Researchers asked consumers a question this year that is worth reading twice, because the wording does all the work.
"If the outcome and the time spent were exactly the same, would you prefer to get customer support from a human or from AI?"
That is the question as SurveyMonkey put it, word for word.
Read what that question removes. It takes away speed, which is the entire pitch for automating support. It takes away the result, so nobody can say the bot got it wrong. It leaves one variable standing: who you were dealing with.
82% of consumers said they would rather have the person.
That number should stop you, because every argument for taking the human out of your business is an argument about efficiency, and this question already granted the efficiency. Same answer. Same wait. Four out of five people still want to talk to somebody.
They Controlled for the Only Excuse You Have
I want to be careful here, because "people prefer humans" is the kind of statement that sounds warm and means nothing. Of course people would rather have a person if the person is faster and better. That is not a finding, that is a preference for competence.
What makes this survey useful is that it closed off the escape route. The respondent is told, explicitly, that the machine is exactly as fast and exactly as correct. Every practical reason to accept automation has been granted in advance. And the preference does not move.
So whatever people are responding to, it is not performance. It is the fact of the person.
You can argue with the strength of the preference. You cannot argue that it is a misunderstanding about speed, because speed was removed from the question.
Your Whole Market Is Automating That Exact Layer
Now put that next to what businesses are actually doing with their year.
In the same body of research, three quarters of marketers said the technology matters more to their strategy now than it did a year ago. Ask where it has already improved their workflows and the answers cluster in a revealing place: 92% say customer relationship management, 89% say marketing automation, 90% say advertising and personalization.
Read that list as a customer rather than as an operator. Customer relationship management is the record of your dealings with people. Marketing automation is the messages they receive. Personalization is how those messages address them by name. Every one of those categories is the layer where the business touches the human being.
That is not an accident, and it is not stupidity. It is where the obvious savings are. The whole industry looked at its costs, found the part that scales worst because it involves people, and pointed the tools at it. Your competitors are doing this right now, on the advice of people who are not wrong about the economics.
They are just all running at the same door, and the research says the customer is standing behind it holding up a hand.
They Already Know, Which Nobody Priced In
The assumption underneath most automation plans is that it will not be noticed. Do it well enough and the customer never finds out.
That assumption is already dead. 70% of consumers say they have noticed AI being used in marketing emails, advertising and customer service. Not suspect. Noticed.
And having noticed, they are not impressed. Fewer than one in four say they like or love seeing it anywhere in that list. Break it down and it gets sharper: roughly a quarter like or love AI in marketing emails, while 43% dislike or hate it. In customer service the split is worse, with 53% landing in dislike or hate.
There is a version of this article that treats that as a reason to hide it better. That is the wrong lesson, and the numbers on disclosure say so. Nearly three in five consumers say it is very important that companies tell them when AI is in play. The people who can already detect it are also asking to be told about it, which means concealment is the one strategy guaranteed to be found out and resented.
This is the same instinct we wrote about in the AI person in your ad isn't real. The problem was never that a tool was used. It is being caught presenting something as more human than it was.
The Layer You Are Automating Is the Layer That Decides Your Reputation
Here is the finding from that research I would actually write on the wall.
Consumers across every region surveyed were asked what influences their opinion of a brand. The number one answer, in every single region, was how well the company treats its customers.
Not the product. Not the price. Not the advertising. How they treat you.
So the sequence a lot of businesses are about to walk into is this. The thing that most determines what people think of you is how you treat them. The moments where you actually treat them are support, service, and the messages you send. Those are precisely the moments being handed to software this year, because they are expensive. And four out of five people would rather that not happen, and seven out of ten can tell when it does.
For a small business this is not a warning. It is the clearest competitive opening I have seen in a while, because you are the only kind of business that can credibly choose the other path. A company with forty thousand support tickets a week genuinely cannot answer them personally. You can. That was always true and it was never worth much, because nobody was making a point of the difference. Now everybody is, by going the other way.
