AI and Customer Retention Practices.
A semi-fictional case study.
It’s a dark and stormy night. About 10 pm. Maria walks into a store. She goes straight to the pharmacy aisle. She is looking for Melatonin, a popular herbal supplement to aid sleep. She knows her favorite brand, she just ran out of it.
Unfortunately, to much of her dismay, the space allocated for her brand of Melatonin on the shelf is emptied out. They must have run out of it.
At this point Maria has a few options:
Grab the same product from a competitor’s brand. It’s right there, one row from her missing favorite one.
Ask a store clerk if perhaps they have it in storage.
Walk out, go home, and watch Reefer Madness, Louis J. Gasnier - 1h and 6min, 1938, USA. Voted the worst movie of all time (1), and guaranteed to work as well as a standard dose of Melatonin, if not better, probably faster.
Walk out. Start wandering in the stormy London night and fall in love with a stranger.
Maria is an actual person. So were many others selected individuals, from different walks of life, who participated in a ‘guerrilla style’ Product Design research study.
O, where is loyalty? If it be banished from the frosty head, Where shall it find a harbor in the earth? (2)
The research focused on customer habits and brand loyalty. The brand is a not-to-be-named vitamins and natural supplements company. Maria’s behavior, as well as the one of other volunteers in the test, was observed by a small team of UX/Product Designers and Marketing professionals via in-store surveillance cameras. Let’s forget about the creepiness of this setup for a moment. It worked. It worked very well, it was dirty cheap to implement, and all the participants were aware of it beforehand. No video recordings were saved.
The research was conducted for an AI-driven Product Locator chatbot app, suggested by a UX/CX consultancy (yours truly) to the vitamins and supplements company.
A, B, C, or D?
What did Maria do? I regret to report that Maria, nor anybody else, chose option D. I would have. Maria instead went for A. So did many others.
A large number of participants, 66%, went for A. 82% percent actually picked up the competitor’s brand product. Some put it back. Staring at it for a while.
So, we ran the same script again. This time we handed the participants a new phone. Maria was told to look for the same item. It was not there. You needed to see Maria’s expression! Precious. Are they f**ing kidding me? What kind of a sick test is this? Just at that time, we sent her an alert to the new phone. “Maria, launch this app”. That’s all it took. Maria launched the Product Locator app. The ChatBot asked her which product was she looking for. She quickly selected it. The chatbot returned with a list of nearby stores, open and with the product guaranteed to be in stock. The bot also added a 20% discount coupon for the purchase.
The loyalty well held to fools does make Our faith mere folly. (3)
Brand loyalty is in great peril in these days of products overload. User Sentiment Data has always been hard to reckon with, so much depending on how you pose the questions and to whom…, these are the 2016 figures from when the project started:
Men and women are loyal to specific brands, but 35.66% of men and 28.77% of women would consider other alternatives that offer better quality. (4)
35.03% of women and 32.95% of men would consider other alternatives to their preferred brand or product that offer a better price. (5)
80% of shoppers would switch stores or brands when offered a compelling promotion. (6)
You get the idea.
The UX Journey
Below is one of the User Journey map I did to make the case for the ChatBot Product Locator app. You can find a larger, legible sample here.
The document maps the user journey of customers with the Product Locator app and without the Product Locator app. It tracks each event, user sentiment observations (via cameras), crucial data points in the Evaluation chart, and opportunities (both analog and digital).
Product Finder ChatBot user journey map
How Does It Work?
The combination of the physical and the digital offers greater opportunities compared to the digital-only scenarios. The ChatBot was designed as a product finder. Its focus is, by design, very narrow. It solves one specific problem: I want to know where to find what I am looking for, right this second. Not online. Not delivered. Just now. I’m out of it and I am ready to buy.
“The store ran out of it” scenario was one of several addressed. But it quickly became the most relevant. The Product Locator took the shape of a ChatBot because it was the most immediate, personal, streamlined and stress-free solution. AI (Machine Learning) is part of the equation here because it provides the best way to automate a chat today. The more people use it, the more training data it digests, the more it keeps learning and provides even more accurate answers. This is also part of a recent trend called microapps.
There is a new fertile space for brands considering developing microapps for customer engagement. This is what’s happening in China:
Tencent has re-engineered the WeChat messaging app in a way that applications smaller than 10 megabytes can run instantly on WeChat’s interface. It is now offering 580,000 mini programs after just one year of development, compared to the 500,000 mobile apps that Apple’s App Store published from 2008 to 2012, according to Hu Renjie, WeChat’s mini program director. “The mini program is a brand new product model which can seamlessly link the offline and the online together,” said Hu, adding that the mini program scheme has attracted 1 million developers. (7)
Product Finder ChatBot sample wire-frames
Retail partners data sharing
For all of this to work, new data sharing partnerships need to be put in place. Brands with a strong foothold in the brick and mortar retail business can better engage with customers if the retail stores carrying their products are open to sharing detail data on sales and warehousing. A similar setup already happens with Google Available Nearby option in their shopping search tab. When it first launched a couple of years ago Google offered retailers to show how to properly prepare and share the data. The vitamins company in question already has a 'preferred retail partners' program running, so it’s not that complex for them. Product brands should consider setting up partners data sharing programs on their own cloud.
Added features and app footprint
Later we also tested a ‘scan here to find it’ QR barcode sticker visible when the shelf is emptied out of the product; scanning the barcode launches and initiates the chatbot. It’s important to stress that the app footprint is tiny. It downloads and launches in a just few seconds (like 3 to 5 secs). The immediacy of the user engagement is crucial in these scenarios.
Keep it simple
Often time the best product solutions are the simplest. Although there was a push to put additional features in the app, Product Design successfully made the case to keep it as narrow-focused as possible. Do one thing, do it right, do it fast. Fast download, fast engagement. Accurate answers. Repeat.
Customer retained. Brand loyalty saved. No need to watch Reefer Madness.