Subscriber Discussion

Retail Analytics

Can someone please provide any actual examples of retail analytics (people counting/heat mapping/dwell times, etc) reducing operating costs for users of these products?

I've seen and heard plenty of vague references to 'reducing costs', 'making sense of data' (i.e. stuff that was occuring that you didn't notice without seeing the analytics data), 'product placement increasing profits', etc - yet I've never seen any actual analysis purporting to show any actual projected savings.

Please note: If you choose to enlighten me, please ask yourself this question first: Could I hire an unemployed individual with at least an average IQ for a week to watch and report exactly the same things that analytics companies claim you need their software to accomplish?

I don't have any examples, though I could probably get in touch with a couple of people who might. But my question is: why in the world would shoppers want someone watching them eye a display? That's creepy, and it would change their behavior around the display, therefore giving useless data. Imagine if you were looking at the Coke display on the end cap and there was some dude sitting in a chair with a notepad writing down notes about you. You'd be less likely to stand there and look.

"That's creepy,... Imagine if you were looking at the Coke display on the end cap and there was some dude sitting in a chair with a notepad writing down notes about you."

There goes Marty's Friday night plans...

I'm not sure if you're suggesting he'd be the watcher or the watchee. Either way, I'm uncomfortable.

Creepiness is tied at least loosely to social norms. Ask someone in 1952 if they thought government controlled surveillance cameras on every corner in every metropolitan city in the US was creepy?

As (or if?) retail analytics become more 'normal' (i.e. widespread) the creepiness quotient gets diluted to socially acceptable levels. :), they would most likely become an embedded entity (i.e. covert - nobody sitting in a chair with a notepad and an inquisitive frown)

But discounting any creepiness factor, my concern is examining the actual value of using these products. I gladly stipulate to their technical efficacy - I'm sure they can do exactly what the sellers of these retail analytics claim they can do - technically. But how does the knowledge derived from using these analytics translate to actual savings by the user?

Marty could never have the watcher job obviously, he couldnt stay quiet for more than 3 minutes before explaining to the shoppers why they should buy Goya Soda bottled in NY vs Quebec originated Pepsi...

I would only take that job with the stipulation that I get to use a bullhorn.

My other question or point of input (on topic) is in regards to a medium sized high end grocery retailer we do a lot of work with; they have no in-store video analytics, but they know every little metric of shopper flow, time at end caps, and every other conceivable measureable. One thing I never understand, and haven't asked about, but how are they actually measuring or reporting that information? Real time obvservation? It always shocks me just exactly how detailed they can break down a customer experience in the store.

That is essentially my point, Sean.... are retail analytics providing anything that isn't already readily available via simple observation at various time intervals?

How are these anlytics sold? Are there recurring licensing fees to use them each month/year per location (which I imagine to be the case), or are they sold at a flat cost?

Show Me The Value

Marty, your example of an unemployed guy weakens the case. It's a strawman and not a serious argument that video analytic developers or proponents (the two that are left) make.

Their contention is that you can spot trends over time - did more people come down this aisle today than last Friday, last week vs October, when we ran promotion X vs promotion Y, etc.

There is some value to some retailers of this. The bigger problem I see is the cost - the need for separate cameras with different angles to do the analytics. The need for such separate cameras in each part of the store to be analyzed. The need to purchase a different system to do the analysis / reporting. Retailers are notoriously cost constrained so this makes it harder to justify.

I'll admit to liberal use of strawman arguments to accentuate my belief that the actual value of retail analytics is unproven. However, my use of the strawman in this particular case was shrouded in sarcasm and not intended as a debate point. :)

I agree that identification of 'trends' can add value to the decision making process when it comes to product placement - but only in very large applications where this data can not be readily determined via simple observation.

I also agree that the cost of employing retail analytics is a huge factor when determining value. The 'information' revealed by analytics has to be provided in a manner than can still lower operating costs from pre-analytics levels (including the cost of the analytic deployment itself). If analytics can save a retailer $X per year based on better knowledge of customers, does it make sense to pay $5X for this information?

One addition note: why are we making the assumption that retailers are not already making the same determinations that analytics purport to be needed to 'see'?

Product merchandising is already a highly studied subject/science, and has been since people started selling stuff to other people.

Show me the value of retail analytics.

So what I've picked up from the propoganda of analytics, is that they outdo humans on the sheer fact that they can handle larger quantaties of data. Although most is sold as alarm-generators.
Does one hang out too long near a pinmachine.
Does that customer get way to close to the racks, indicating (s)he might stick somewhat in their jackets.

I have no personal experiance with analytics other then a blue monday toying around with Bosch it's IVA.
Is it actually possible to track single individuals trough the use of multiple camera's ?

Rogier, I started a new discussion on "Who Can Track an Individual Across Multiple Cameras?"

To reinforce the notion that retailers already know huge amounts of info about individual shoppers, please read this article from the NY Times from a year ago - How Companies Learn Your Secrets

It's clear that retailers are starving for data of any kind that can help them make marketing/merchandising decisions allowing them to increase profits. They seek out habits that humans natuarally exhibit so they can exploit this human trait for their benefit. Big retailers are always looking for an 'edge'. This makes them vulnerable to nice sounding solutions that purport to give them any kind of edge - even unproven 'solutions'.

Can retail analytics detect the patterns inherent to habits? Probably. But are analytics (and their associated costs) required to reveal this data?