10. Conclusions
• Increased impressions per user and stable clickthru rate are driving traffic, order volume, and
revenue growth
• Holiday shopping accounts for unusual user
behavior
– Holiday sales account for dip in average order size
– People wait for sales/holidays to buy
• Overall, trends seem promising when holidays
taken into consideration
– Increased impressions don’t cause significant drop-off
in click-through rate and conversion rate
– Revenue doubles after holidays
14. Observations
• Click thru rate (users_clicked / users_impressioned) relatively stable after
holidays (slight decrease)
– However, clicks per user (clicks / users_clicked) increases quite a bit
– Clicks / users_impressioned decreases thru holidays (from 0.035 to 0.02) then
jumps and stabilizes (0.03)
• Conversion rate (num_orders / users_impressioned) also relatively
constant whereas num_order per impression is amazingly flat/constant
– Num_orders / users_clicked (clicks) spikes during holidays, slowly decreases
• Impressions per user on the rise after drop over holidays
• Avg lag hrs spikes during holidays
• Avg order value (revenue per order) decreases during holidays (sales) and
then linearly increases after
• Revenue per user_clicked (sum(order_value) / users_clicked) slightly
decreases over time
– Same with revenue per impression
• Number of orders per day increases, drops off in days after holidays, then
begins to increase again. Similar behavior as revenue per day,
users_clicked per day, and clicks per day