Subscriber Discussion

Are Heatmaps Useful?

U
Undisclosed #1
Apr 04, 2016

I am a young amateur who surfs alot to learn security and surveillance stuff. I was wondering about some products that provide Heatmap in their video analytics suite that seems to aim at retail markets. Is it really that useful? If not, what other applications is there for heatmaps?

Jv
J v E
Apr 04, 2016
IPVMU Certified

We use UDP products like IPN3502HD that can provide a heatmap just from the camera itself. No additional software needed.

U
Undisclosed #1
Apr 04, 2016

Hi Frank,

Thank you for the response. Is it really useful? Are there any other applications for heatmaps?

JH
John Honovich
Apr 04, 2016
IPVM

I think it is rarely used outside of retail since most other users do not find a lot of benefit in it.

I do think retailers can find it useful much like online heatmaps are useful for websites.

For example, here is our home page's heat map:

I find online heatmaps useful primarily when we change things, otherwise the heat map tends to be constant for long time periods.

I am curious to hear what retailers are using with heatmaps. We don't hear a lot of real world use cases.

U
Undisclosed #2
Apr 04, 2016

UD1 - thanks for framing your question in the manner that you did... I have always found the value of heat mapping to be questionable.

You appear to accept the fact (as I do) that this technology exists and that it can be configured to physically do what the makers of these products say it can technically do: i.e. map heat signatures via video, based on customer traffic patterns and dwell times in certain areas.

However, you appear to be questioning the actual value of paying for and using this technology in the retail environment.... i.e. what is the ROI of employing this technology?

Excellent question, imo! And one that I've never heard a satisfactory answer to... ;)

As an old-timer, non-amateur in the industry, I like to see that you are questioning the value of this technology rather than just being wowed by the technology itself - which I maintain appears to be the goal of most of the marketing I've seen regarding anyone offering heat mapping in retail environments.

Personally, I find it telling that there are no case studies/testimonials regarding specific 'savings of monies' that can be attributed to a technology that has been deployed in the field for years.

Some (ok, I) would maintain that the reason for this is that any data to suggest that using these products will save a company any money is empirically unproven.

As an example, check out this story on Prism Skylabs - an industry provider of heat mapping analytics for retail environments - from 2014: http://www.businessinsider.com/how-retailers-track-shoppers-in-heat-maps-2014-1

Please note how the story shows pics of all the heat mapping and how the 'blue' areas show that nobody is seeing/touching/buying the stuff located there - and how the writer is wowed by these pictures and the information that they purport to give to the retailer.

In all the stories like this I've read, not ONE author has ever taken a second to step back and note that the 'information' that the blue areas in heat mapping provides is already known to these companies by simple lack of sales of items in that blue region! So why am I paying money every month for analytics to tell me what I already know??

I've asked that last question (admittedly maybe in a hostile manner) openly for years - and I have yet to have anyone who sells this type of product convince me that this technology - at least as it relates to retail environments - is anything other than the emperors new clothes.

Finally - I've harvested 2 key paragraphs from that puff piece to focus on (below):

1. From the middle of the article:

"Experimenting with ideal store layouts isn't a new idea. Groceries, for instance, tend to place their highest margin items at the very front of the store and by the checkout aisle. Research has also shown that people are more likely to purchase items on the right-hand side of an aisle. But technology like Prism's makes the analysis possible on an entirely different scale." (note it never defines what the 'different scale' is, nor how it actually provides any more data)

2. Last paragraph of the article (and my personal favorite):

"Prism Skylabs hasn't yet quantified the effect of its technology on sales, but Crosbie (of Prism) guesses that it's "significant." He adds that retailers have spent ages doing manually what Prism is now doing with cameras and computer software. "This has taken a lot of guesswork out," he says."

Where is the data?

JH
John Honovich
Apr 04, 2016
IPVM

Prism is also guessing that they have a sustainable business...

The Prism thing is more difficult since it is an add on subscription service. Compare to Hikvision and Mobotix (to name 2) that give heatmapping away for free. That's a lot less risky / difficult to justify.

For example, I would not pay extra for heatmapping for IPVM's website but it's included in our service so I use it once in a while.

U
Undisclosed #2
Apr 04, 2016

Good point.

If the technology came free, I can see using it as a validation of what is already indicated by the retailers own sales figures.

The retail industry has long known (and quantified) best practices for displaying product - the research has already been done.

Someone (anyone) please school me on why I am wrong in my determination that heat mapping is only a visual validation of information that is already known.

U
Undisclosed #4
Apr 05, 2016
IPVMU Certified

The statement

...why I am wrong in your determination that heat mapping is only a visual validation of information that is already known.

and more specifically

...the 'information' that the blue areas in heat mapping provides is already known to these companies by simple lack of sales of items in that blue region! So why am I paying money every month for analytics to tell me what I already know??

Why would you think that lack of product sales indicates lack of foot traffic in proximity to the product?