What to Automate Without Guilt
None of this is an argument for doing everything by hand. That would be a bad business and it is not what the research supports. The useful move is to sort your touchpoints rather than take a blanket position, and after a lot of client conversations this year I have landed on a rule that holds up.
Automate the logistics. Nobody wants a relationship with your calendar. Appointment reminders, booking confirmations, delivery notifications, routing a form to the right inbox, chasing a missing document, data entry, and the entire back office are fair game. Customers do not experience these as being treated well or badly. They experience them as working or not working, and software is often better at them than a busy person is.
The same goes for everything the customer never sees. Reading your own data, drafting a first version, analyzing reviews, summarizing ninety days of search terms. We made the full case for pointing the tools there in stop asking AI to write your ads, ask it to read your account, and nothing in this research argues against it. A customer is not offended by how you prepared.
What to Keep Human, on Purpose
Then there are three categories where the person is the product, and the test is simple. Ask whether the moment involves judgment, money, or repair.
Judgment is any moment where the customer does not know what they need and is asking you to tell them. Which service is right for my situation. Whether this is worth fixing or replacing. What you would do. This is the conversation where your twenty years actually shows up, and a confident generic answer is worse than a slow specific one.
Money is quotes, pricing conversations, anything to do with a bill. People want a human accountable for a number. A disputed invoice handled by a chatbot does not read as efficient, it reads as a company arranging not to be reachable about its own charges.
Repair is when something has gone wrong. This is the one I would defend hardest. An apology from an automated system is not an apology, because the entire content of an apology is that a person took responsibility. Handle a complaint well and you frequently end up with a more loyal customer than if nothing had gone wrong at all, which is the argument behind your support inbox is a sales channel. Automate that moment and you have converted your best retention opportunity into a cost saving worth a few dollars.
Logistics, automate freely. Judgment, money and repair, keep the person. That single sort will do more for you than any policy about which tools are allowed.
An Advantage Nobody Can See Is Not an Advantage
Now the part most businesses will get wrong even if they get everything above right.
If you decide that a real person answers your phone and handles your quotes, and you never mention it anywhere, you have taken on a real cost for zero competitive return. The customer comparing three companies cannot see your staffing model. They assume the default, and the default is rapidly becoming a bot.
So say it. Put it where people are deciding. A line on the contact page that a person answers, with a name. A sentence in the ad. A note in the booking confirmation about who will actually call. This is the same category of signal as the one in your footer is the last place they check to see if you're real, and it works for the same reason: buyers are actively hunting for evidence that a real business is on the other end.
The disclosure finding makes this more valuable, not less. Three in five people want to be told when AI is involved. A business willing to say plainly where it uses software and where it does not is answering a question your competitors are hoping nobody asks.
And it earns you something in the platforms too. Google's Local Services Ads already rank on review quality and responsiveness rather than budget, which we covered in Local Services Ads don't reward the biggest budget. Treating people well has measurable distribution consequences now, not just reputational ones.
Be Honest About Who You Sell To
One caveat, because the number is not uniform and I do not want you acting on an average.
The preference for human contact runs hardest in older cohorts. 91% of Boomers said they prefer human support. Gen Z is measurably more open to automation, though still not enthusiastic about it. Trust in AI-generated answers is low everywhere, with no region showing more than about a third of people saying they trust AI summaries a lot or completely.
So if you sell home services to homeowners over fifty, this is close to a strategic mandate. If you sell software to twenty-six-year-olds, the picture is softer and a good automated flow may be genuinely welcome. Look at who actually buys from you before you decide how hard to lean on this.
We spend all day helping small businesses compete against companies with far bigger budgets, and the honest answer is usually that you win on the parts that do not scale. This year that stopped being a consolation and started being measurable. If you want help working out which of your touchpoints are worth protecting and how to say so where buyers will see it, that is what we are here for.
