Ask most sales leaders what AI has changed and you get a list of tasks: research, list building, sequence copy, call summaries, CRM notes, forecast commentary. All of that is real, and all of it is genuinely useful. It has removed hours a week from a seller's diary.
Now ask the same leaders which of their deals were won by any of it. The answers get vaguer, because that is not where deals are decided.
The line is judgement, not effort
Automation is very good at work that is repetitive and where the right answer is knowable in advance. Selling is neither. A buyer's real objection is rarely the one they say out loud. The person who has to sign is often not in the meeting. The reason a deal stalls is usually political, not commercial.
These are the skills that remain stubbornly human.
- Understanding another person, including the parts they have not said.
- Asking a question nobody else in the process has asked.
- Listening properly, which means changing what you were going to say next.
- Building trust slowly, in a way that survives a bad week.
- Reading a room, or a silence on a call.
- Communicating a complex idea simply, to someone who is not technical.
- Genuine curiosity about a business that is not your own.
- Handling an objection without becoming defensive.
- Influencing a decision you are not in the room for.
- Resilience when the quarter is not working.
- Taking coaching, and actually changing.
- Understanding what motivates the person opposite you.
- Creating a commercial relationship rather than a transaction.
- Closing.
Why this matters for hiring, not just for selling
If AI has compressed the mechanical part of the job, the proportion of a seller's value that comes from the human part has gone up, not down. That should change what you test for.
Most interview processes still spend the majority of their time on the compressible half: what tools have you used, how did you build a list, what does your day look like. Those questions are easier to ask and easier to score. They are also increasingly the part of the job you can hire around.
How to test the part that matters
Discussing selling is not the same as observing it. Replace one competency interview with a live discovery conversation against a realistic scenario, briefed properly in advance. Score four things, in writing, with two interviewers:
- Question quality. Did they ask something you had not anticipated?
- Listening ratio. Did the answer change the next question?
- Thread-following. Did they chase the interesting thing or return to the script?
- Summary accuracy. At the end, did they describe your problem better than you did?
Then ask for a one-page follow-up. Look for whether they captured the actual problem, proposed a sensible next step, and wrote like a person rather than a template.
The uncomfortable one
Ask every candidate about a deal they lost that they should have won. People with genuine selling ability answer this quickly and specifically, and the answer is usually about something they did. People who have never really examined their own performance blame procurement, pricing or the product.
The short version
Technology has changed how selling gets done. It has not changed why anyone buys. Hire for both, and weight the second one more heavily than your current process probably does.