Maybe the foot traffic is decent but the product sucks!

That's one thing it tells you.

Or the reverse, low foot traffic with ok sales, maybe you have overlooked a real winner that could be moved out front .

Those are the most obvious examples.

U
Undisclosed #2
Apr 05, 2016

"Why would you think that lack of product sales indicates lack of foot traffic in proximity to the product?

Maybe the foot traffic is decent but the product sucks!"

I don't think that. Foot traffic is relative - dwell time is not.

If a product is placed - for example - in a spot where everyone has to pass it when entering and exiting the store, then foot traffic doesn't really mean anything... but dwell time always does. It represents interest. If the product sucks, there will be less dwell time - with a representative lack of sales.

(1)
U
Undisclosed #4
Apr 05, 2016
IPVMU Certified

If a product is placed - for example - in a spot where everyone has to pass it when entering and exiting the store, then foot traffic doesn't really mean anything... but dwell time always does...

There are other places besides the front of the store.

Why don't you think the information of "1 out of 10 who saw product X purchased it" vs "7 out of 10 who saw product Y" purchased it? Meaning you might want to consider dropping X and giving its space to Y.

Sales numbers alone don't tell you that, yes/no?

U
Undisclosed #2
Apr 05, 2016

My referenced sentence was written simply to clarify that heat mapping is more about dwell time than foot traffic - which was the term you used.

Also, I'm not sure that anyone selling heat mapping claims to be able to perform your example of 'the information that 1 out of 10 people who saw product X purchased it vs 7 out of 10 who saw product Y and purchased it'.

Ignoring the vast amount of variables that could come into play in your scenario, this is far more specific than the claims I have heard/read - and I think that even if this could be done, it would require other types of analytics and additional component integrations (i.e. people counting, POS integration with both analytics, etc) - even more costs involved to weigh against any return actually received.

Retail merchandising and product placement is not guess work - and hasn't been for decades. I think it is a false narrative to insinuate otherwise - and most pitches I've seen for heat mapping do exactly that.

Google the underlined subject above - there is no shortage of known best practices based on a wealth of historical data.

U
Undisclosed #4
Apr 05, 2016
IPVMU Certified

You are making many points about many things, but I am only responding to your request of

...why I am wrong in your determination that heat mapping is only a visual validation of information that is already known.

In short, what I am saying is that a heatmap gives you specific information that is NOT already known is NOT necessarily a direct correlate of sales.

I used 'foot traffic' as the simplest metric, but using 'dwell time' is fine.

So, as an example if for a large end cap display of product X, people dwell 'red' but sales are 'blue', this could indicate an oppurtunity: e.g. maybe the product is enticing but the price is too high.

It shows interest in a product, which means there is a potential. Which is actionable information, right.

Please respond to this example specifically.

As for the 'retailers know everything about merchandising already' comment, regardless of what they know, no one knows whether a product is a hit until it hits the shelves, just like music producers, with their immense knowledge of musical trends and tastes cannot be sure they have a hit song before it hits the radio.

(1)
U
Undisclosed #2
Apr 05, 2016

First of all UD4, thanks for this discourse - I am not trying to insult you or your intelligence, I just don't see value in retail analytics - primarily because the sellers of such have never provided actual proof of any.

This is where my disdain and frustration is focused - not on your words. I actually value the ability to mix it up with you here regarding a subject that has bugged me for a long time.. :)

"So, as an example if for a large end cap display of product X, people dwell 'red' but sales are 'blue', this could indicate an oppurtunity: e.g. maybe the product is enticing but the price is too high."

This is a good example of something the sellers of these analytics might say - and I agree with the thesis of the stated scenario: it could be an opportunity for price adjustment. But it could also be many other things - like:

1. Can the customer discern the value of this product based on how it being displayed - or conversely, does the product need to be 'demo'ed by salespeople for customers to 'get' the product's value?

2. Is the packaging design wrong/not working?

3. Is the product naming/branding wrong/not working?

4. Is the end cap right next to the only heating vent in an otherwise very cold store?

I am being facetious with #4, but I use it to show that inferences can be made based on a myriad of data - but the interpretation of this data (or the lack of the ability to really do so) is where actual value lies... i.e. what does this data really mean?

If the data provided by heat mapping can be shown to increase sales, then where is the empirical evidence after these many years of their existence in the market?

U
Undisclosed #2
Apr 05, 2016

I am sorry. I am physically unable to help myself:

MI
Matt Ion
Apr 06, 2016

So, as an example if for a large end cap display of product X, people dwell 'red' but sales are 'blue', this could indicate an oppurtunity: e.g. maybe the product is enticing but the price is too high.

I had this exact same thought. The heat map in and of itself may not give a lot of information, but it could certainly fill in some blanks that exist in simple sales numbers.

