Symptoms
Some customer experiences are like that one person in the meeting who is beautifully dressed, deeply confident, and quietly wrong. On paper, everything looks respectable. The journey is familiar. The numbers are still decent. The complaints have not yet risen to biblical levels. Customers are still using the service. Leadership is still sleeping at night. Nobody is running through the corridor waving a ruined dashboard and yelling that the sky has split open.
So the business relaxes. It tells itself a soothing little bedtime story. We are stable. We are solid. We are performing. We are, apparently, fine.
This is usually the part where I start narrowing my eyes.
Decline rarely arrives wearing a silky cape and carrying a fog machine. More often, it slips in through the side door with a clipboard and a collection of small irritations. A few more explanations than there used to be. A few more defects clustering around the same points in the journey. Add a dash of customers who are not furious exactly, just tired. And a healthy helping of teams quietly stitching the experience together with workarounds, verbal duct tape, and the organisational equivalent of “just jiggle the handle a bit.”
Nothing dramatic. Nothing headline-worthy. Just that faint but growing smell of strain.
And this is where comfortable companies get themselves into trouble. They assume that because the experience is still broadly accepted, it is also broadly safe. They look at the centre of the dashboard, nod approvingly, and miss the muttering at the edges.
The edges, inconveniently, tend to know first.
Diagnosis
The core mistake is gloriously common and wildly expensive.
Organisations confuse familiarity with fitness.
An experience can be well known, broadly adopted, and still quietly becoming easier to outperform. It can still be functioning while becoming more effortful, less elegant, more brittle, and strangely dependent on human heroics nobody has budgeted to honour properly. Customers may still use it. Teams may still manage it. The metrics may still flatter it. That does not make it future-proof. It simply means it has not yet been forced into an honest comparison with something better.
That is the sting in Andy Jassy’s point. If AI is set to reinvent customer experience, then even the experiences people already know, already use, and may even broadly prefer today are not safe. In fact, they may be the ripest targets of all. Not because they are visibly disastrous, but because they have grown comfortable enough to avoid scrutiny.
This is why averages can be such charming little fraudsters. They smooth everything out. They make strain look civilised. They turn awkwardness into percentages and recurring irritation into “nothing material at this stage.” Meanwhile, out in the wild, the real clues are gathering. The repeat contact that still looks too minor to escalate. The policy explanation that grows longer every month. The customer effort hidden behind a technically successful interaction. The tiny defect that keeps reappearing like an ex who still thinks they deserve a key.
A comfortable experience is not always a strong one. Sometimes it is just an ageing one with very good lighting.
Resistance
So why do companies cling to these experiences like Victorian widows clutching pearls?
Because comfort is flattering. It tells leaders they are safe when what they really are is unchallenged. It whispers to teams they are coping when what they are actually doing is compensating. It declares to the organisation that broad adoption equals durability, when often it just means customers have not yet been given a sufficiently seductive alternative at scale.
And let us be honest, there is something emotionally awkward about questioning an experience that is still performing “well enough.” Nobody wants to be the person dramatically clearing their throat in the meeting and suggesting that the apparently healthy customer journey may in fact be wearing a corset over a slow internal collapse. It feels impolite. Premature. A touch inconvenient.
Then there is the human habit of adapting around nonsense. Teams are remarkably good at this. They explain more. Patch more. Rescue more. Smooth over more. They become so practised at keeping an ageing experience presentable that the business starts mistaking their effort for proof that the design itself is still sound.
It is not sound. It is being lovingly held together by people who deserve flowers, bonuses, and perhaps a brief lie-down.
The company sees continuity. The frontline sees glue.
Incision
So what do sensible adults do about this before the market arrives with a chair and starts rearranging the furniture for them?
First, they learn to read the edges properly. Not every anomaly deserves a panic attack and a themed war room. But repeated anomalies deserve respect. If the same frictions keep surfacing in the same moments, if the same workarounds keep saving the same journeys, if the same customer confusions keep requiring interpretive dance from the frontline, then the experience is trying to tell you something. Weak signals are only weak when you insist on pretending they are unrelated.
Second, defect elimination has to grow up a little. Too many organisations still treat defects like crumbs to be brushed off the table before visitors arrive. Tidy the issue. Close the loop. Move on. Lovely. Useless. A defect is not just an annoyance. It is a clue. It tells you where the design is under strain, where the journey is asking too much effort from customers or staff, and where “still working” is beginning to lean suspiciously hard on manual rescue.
This is where I become deeply annoying and go full Sigma. Not because I enjoy tormenting executives with pattern analysis, though there are worse hobbies, but because defects are directional. Low-volume irritants matter when they cluster. Repeated friction matters when it points to the same structural weakness. What looks minor in isolation can be a rather elegant warning in aggregate.
Third, stable metrics need to be pressure-tested against lived reality. Do not just ask whether the numbers still look acceptable. Ask what they are hiding. What is being absorbed before it becomes visible. What is becoming more effortful while still appearing statistically respectable. Which parts of the journey technically function but increasingly feel like an obstacle course designed by committee.
Fourth, the strategic question itself must improve. Stop asking only, is this experience still working. That is a lazy question and usually receives the exact level of answer it deserves. Ask instead: how vulnerable is this experience to reinvention? How easy would it be for a competitor, a smarter design, or an AI-enabled journey to make this experience suddenly feel clunky, effortful, and faintly ridiculous by comparison? That is the question with teeth.
And finally, put the signal into the plan. If the edges are showing strain, it belongs in improvement work now, not once the pain becomes socially impressive. Investigate the recurring friction. Map the repeated defects. Pressure-test the assumptions. Review the journeys that are still numerically fine but increasingly held together by customer patience and frontline charm. Do not wait for the customer experience to become embarrassing before taking it seriously. That is not foresight. That is procrastination in business casual.
Prognosis
The organisations that do this well become much harder to surprise. Not because they possess mystical powers or a secret Oracle hidden behind Finance, but because they stop requiring pain at scale before they allow themselves to think. They get less hypnotised by stable averages and more interested in subtle strain. They treat comfort as a condition to interrogate, not a strategy to admire. They understand that “customers still use it” is not the same as “customers will still prefer it.”
That changes everything.
Defects stop being housekeeping and start becoming intelligence. Frontline observations stop being operational grumbling and start becoming strategic signal. Improvement work gets sharper because it is tied to future vulnerability, not only past failure. Leadership gets more credible because it is no longer waiting for obvious collapse before asking better questions.
And perhaps most importantly, the business stops making customers and frontline teams be the first to feel what leadership could have noticed earlier.
Because that is the real point, is it not?
Comfort is lovely as a feeling. I enjoy it. You enjoy it. A nice blanket, a warm drink, a customer journey that is not actively on fire. Splendid.
But comfort is far less impressive as a strategy.
The most dangerous customer experiences are often not the visibly broken ones. They are the familiar ones. The ones still broadly accepted. The ones still producing just enough decent data to avoid a proper interrogation while the edges are already whispering that something better could replace them with alarming ease.
Comfort is not protection. Sometimes it is simply the last polite phase before reality stops knocking and lets itself in.