Gear Chats and AI: the New Way to Buy

If you have spent any time shopping for outdoor gear in the last year, you’ve no doubt noticed the ever-present chatbots in the bottom right corner of vendors’ web pages inviting you to ask questions.  Shoppers all wonder, exactly what is this? Is it a person, AI, or both? How does this whole thing work? No matter the answer, the chatbot growth rate is expected to be 23 percent per year.

The appeal is that either a real person like a river guide or a ski expert guides you on a purchase decision. In some cases, it’s not a person, but rather a cleverly designed AI. As it turns out, I’m probably a very good person to help explain this mystery. One of my graduate degrees is in Computer Science, and during my years at IBM, I worked with super AI Watson in cutting-edge pharmaceuticals. Watson famously won Jeopardy in 2011 and became world famous as a leader for natural language processing. I still use AI many times a day as a principle investigator on an astrophysics Dark Matter research project.

I am also a gear expert, and have worked for a chat company as the expert you get to when you do chat.

You may not be aware of just how many of these AI/chatbot systems are out there. They cover a far larger foot print than the outdoor gear industry. They are everywhere now, and as you may already know, Amazon, UPS Virtual Assistant , and many electronics dealers all use them. There are even websites for people to find jobs as a gear expert and become the person who chats with customers.

Here, we are just looking at the outdoor gear sector. There are many players in this arena too. From the largest Gearinc  with thousands of gear experts, to new up-and-comer Withremark that has less than 100.  REI and Backcountry Gearheads are also very popularly used chat services.

To really understand what’s behind the veil, I decided to join one of the gear expert companies, and see what life on the other side of the chatbot symbol was like. I had an odd applicant problem, however. I was wildly over qualified, so I wondered if anyone would take me seriously. We’ve all been rejected for having too much experience at some point. In my case, I am currently a ski patroller who’s going to senior level this season and I am a bike patroller on the same hill over the summer. Being on bike patrol resulted in my becoming a fully Certified Shimano Mechanic, a Trek guide (highest level) and an equal investment in Sram, Fox, and several other major manufacturers.  Before ski patrol, I was a triathlete, and won my nationals in sprint and international distances between 2005 – 2007 for the 200-pound weight class. My wife and I also compete in open ocean distance swimming races, have an extensive collection of great climbing gear, and I used to guide teams out on their first snowshoe and camping expeditions in the Appalachian mountains.

After a little inquiry, I got an invitation to join one of the chat companies as a gear expert for outdoor brands. Everyone joining the team introduced themselves to the team on the internal meetup group online. Before long, I was in and chatting away and I was trained and assigned to about a dozen client companies. Most of these were names you’d know very well.  I quickly realized my gear expert company had two categories of chat support, AI for the budget-conscious, and real human gear experts for the companies with more marketing capital. I will tell you about both.

Life inside the Chat World: Customers

There are basically three types of customers who shop gear vendor websites.

Tier 1 – These are customers who are just randomly browsing, with no real mission, and perhaps just curious onlookers. We’ve all done it.

Tier 2 – These are shoppers who have a mission, of which the vendor site is only a part. They are looking for a specific product like a tent or skis and check in 4 or 5 sports websites for the lowest price, or for a sale, or perhaps additional product information.

Tier 3 – These shoppers almost always already have something in the cart for potential purchase when they chat. On the gear expert side, we see that right up front. They may be past or repeat customers, and they are the most serious buyers any vendor can get. In the end, they buy more than the other two tiers combined.

Life inside the Chat World: AI

Our chat company’s AI offering was orchestrated by a small group of company programmers who had backgrounds similar to my own. Their job was to fabricate the AI framework for clients based upon a knowledgebase they assembled and likely customer inquiries.   The knowledgebase was a hybrid of information garnered from the vendor’s website, and a series of discussions with the client. AI would hopefully unite the right question to the right knowledge base item. This is accomplished via the classic AI model, which I will reproduce in layman’s terms below.

The challenge for programmers is that middle red ring. You see, context is everything. As we all know, many English words have multiple meanings.   The word “bat” could be an aluminum slugger, or a small cave-dwelling mammal, or something women do with their eyes. Humans are far better at context than computers, and those older grey-haired people have more experience than younger adults, who are far better than their children. Experience, particularly human experience, is very hard to replace.

Life inside the Chat World: Gear Experts

Once I was on the job, I realized that there were a few things I had to get used to.  First was the way I got my work. Everything I needed was on my laptop. I had a single pane of glass application with my knowledgebase, the customer cart view, and the vendor site the customer was working with as a shopper.  The shopper’s cart was almost always full with the items we were about to discuss.  I had a vendor-specific knowledgebase that covered many questions, and a recommendation tool that was up-to-date with stock availability, item color, and size. Nonetheless, work was never steady. This was largely an evening and weekend job, where I could get as many as 12 – 15 chats per hour. The pay was just under $4 a chat, but I never knew when a chat request would pop up on my screen.  Many hours passed with only 3 or 4 chats in 60 minutes. I ended up serving about 200 chats a week on average.  After each chat, the customer rated me, which made up my customer service score.