Another possible example: the retailer may notice some items in a given aisle aren't moving... heat map shows the foot traffic is mainly just passing the aisle by, because there's nothing at the beginning to draw them in. It may be valuable then, to put some more attractive items near the start of the aisle to entice the shopper to start down that row. Maybe take the items from the center area and move them toward the end caps, where they're more visible to shoppers passing by.

U
Undisclosed #2
Apr 06, 2016

Retail Merchandising is a profession - lots of different colleges offer Associates, Bachelors (of Science, not Arts) and Master's Degrees in this field.

Large retailers do not guess where they should place each item - they use detailed planograms (either their own, or supplied by the manufacturer - or a combination of both) to maximize sales.

Merchandising and floor/shelving layouts are designed to 'move' customers through the store so that higher-margin items are displayed prominently in the high traffic areas. i.e. The stores themselves create the high traffic areas - and they put stuff that they make the most money from in these areas.

Each square foot of retail space has targeted sales goals based on a host of historical data. Adjustments to displays and positioning are done based on data (i.e. sales, market trends, emerging products, etc) - not on the whims of what an individual store manager might personally find appealing.

Back to my original thesis: Retail stores do not need heat mapping to tell them where their high-traffic locations might be - because this is information they already know.

MI
Matt Ion
Apr 06, 2016

And yet, with all that knowledge... there's still the human factor to be accounted for. People are strange things - just when you think you have them figured out, they go and do something completely unexpected. What makes perfect sense in a dozen other stores may not apply exactly the same way in the 13th outlet.

A substantial portion of our clients are small retail (gas stations, convenience stores). I regularly see them receive new planograms from Head Office... they rearrange things as required... and then often rearrange things again because their customers that they see every day, have different needs than what Corporate thinks the average customer wants.

Even when it comes to camera placement... we did dozens of gas stations where the single greatest concern was being able to get plates from cars at the pumps, to be able to track down gas-and-dash thieves. And then, defying conventional wisdom, there was the one station in a small town, just off the beaten track, that the owner really didn't care about any cameras on the pumps, because he almost never had fuel thefts. Instead, he wanted multiple angles of the paypoint, including a covert camera above the cash drawer, because he was constantly having problems with staff skimming.

Avatar
Marc Pichaud
Apr 04, 2016

Who is paying really for heatmap ...I don't really know. It's always fun to see it during demo like a colored Rorschach test but nobody will commit on the results.

UI
Undisclosed Integrator #3
Apr 04, 2016

In the old days......retailers used to pay marketing companies to watch the video. It was a human process and expensive. It absolutely makes sense to understand what ads or placements work. Why has Neilson lasted to tell what programs are being watched. It won't change my watching habits.

UI
Undisclosed Integrator #5
Apr 05, 2016

I think really only in the retail industry would they be useful...and if a company pays for it...they are going to have to enforce it being used.

I forget what supermarket it was, but they put sensors above each check out lane to count the people, and the software would alert managers how/when to open more checkouts via a tv screen. Needless to say, there was only 2 lines open, and 10 people in each, and it was like 8pm. So this supermarket invested thousands of dollars in this system only to have it not be used.

It is a really cool technology, only if it is implemented and enforced correctly.

U
Undisclosed #2
Apr 05, 2016

"It is a really cool technology, only if it is implemented and enforced correctly."

I will grant you that the technology may be 'cool'. But your scenario is another example of the types of false narratives used by those that sell retail-specific analytics - like people counting used as an 'answer' to the pain point of having customers stand in line for long periods of time.

Retail establishments staff according to known norms - i.e. they have more cashiers working on weekends during the daytime than they have working on Tuesday nights in the late hours before closing (assuming typical retail store hours).

They do not have extra staff lying about in break rooms waiting to be summoned to their check out stations when customer lines start to build up. Can staff be shifted from other duties to accommodate swelling numbers of customers? Sure they can - and they already do this when they see this occurring (in good retail places anyway).

Which leads to the other side of this false narrative - that the manager of the store is locked in some room somewhere not being observant of any growing lines in the check out area, and needs an 'alert' to provide him/her with this important information.

"The emperor is not wearing any clothes."

U
Undisclosed #1
Apr 06, 2016

Thanks for all the responses. As an amateur I really enjoyed and learned a lot from these discussions.

The reason I put out this question is because heatmap seems good fit for retail business but the lack of real world feedback is so deafening, and this aroused my curiosity. Thus I turned to experts here that might provide me some insights which I have no experience.

IPVM is an invaluable site to me =)

MI
Matt Ion
Apr 07, 2016

Und2 makes a good point, "Retail Merchandising", retail traffic and floor layout, is a professional all to itself. There are people who excel at it, who are employed by most large chains. They may see heat mapping like this as a valuable tool, or a threat to their own value... but either way, the big outfits spend huge money on these metrics.

If you're a small-scale retailer, though... it may be very much worth it for your VMS and/or cameras to have this capability. Certainly cheaper than hiring a professional.

New discussion

Ask questions and get answers to your physical security questions from IPVM team members and fellow subscribers.

Newest discussions