Retail

Driving long-tail purchase decisions at retail

Michael Brill Technology startup exec focused on AI-driven products

May 9th, 2014

Has anyone come across interesting approaches to or research on retail decision-making processes? Are there conceptual frameworks to take someone through 100, 1000 or 10000 choices of highly-attributed products and do it quickly and with a high level of purchase satisfaction?

Low-level examples may be things like faceted search, collaborative filtering, decision trees... but how we buy things seems like it's a big enough issue (like a third of the global economy) that there's got to be some great higher-level thinking on the topic.

Jeb PhD Decision & Data Scientist / Experimental Psychologist / Business Intelligence

May 9th, 2014

Dan Ariely's "Predictably Irrational" might be worth a look... it's not specifically a business book, but it discusses among other things how giving consumers fewer choices results in more purchases and greater satisfaction, how to best create that set of items to drive action, etc.

David Hattenbach CEO Secret Ingredient Marketing, Hugg Technology

May 9th, 2014

Hi Michael, I can tell you that purchase decisions typically don't happen like you may think. The vast majority of people are not "utility maximizing" consumers. In fact, it is often quoted that 70% of consumer decisions happen instantaneously at the store shelf. We now know that most of us make decisions based on our emotions/intuition rather than reason. We used to think that consumers think, feel then do. We now know that they feel, do, then think. There are many good books out today on consumer decision making. "Predictably Irrational" is probably the most famous. You could also research the topic of Behavior Economics. Hope this helps. David Hattenbach CEO, Hugg Technology

George Dawson

May 9th, 2014

Hi Michael,

I have been in retail for close to 20 years and echo David's comments.  One of the seminal works in the space is Why we Buy by Pablo Underhill.  Interestingly, the online world has lead to some interesting spins on retail choice/decision making.  Look at the most successful online players (Amazon, Alibaba, eBay, Wayfair) who have almost limitless choice.  The key in the online vs. offline space in my mind is very good site search/long tail SEM (check out Adchemix and recently acquired Adchemy) to get customers to a "more" relevant selection quickly.  

I hope this is helpful.

George  

Jeb PhD Decision & Data Scientist / Experimental Psychologist / Business Intelligence

May 9th, 2014

In your specific example, an object-to-object collaborative filter will return a ranked list of "stuff you should like" provided you can give it one or more examples of something the user likes, *or* a tagged set of attributes associated with those products that the user really does require (e.g., "I only like red and fruity.") 

From there, I think Ariely's work is very much on point: provide three items side-by-side to choose from, one of which is a relatively poor match to provide contrast (in this case, probably a higher-priced item), then you set the remaining two up to structure the choice such that it's an easy comparison on a single attribute ($5 price difference, from two different regions, etc.)  To my knowledge, the bulk of the research shows that consumers are more likely to make a choice if you reduce the number of apparent items as well as the number of perceived vectors they're comparing on (I believe Ariely's related example used vacation packages) so, e.g., the consumer is really just choosing whether they want a $20 Italian red and fruity wine or a $25 Argentinean red and fruity wine, because the $40 Shiraz Reserve is (purposely) too expensive. Alternatively, you might increase average purchase if your contrast item is priced lower rather than higher.

I've built somewhat similar applications at Match.com, Selloscope, and DigiWorksCorp and had a high degree of success, so if the machine learning approach isn't giving you a usable consideration set, it might be worth considering alternative methodologies.

Rob G

May 9th, 2014

i like Jeb's suggestion... you need to show em a few tall, gap-toothed hillbillies, Michael! 

Josh Raymond

May 9th, 2014

check out mindtools.com they have a bunch of great decision making tools and strategies that would resonate in the retail sector.

Michael Brill Technology startup exec focused on AI-driven products

May 9th, 2014

Thanks guys. I have read both Predictably Irrational and Why We Buy. Neither really seems to address the challenges associated with significant choice (and Why We Buy seemed more about store layout and merchandising).

Although I'm working on a domain-neutral platform, my reference vertical is retail/restaurant wine where basically nobody knows what they're doing and there is effectively unlimited choice in a wine store and merely excessive choice in a restaurant. We've tried to figure out what motivates decisions and it's complex. We have ~ 20 factors we look including things like sensory, perceived value, scarcity, brand values, discovery propensity, risk tolerance, purchase context/intent, etc. Of course humans can't possibly work with such discrete items and search/faceted search doesn't work because people don't know what they're looking for. Machine learning doesn't work because it cannot explain *why* someone should buy this and not that. There is definitely promise in getting them to compare products (e.g., find a product similar to this but under $40), but the challenge still remains how to get them to the base product in the first place!

So my approach now is mimicking how a sales person would work. Determine rough user intent, lay out a few categories, get feedback/constraints, propose products, handle objections and do final product selection. It's super-flexible and a zillion times better than any alternative, but I can't help feeling that there is something with a bit more rigor that has been implemented. 



Marc Intrator Technology Evangelist, Experienced Business Professional and Visionary Entrepreneur

May 9th, 2014

Are you looking for an AI Technology solution?  If so,  I can put you in touch with a very successful technology/company but I would first need to speak with you to confirm my understanding.  Marc Intrator Evergreen Education Partners LLC 516-513-2846

Michael Brill Technology startup exec focused on AI-driven products

May 9th, 2014

@Marc, not sure what I'm looking for. Any approach that cannot provide an explanation of recommendations is probably off the table. That rules out most machine learning. I haven't looked at expert systems in a long, long time. Looking at existing retail advisor implementations might be helpful.

Michael Brill Technology startup exec focused on AI-driven products

May 9th, 2014

@Jeb... that's *kind* of how it works now except we pre-define the "red and fruity" concepts because most people cannot do this very well. The app is built around Cards where each card represents some concept such as "Top 10 Values on List" or "Great Wines with Seafood" ... cards can also be customized by user (e.g., wine pairing card asks several questions before generating advice). Conceptually they just represent a query predicate but many embed processing code (e.g., the wine pairing card). Cards can also be grouped into pages to provide additional structure (e.g., a person stands in front of an iBeacon-enabled Spanish wine section of a store and we push a Spain page that contains a number of cards). Cards can also contain support/educational content, promotions, etc. It's kind of like having a web overlay onto a physical shopping experience but without the web suckiness. The punchline is that I'm trying to get users to a specific product category (cluster of attribute values) quickly because we can drop from 2,000 to 50 products if we can get them to buy into the higher-level category. But it's still 50, not 3.

Once we're down to the 50 (or 20 or 10), I like the concept you raise about comparisons although I don't know how to fulfill the role of a user agent while simultaneously hiding some products and purposefully presenting non-competitive products. It'd be like Match.com showing one good-looking guy named Joe, a good-looking billionaire guy named Joe and a gap-toothed hillbilly named Skeeter.

But I will chew on this as a bit of manipulation isn't a bad thing if it drives the user to a good decision that they're satisfied with.