A big advantage for ecommerce sites is the ability to track minutest details of customer activity. For the average store, however, swing door operators is the extent of their analytics. Now it’s time to bring analytics deeper into store isles.
One idea consists placing infra-red sensors at the ends of isles. These would trip each time customers walk in or out. Unfortunately this would only show that people walked into isles but fail to give insight about what they did there.
Another approach is being explored by Rayid Ghani at Accenture Technology Labs. Their approach is to use video cameras to track in-store motion but the technology is still in development.
The approach I like best utilizes RFID.
RFID technology is getting cheaper by the day and is estimated to cost around 5 cents a tag in the near future. With prices that low I would slap tags all over the store. Let’s consider using RFID technology at the grocery store.
Placing RFID tags on carts with receiver tags at strategic locations all over the store is a salivary idea for any analytics junky. The implications are fantastic and such a network of sensors would allow us to perform all kinds of rich analytics. What really excites me is the idea that if we placed receiver tags at checkout lanes then when I give the cashier my preferred card the system would immediately match my cart id with my store history. This opens up a whole new layer of analytics- now we can track both averages and specifics. Averages apply to the average customer but specifics apply to me as an individual, and both are useful measures.
Listed below is a partial list of what RFID data could potentially reveal:
- We can now track how many people walked in an isle
- If position of product SKU’s is known (as it should be) then it’s possible to see what shelves caused people to pause.
- We can trace the average walk-paths of customers. In the old days store managers would literally look at the floor to see wear and tear for rough estimates on walk-paths.
- As a reverse metric we can measure what items got little customer attention. If those items typically sell well in other stores we can insert them along popular walk-paths.
- We can now calculate conversion rates. Say an item on the popular walk-path gets 50 pauses a day but is bought only 2 times this would give us a conversion rate of 4%. Another item outside the popular walk-path gets 12 pauses but is bought 3 times. The conversion for this SKU is 25%. As store manager I would switch these two items.
- I hate frozen isles, they are far too cold and cause me to walk real quick. Up until now grocery stores didn’t have any idea how many dollars were being lost because of the sub-arctic temperature. Knowing my average speed and tracking my speed in frozen isles they would realize I find the temperature uncomfortable (caution: this could also mean I didn’t need anything from the frozen isle).
- Habitual by-pass isles: Let’s say I never walk through isle 8. If my shopping list indicates I might like items on isle 8 then the store could send me a special promo for an item on isle 8 tempting me to break my habitual walk-path.
- Moving price analysis: Let’s say SKU #1332 gets really high pauses but fails to convert, this may indicate that customers are interested but find the item too expensive. The store should consider dropping price a little if they want to move this item.
- Price sensitivity analysis: Let’s say customer A and customer B have the exact shopping receipt. Normally the retailer would look at the shopping receipts and give both customers the same preference ranking. Now let’s dig deeper and look at RFID data. We find customer A pauses only at discounted SKU’s while customer B pauses equally at discounted and non sale items. Clearly customer B is not motivated only sales alone, thus for the retailer is more valuable.
- Browsers Vs. agenda specific shoppers: Browsers tend to walk through every isle while agenda shoppers only visit isles with items on their list. I’m not sure how a retailer would use this metric but it’s there for them.