My boss was a terrific guy, a former Olympic team competitor who always treated me well. I appreciated our many meetups on Zoom. He let me know that I was the 3rd highest chatter (by volume) in the company, with a 9.6/10 customer score and a 58 percent sales record on a major sports clothing vendor. He said that this was “incredible.”  I realized before long that the nicest part of this job was that it satisfied my inner desire to help others. One customer said, “You are the Jedi master,” after I helped him, which was more rewarding than money. Helping others was a key part of the job.

At the same time, a significant number of people were there to chat with no intention of buying anything.   If I had to guess, 15 percent of the chats I served where customer service redirects. These were ordering problems, financing, or other things outside the realm of a gear expert. Another 15 percent where unauthorized retrofits on ebikes, chats that really should go to a mechanic, and “please decode my bike’s serial number.”  Late one night, I had a customer ask me, “What bike do I have?” and this kind of thing was not as rare as I thought.

Life inside the Chat World: Vendors

My dozen vendors included Ebikes, regular bikes, board shorts, bike clothes, skis and even bike selling sites. They were all enduring an economy that could be better, so raising sales numbers would be a blessing.  The attraction of our group was the dual offering of both AI and/or real people to serve chats.   But there were problems that plagued the entire process. The largest issue is available inventory. It’s hard to sell when the vendor is out of stock on popular sizes and models.  At this point, the bike industry is going through hard times, and it’s not a secret. I even had out-of-stock templates in the general knowledgebase because of the frequency of that occurrence.

Life inside Chat World: Chat Company Management

It did not take me long to realize that all chats were saved and monitored, and our performance was evaluated by more than just Volume x Customer score x Sales percentage, which we saw in the applications.  There were clearly two silos, one between the vendor and the chat team management, and another between the customer/buyers and the gear experts.  This actually was what created some of the biggest problems in the system.  

Steve Jobs extolled the importance of the customer experience above nearly everything else. A typical chat company hook for marketing would be to tell a potential client sports vendor that “Our experts help your customers get to the right piece of gear.”  This sounds great in a business grad school classroom, but reality is not so simple. In a struggling economy, the customer experience isn’t actually the search for the right piece of gear, but rather the lowest cost item that will competently complete the mission.   

It’s up to the gear expert to bridge that confidence gap. I recall one day at nationals in triathlon, when a friend looked at all the pricey bikes on the expert section of the rack and commented, “This is a high tech arms race.” He was right.  All the expert bikes at the 2005 triathlon nationals were over $5000, ultra light, and high tech. It’s more like $10,000 today. So the right piece of gear would likely be the most expensive, lightest bike the buyer could afford. But as I said, a poor economy and COVID shortages have changed the rules. Now athletes consider used bikes, and direct-to-consumer models that significantly lower cost. The customer experience today is the competent gear and right price, not just right gear alone.

More to come....

David Flynn

David Flynn wrote:

The customer experience isn’t actually the search for the right piece of gear, but rather the lowest cost item that will competently complete the mission. 

AI still has a long way to go: every time I try to use a chatbot to discuss a product, it ultimately redirects me to a number where I can call an actual person; every problem I put to ChatGPT has made it choke.

Because I appreciate personalized service, my webstore has my contact info easily find-able, and I'm always glad to take a call, email, or IG message and answer questions.  I hope that even as my brand expands, I'll still find time to talk to people personally.  I get that's harder for big companies, but the abiliy to reach someone who actually works there and has their hands in the process is priceless to me.  The suppliers I buy from have sales people who actually handle the product, and take time to engage over the phone and ensure you're getting what you want.  There's no way AI will ever get close to that, because no AI has ever hung at a sling belay on El Cap on a windy afternoon wishing their partner would hurry up, or set up a ledge in a sudden thunderstorm, nor had a flat tire twenty miles from home.  Those things can be described, but not felt.  AI doesn't feel.

*flagged as bot

;)

I haaaaate the AI bots and just spam them with insane questions until I get a human operator. 

There is nothing inherently interesting about AI, and chatbots are one of its most boring forms. 

If there's something interesting here, it's the fact that we have chosen for decades to build a system of human interaction in which it is an easy and sensible upgrade to replace a person with a bot. 

take TAKE wrote:

There is nothing inherently interesting about AI, and chatbots are one of its most boring forms. 

If there's something interesting here, it's the fact that we have chosen for decades to build a system of human interaction in which it is an easy and sensible upgrade to replace a person with a bot. 

It's less that it's easy and sensible, but rather that an AI replacement promises to offer a 'good enough' replacement such that the lost revenue from poor customer service will be, it's thought, exceeded by the payroll savings from firing workers.

David Flynn wrote:

Wow this was informative and interesting!!