Last week, filling idle time one evening, I decided to conduct an experiment with ChatGPT. Earlier in the day, I had been looking at a ChatGPT led training program for sellers. As part of the program design, it suggested it engaging sellers in role plays.
I was curious about how that might work, so I decided to experiment. After describing a potential selling situation, I suggested a role play, where ChatGPT was the seller and I was the customer.
You can see my adventures with this role play in ChatGPT As A Seller! See ChatGPT Pitch! It was a fun, and funny exercise.
But at the same time, we should be alarmed by what it tells us about how far off target we are in how we sell, how we engage our customers.
Despite how patient I was as a customer, despite how much I tried to “coach” ChatGPT to engage me in a discussion about the issues that most concerned me (as customer), it would always go wrong and fail–miserably.
At first, as we begun the simulation, it wouldn’t allow me to get a word in edgewise. It presumed it knew my issues and presented how it’s solution would address them. It didn’t probe to ask if those were my concerns it just rolled on talking about what it thought I should be interested in.
We went back and forth, it would apologize then revert to the same behavior.
Finally, I said, “Would you please take the time to understand my issues and probe how they impact me?” I suppose I was the “Dream Customer,” for ChatGPT, most sellers would be thrown out of the customer’s office.)
It tried, I identified a specific issue, It asked me one probing question to understand the impact of that on me. And then, rather than probing and exploring it more deeply, or identifying other issues, it went on to pitch it’s solutions to me.
In one final effort, I asked ChatGPT why I should buy, what the impact of solving my problem might be on my results. And again, it failed. It didn’t understand how to develop and present value.
While it was a fun and interesting session, and I went back and forth, trying to see how it might change it’s approach, it was very consistent in it’s responses. It struggled to focus on understanding my issues, probing them, helping me better understand the impact–it would constantly pitch it’s solution.
How did ChatGPT come up with those responses? What had ChatGPT trained itself on, to be so off target in it’s approach?
Clearly, whatever data ChatGPT has access to, in it’s massive databases, it so the dominant patterns of how to engage customers in conversations, stated differently, it had trained itself based on the dominant patterns it saw in “sales conversations.”
Despite all the patience I, as customer, displayed, despite all the coaching I provided, it persisted in focusing on pitch driven discussions–not focusing on the customer concerns, but more on what it had seen, in it’s analysis, of what sellers are “supposed to do.”
We should be alarmed at what this demonstrates about how far off base our customer and “selling” conversations are!
We’ve years of thoughtful research giving us data about this problem, yet we persist in the same behaviors. It’s no wonder we drive customers to digital and other sources when they consider buying.
We should be alarmed that powerful tools, like ChatGPT, analyzing terabytes worth of data sees the dominant patterns of “selling conversations” focused on seller self interest and not helping buyers. ChatGPT pattern recognition is basically a mirror of our selling behaviors.
We experience plummeting response rates, plummeting win rates, plummeting customer engagement, plummeting quota attainment. Gartner has been posing the question on customer preference for “rep free” buying experiences for several years. And in that time, it’s gone from roughly 40% preferring rep free buying experiences to well over 70%–and I suspect it’s continuing to increase.
When are we going to start recognizing that our current approaches are broken?
While we may be achieving our goals, the costs of doing this are skyrocketing. Alternatively, we may be underperforming our potential—doing a disservice both to our customers, our companies, our shareholders.
And worse, we are enthusiastically jumping on the ChatGPT bandwagon, leveraging it to help us sell, whether through writing emails, content development, script development. It’s trained on the wrong engagement strategies and can only produce output that continues those failing strategies–at scale.
As you reflect on this try two things. First, try a role play with ChatGPT, yourself. Pose a rea;istic scenario, see how ChatGPT sells you. Is that the way you want to be sold to? How much does it mirror what you and your people are doing?
Think about what you might do differently, what do you change, how do you engage customers in ways that are more impactful. Test those ideas with ChatGPT–through a clever series of prompts, you can actually get some great insights.