June 2026 · Anthony Sogg

Don't Let a Robot Be Your New Greeter and Ambassador

Forrester says a third of brands will automate away their own trust in 2026. They'll do it at the one moment they can't afford to.

Forrester made a 2026 prediction that reads like a dare: one in three companies will erode customer trust by rolling out self-service genAI before it's ready, in the exact contexts where it's least likely to work. The same forecast warns of a 20% jump in privacy-related class-actions as AI, breaches, and regulation pile into each other. Treat that as the weather report for the year ahead.

I've said before that capability and trustworthiness are two different measures, and the gap between them is the whole job. As a consumer, and customer experience "expert" (ew, when did that happen? How did I get here? When did I become the adult? This is not my beautiful house, this is not my beautiful wife…and the days go by…sorry, I digress) poor, self-service anything, especially customer support, is one of the many horsemen of the marketing apocalypse.

Humans are, as I've mentioned before, irrational and emotional consumers. And that's good for brands and marketers. Marry that with cost cutting efforts, poor deployments, and the sterility of a genAI bot and you're actively killing your conversion and retention opportunities.

Whether the emotion is excitement or frustration, reading that emotional context is vital to keeping a customer for life. It could be the difference between upselling or preventing defection.

Think of how many times you, yes, you dear reader, have been shopping and came close to buying the wrong thing for your need because the product information was unclear or the difference was too subtle. A person was likely able to help you with that and understand the context and implied needs to help get you to your answer.

Flip side, how many times has something gone wrong, you're frustrated or sad because you've been so excited to get that thing or service and it let you down? Customers often don't need good will, though it certainly helps and is worth it in the long run, but someone acknowledging the issue, commiserating, and making them feel seen, validated, and taken care of goes so much farther than the costs cut from a poor CX experience.

Those experiences? That's the front door of your brand house. That's your greeter, your ambassador, your representative and agent. Remember when agent meant acting on behalf of? Do you really want a cheap bot being that agent for your brand?

The savings are this quarter. The damage is next year.

The driver here isn't customer value. It's cost. And that's bad.

We've embraced the Jack Welch cost cutting philosophy of making our numbers look better because it's "good business" which means "it's an easy loophole". Newsflash, it's not good business. It's lazy and it sounds like the starting gun for a race to the bottom.

An agent that deflects a ticket or qualifies a lead looks like money saved on a spreadsheet, so the pressure is to bolt it onto the first hello and walk away. Forrester is blunt about it: cost-cutting is exactly what shoves these tools into the rooms where they fail.

But first contact is the one moment you cannot fumble. Ask Zefram Cochrane and the Vulcans, IYKYK.

It's where a prospect decides whether you're worth their time, and where a burned customer decides whether they're done with you. We are taking the highest-stakes, lowest-patience moment in the entire relationship and handing it to the part of the stack we trust least, because that is where the math looks prettiest.

The savings show up now. The damage shows up late. Short term wins at the expense of long-term stability. That lag, and our inability to defer gratification, is the whole reason it keeps getting approved.

Trust doesn't show up on the invoice

So the dashboard looks green, but that's because we've made it look that way. It's no different from every younger car enthusiast erasing OBDII codes or putting duct tape over the check engine light. If I can't see it, it's not broken.

The reality? It's a bank ledger.

Every interaction either deposits trust or withdraws it, and a bad automated hello is a withdrawal you make before you've banked a single deposit.

Great, the number of tickets is at an all-time low. You know what else is starting to trend that way? The people who quietly decided not to come back, which is why that bill stays invisible right up until retention sags two quarters later and nobody thinks to blame the chatbot.

And there's a second bill behind the first. Forrester's class-action line is just the legal version of the same mistake: an agent at the door, moving fast, reading consent flags that it was never properly wired to honor. A suppression rule skipped. Data usage nobody thought through, let alone agreed to.

The seam between what the agent can do and what it's allowed to do is where the lawsuit lives, and at the front door that seam is wide open with nobody watching it.

Put it behind the human, not in front

None of this is anti-AI. It's about sequence, collaboration, and intent.

Park the agent where mistakes are cheap and reversible: behind the human, informing and mining insights, teeing up the next-best-action, briefing the rep before they pick up.

Let it earn its way forward.

The brands that land in the trustworthy two-thirds won't be the ones who refused to deploy. They'll be the ones who refused to react without a real strategy and analysis of the impact of that deployment. Their bumps in the road mean a mulligan between friends built on trust, not a quiet end to an entire relationship.

The case for automating the first hello is the prettiest slide in the building, because the savings land this quarter and the damage lands next year. A third of companies are going to take that trade in 2026.

The question was never whether the AI can handle the front door. The real question is what are the (in)tangible costs of a mistake, who gets to pay that bill, and will there be a chance to make another mistake thereafter.

Anthony

← More writing · If this was useful, or wrong in an interesting way, tell me: hireme@crmjedi.com.