over other week days, according to Client 5 and their 64,000 data points.
[caption id="attachment_2813" align="alignnone" width="550" caption="All Clients. 244,000 data points. Not so much variation over the week."]
[/caption]
Overall, what does the data tell us? As we expected, over all the data points, there is not a significant score difference depending on which day of the week you fill a survey in. The difference is within a range of 1 - 2%, which can be put down to some variability in the data. In short, we don't support Mr D'Arcy's claim that there is a "Blue Monday" effect, although his hypothesis is interesting.
But there are some patterns that intrigued us. Rather than a day of the week skew, we believe the numbers could be explained by variability in service on certain weekdays. These may warrant closer inspection by our clients. Some notes we made:
- We did notice that for one b2b client, surveys completed at the weekend were very negative. Maybe due to system errors happening on a weekend with no-one to fix them?
- Several of these clients listed have a survey sent after a fixed number of days after a transaction - for example 14 days. So the NPS results might refer more to service issue patterns on the weekday itself. For example, weekend parcel deliveries might be less reliable than a weekday delivery.
- It could be that customer service workers are more on the ball on say, Thursdays, rather than Mondays, so the effect is more on the delivery side than the customer perception.
- One bright spark in our office suggested that this data be used to fine-tune which workday to send a survey, if a client wanted to show a higher Net Promoter Score. We quickly squashed that idea. NPS should be all about consistency - in which case it does not matter, as long as you keep repeating the same environment. In any case, instinctively we feel that it should not make a difference.
A Day in the Life
For CustomerGauge clients: we are happy to share the details of this exercise, and if demand is high enough we will create a special report to help understand Net Promoter by day of the week (sent, completed, or transaction date). Give us your feedback.
Please also use the comments to discuss your thoughts - did we miss anything? What other information would you like to see? Should we call the Mythbusters?
Information for fact-fans: Data points taken over last 12 months. More than 300 responses in each weekday bucket, with most more than 1000 data points. NPS based on standard 0 - 10 scale. Client NPS results: Min 5.6, Max 67, Average 28.4, Median 18.5, all based on continuous 7 day surveying.
Useful references: Data Dredging, Calendar Effects. See also our NPS article: "Is there a Dutch Effect?" - dealing with cultural bias
