Were Kettles Sales Calls Effective In Tesco

1
Were Kettles Sales Calls Effective In Tesco?

• Nigel Marriott
• Chartered Statistician
• May 2007

2
Data Used In Analysis

• Total Kettle Crisp Brand sales per store only.
• Value sales uplifts from control period as
  calculated by Storecheck.
• Sales from all 1454 stores were included
• Stores grouped into Call/No Call only.
• The type of sales call was not analysed.

3
Sales teams called upon 188 stores in the 4 weeks
prior to Xmas. These sales have been compared
with the stores not visited. Rather than use
weeks, I have grouped the weeks into 4 weekly
periods as shown for clarity. We see that sales
trends in the two periods prior to the Calls were
similar for both groups of stores. This means we
can compare the two groups during the sales call
period.
4
What we want to see are similar uplifts in the
Baseline PreCall periods and higher uplifts in
the stores called upon in both the Call
PostCall periods. Statistically this is known
as an Interaction effect between Time Period
Call Status. Based on this chart it appears that
the sales calls led to an additional uplift of
72 during the Call period and -14 during the
post call period.
5

• A 2-factorial ANOVA is the simplest way to
  see if the sales calls achieved a statistically
  uplift. The two effects we are interested in are
  Time Period Call Status. In addition, there is
  a 3rd effect which is the Interaction between
  Time Period Call Status. We want all three
  effects to be statistically significant since
• We want uplifts to be higher during the Call
  period (Period effect)
• We want uplifts to be higher in stores that
  called upon (Call Status effect)
• We want uplifts to be higher in stores called
  upon during the Call period (Interaction
  effect)
• For these effects to be statistically
  significant we would be looking for P-values
  under 10 for all 3 effects and this is
  definitely the case here. Therefore we conclude
  that the sales call did lead to a statistically
  significant uplift.

6
Conclusions Recommendations

• Kettles sales calls in the run up to Christmas
  2006 had a statistically significant effect on
  the total value sales of the brand.
• The effect was an additional uplift of 72 in the
  stores called upon.
• The ANOVA test used here to establish
  statistically significant differences is quite
  easy to code into a software package to allow
  users to perform their own statistical tests.
• However, the 2-sample T-tests and the ANCOVA
  tests would also need to be coded as sometimes
  these tests are more appropriate than ANOVA.