A few years ago, I sat in on a support call review for a mid-sized appliance brand.

The agent spent eleven minutes walking a customer through how to reset their smart oven's Wi-Fi connection.

Eleven minutes. For a Wi-Fi reset.

The customer wasn't frustrated because the process was hard. They were frustrated because nobody had told them how to do it before they had to call someone.

That moment stuck with me. Because it's not really a story about one oven. It's a story about how support works almost everywhere now.

The scaling problem nobody talks about

Here's the uncomfortable truth: support teams don't scale the way the rest of the business does.

Sales scales with better targeting. Marketing scales with better channels. But support? Support scales by hiring more people to answer the same ten questions, over and over.

Every new customer is a new set of "how do I" moments. Every new product feature is a new wave of confusion. And every market you expand into brings new languages, new expectations, new time zones.

Most companies respond by throwing bodies at the problem.

That works—until it doesn't.

The hidden cost of rising ticket volumes

Ticket volume is the metric everyone watches. But it's not the real cost.

The real cost is what happens around the ticket.

A customer who can't figure out a feature doesn't always file a ticket. Sometimes they just... stop using the feature. Or the product. Quietly.

No complaint. No refund request. Just churn that shows up three months later with no clear cause.

I've seen companies spend months trying to figure out why retention dipped, when the answer was sitting in their support logs the whole time: a spike in confused questions about one specific feature, right before usage of that feature flatlined.

Tickets are the visible tip of the iceberg. The silent drop-off is the part underwater.

Why FAQs and knowledge bases stopped working

Knowledge bases made sense in 2012.

The problem is that they assume something that's increasingly untrue: that customers know what to search for.

If you don't know your product has a "calibration mode," you're not going to search for "calibration mode."

You're going to search for "why is this thing making a weird noise," get zero useful results, and call support.

Knowledge bases are built around the company's mental model of the product. Customers operate from their own mental model—usually built around whatever problem they're staring at right now.

That mismatch is where most self-service fails.

So what actually works? AI Product Guidance

This is where AI Product Guidance comes in—and it's genuinely different from a chatbot bolted onto a help center.

AI Product Guidance means the AI understands the product itself: its features, its common failure points, its setup flows, its quirks.

Not just a database of articles. An actual model of how the product works and how people get stuck using it.

So when a customer says "the light keeps blinking and won't stop," the system doesn't search for the word "blinking."

It recognizes that as a known error state, understands what causes it, and walks the person through the fix—conversationally, in plain language, right where they are.

Catching problems before they become tickets

The companies doing this well aren't just answering questions faster.

They're answering questions before the customer thinks to ask.

Picture a customer setting up a new device. Three steps in, usage data shows they've stalled—same screen, same step, for two minutes.

A well-designed AI layer notices that pattern and gently offers help. Not a popup that screams "NEED HELP??" Just a quiet, contextual nudge: "Having trouble with this step? Here's what most people miss."

That's the shift. From reactive support to support that meets people exactly where confusion happens.

Conversational AI, voice AI, and the death of the static manual

Manuals are written once and read by almost no one.

Conversational AI flips that. Instead of a 40-page PDF, the customer just asks a question in their own words and gets an answer shaped for their exact situation.

Voice AI takes this further—especially for products used in hands-busy, screen-unavailable moments. Think kitchen appliances, cars, industrial equipment.

Nobody wants to stop mid-task to read a screen. But asking a question out loud while your hands are covered in flour? That's natural.

The brands getting this right aren't replacing human support. They're removing the friction that used to exist before a human ever needed to get involved.

Onboarding: where the real ROI hides

If I had to point to one area where AI Product Guidance pays off fastest, it's onboarding.

The first 48 hours with a product shape everything that comes after. Confused early, churned early. Confident early, retained for years.

I've watched companies cut early-stage support tickets dramatically just by giving new users a conversational guide during setup—one that adapts based on what the person is actually doing, not a generic checklist everyone gets.

The result isn't just fewer tickets.

It's customers who actually use the product they bought, instead of letting it sit half-configured in a drawer.

A practical example: the "second device" problem

Here's a scenario I've seen play out across multiple consumer electronics brands.

A customer buys a smart home device. Setup goes fine. Three months later, they buy a second one.

Now they're trying to add it to an existing setup—a completely different flow than first-time setup, but most support content doesn't distinguish between the two.

The customer gets stuck, searches the knowledge base, finds first-time setup instructions, follows them, and breaks their existing configuration.

Now you've got two problems and one very annoyed customer.

AI Product Guidance that understands context—this customer already has a device, this is an "add a second one" scenario—can route to the right flow immediately. No ticket. No broken setup. No annoyed customer.

That's not a hypothetical. That's a Tuesday.

Where companies get this wrong

I want to be honest about the failure modes, because there are plenty.

Mistake one: treating it like a chatbot project.

If the goal is "reduce headcount," the AI gets trained to deflect, not to help. Customers feel that instantly, and trust evaporates.

Mistake two: launching without product knowledge.

A conversational layer with no real understanding of the product is just a fancier search bar. It'll confidently give wrong answers, which is worse than no answer.

Mistake three: ignoring the handoff.

When AI guidance can't solve something, the handoff to a human needs to be seamless—with full context carried over. Nothing destroys trust faster than repeating your entire problem to a human after the AI already "understood" it.

Mistake four: set-and-forget.

Products change. New features ship. If the AI's understanding doesn't update alongside the product, it starts giving outdated guidance—right when customers need it most, during a new feature's rocky early days.

What this means for customer experience, long term

Step back, and the pattern becomes clear.

The companies pulling ahead aren't the ones with the most support agents.

They're the ones where customers rarely need an agent—because help showed up at the moment of confusion, not twenty minutes later in a queue.

That shift changes how customer experience teams are measured. Ticket count becomes less important than "time to resolution" and "did the customer ever need to ask."

It changes product teams too. Confusion patterns surfaced by AI guidance become a feed of product feedback—every recurring question is basically a usability bug report, delivered in real time.

And it changes retention. Not through grand gestures, but through dozens of small moments where something that used to be frustrating just... wasn't.

The quiet shift

Nobody posts on LinkedIn about the support call they didn't have to make.

There's no celebration for the oven that connected to Wi-Fi on the first try, or the new user who breezed through setup without ever opening a help article.

But that absence—that quiet, uneventful experience—is exactly what's separating companies that scale gracefully from companies that scale painfully.

The best customer experience is the one nobody talks about, because nothing went wrong.

So here's the question worth sitting with: how much of your support volume today is really about answering questions—and how much of it is about your product failing to answer those questions on its own?