The overwhelming theme from your feedback over the past day is this:
- Please show me how to calculate Merchandise Productivity.
Here, then, are the specific instructions for how to calculate Merchandise Productivity for April 2016.
Step
1 = Select all customers who bought exactly two times between April 1, 2015 and
March 31, 2016.
Step
2 = Calculate the average amount spent by this audience between April 1, 2016
and April 30, 2016 (include all of those who did not buy from 4/1 to 4/30 in
your calculation). Pretend this amount is $11.00.
Step
3 = Select all customers who bought exactly two times between April 1, 2014 and
March 31, 2015.
Step
4 = Calculate the average amount spent by this audience between April 1, 2015
and April 30, 2015 (include all of those who did not buy from 4/1 to 4/30 in
your calculation). Pretend this amount is $10.00.
Step
5 = Your comp segment measure is ((Step 2 / Step 4) – 1) = ((11.00 / 10.00) –
1) = 10%. This means that in April 2016, your merchandise productivity is 10%
better than in April 2015.
Step
6 = Repeat Step 1/2/3/4/5 for each month you wish to measure merchandise
productivity, sliding the window back a month for March, then back another
month for February, and continue back as far as you like.
I like to run these analyses with "numerous dependent variables". Instead of total sales, as outlined above, I will modify Step 2 and Step 4 (not Step 1 and Step 3) for the following:
- Phone Transactions.
- Paid Search.
- Organic Search.
- Affiliates.
- Comparison Shopping Engines.
- Mobile Phones.
- Tablets.
- Social Media.
- Email Marketing.
- Stores.
- Specific Stores (i.e. Bellevue Square).
- Outlets.
- New Items.
- Existing Items.
- Each Product Category.
- Low Priced Items.
- Average Priced Items.
- Expensive Items.
- Items Purchased Via Free Shipping.
- Items Purchased Via Discounts/Promotions.
- Items Purchased Via Liquidations/Clearance.
- Fashion Items.
- Basic Items.
Ok, what questions do you have? About a third of my consulting projects surround the methodology above. 80% of the time, I am able to identify a merchandising issue that is holding the business back, so the methodology works well.
I also like to run the analysis, going back about four years, allowing me to create comp segment metrics for the past three years (on a year-over-year basis). The trends allow me to see when specific issues arise ... it is very common for issues to arise at the start of a new season when a new merchandise assortment is introduced.
What is your thought process behind choosing customers with exactly 2 orders? Say you have a distribution that looks like this:
ReplyDelete% Cust 12 mo Ords
65% 1
19% 2
8% 3
4% 4
5% 5+
You are basing your merchandising productivity analysis on less than 20% of your customer base. I'm not necessarily disagreeing with you, just curious about your justification.
The 2x customer from last year is chosen because it tends to be the least biased of the segments. 1x customers are too heavily influenced by the method of customer acquisition last year - choose too many online buyers one year and too many offline buyers another year, and the downstream comparison isn't fair. Among 3x+ buyers, loyalty programs and discounts/promotions tend to heavily influence subsequent behavior.
ReplyDeleteSo that's why I pick 2x buyers.
Now, I run 1x and 2x and ... and 5x+ segments in my analysis, so I get to see the results across all frequency groups. I get to see if they are different or not. And when they are significantly different, I dig in an research why.
I frequently get this question ... I don't care what segment you use, pick a historical segment that you perceive to be equal and that you perceive makes the most sense. I don't care what segment you pick. I care that you do something!!