There is a moment, if you pay attention, when you can feel a product starting to turn.

It is hard to pinpoint exactly. The product still works. The core experience is still recognisable. But something has shifted in the texture of using it. There is more friction than there used to be. More noise between you and the thing you came to do. More of the experience serving someone other than you.

You have probably felt this recently with search.

Google was transformative for a reason that is easy to forget now: it made finding information on the internet straightforward. You typed a question, you got results, you went to a page that answered it. The transaction was clean and fast and felt almost miraculous compared to what preceded it. It became a verb because it described something genuinely new in human experience.

Now try using it to find a straightforward answer to a practical question. You will navigate through sponsored results at the top, AI summaries of uncertain reliability in the middle, and a collection of pages that have been engineered to rank rather than to inform. When you eventually reach a website, you will frequently encounter the same pattern: cookie banners, newsletter popups, ad placements, and paragraphs of SEO-optimised preamble before anything resembling the answer you were looking for.

The thing that was built to cut through the noise has become noise.

And then comes the AI alternative, and suddenly we hit the reset button.

The arrival of conversational AI as a search replacement is not, when you look at it clearly, a story about artificial intelligence. It is a story about a pattern that repeats itself with remarkable consistency across the history of consumer technology.

A product emerges to solve a problem that existing solutions handle badly. It solves it well, with a clarity and simplicity that the incumbent cannot match, partly because the incumbent is burdened by years of accumulated complexity, business model compromises, and organisational inertia. The new product grows. Users love it. The simplicity that made it lovable is the thing that defines it.

And then the business pressures arrive. Revenue needs to be generated at a scale that the simple, clean experience cannot support. Advertisers discover the audience. The algorithm that once served users starts to serve multiple masters simultaneously. Each individual compromise seems reasonable in isolation. The cumulative effect, experienced over months and years by users who remember what the product used to feel like, is the transformation of something they loved into something they merely use, and eventually something they actively resent.

The AI tools are already on this path. Ads have arrived in ChatGPT. They are relevant to your query, subtly integrated, easy to rationalise as useful. They will not stay subtle, and they will not stay few. The economic logic that turned Google's clean search box into its current form is the same economic logic now operating on the AI layer. The timeline may be different. The destination is not.

What is being described here is not a new observation. Economists and technology historians have noted versions of it for decades. But it rarely gets discussed in product circles with the seriousness it deserves, perhaps because acknowledging it requires a degree of honesty about where most products are headed that is uncomfortable to sit with.

The pattern works like this. A new entrant identifies the gap between what an existing category promises and what it delivers. It builds the promised version, without the incumbent's accumulated baggage. Users reward it with loyalty and growth. That growth attracts investment and revenue expectations. Meeting those expectations requires compromises that erode the original experience. The erosion creates the gap that the next entrant will identify and promise to close.

Uber identified the gap between what urban transport promised- reliable, fairly priced, convenient- and what the taxi industry delivered, and built the cleaner version. For a period, it was genuinely better: faster to book, easier to pay, more predictable in price and quality. Now, depending on where you are and when you are travelling, the price gap between Uber and a traditional taxi has largely closed; the surge pricing mechanics produce frustration that mirrors the worst of the old system, and the driver experience that underpins the whole thing has its own set of problems that have attracted exactly the kind of regulatory scrutiny the original model was designed to circumvent.

Airbnb told a similar story. It began as a genuine alternative to hotel stays: cheaper, more personal, more embedded in the local character of a place. Now, in most major cities, the prices are comparable to mid-range hotels, the cleaning fees are a source of widespread frustration, the cancellation policies have become increasingly restrictive, and the platform's relationship with the cities it operates in has become adversarial in ways that have attracted significant regulatory intervention. The gap between Airbnb and what it was built to replace has narrowed considerably.

This is not a coincidence. It is a structure.

The easy narrative is that companies start with good intentions and are corrupted by the pursuit of profit. That narrative is not entirely wrong, but it is incomplete in ways that matter for anyone trying to build something that lasts.

The more accurate picture is that most of the individual decisions that degrade a product experience are defensible in isolation. Adding one ad placement to a search results page generates meaningful revenue and does not, by itself, ruin the experience. Raising prices slightly in response to increased operational costs is rational. Adding a feature requested by a large enterprise customer makes business sense even if it adds complexity for individual users. Complying with a regulatory requirement introduces friction that nobody wanted.

The problem is that these decisions accumulate, and their cumulative effect is felt by users who experience the product as a whole rather than as a series of individual trade-offs. The user does not know that the fifth ad in their search results is there because of a specific revenue target from a specific quarter. They know that finding what they are looking for is harder than it used to be.

By the time the degradation is visible enough to produce genuine user backlash, the product is usually too embedded in the business model changes that drove it to reverse course without high cost. The ads are load-bearing. The price increases are baked into investor expectations. The complexity is woven into the architecture. The product that was once simple cannot easily become simple again.

There is something worth dwelling on in the rhythm of these cycles, because it carries a cost that rarely gets accounted for properly.

Each time the reset button gets pressed, users invest trust in a new product. They learn it, integrate it into their workflows, recommend it to others, and build habits around it. When the product degrades, that investment does not simply disappear. It turns into scepticism, which is harder to overcome than ignorance. The user who was burned by the last clean product that became cluttered is not as easily won by the next clean product that promises the same thing.

The aggregate effect of enough cycles is a population of users who are increasingly cynical about new product promises, because experience has taught them that the promises are temporary. The clean search is this year's Google before it became last year's Google. The AI assistant that answers your questions directly is this year's experience before it becomes next year's experience.

That cynicism is a real and underappreciated product challenge. Building trust with users who have been through enough cycles to recognise the pattern is harder than building trust with users who have not. And the window of genuine user enthusiasm, the period before the pattern becomes obvious to the people experiencing it, is shorter each time.

I want to be clear that the purpose of this analysis is not to produce fatalism. The pattern is real, and it is powerful, but it is not inevitable in its specific form, and the speed and severity of the degradation are not fixed. What happens depends significantly on the choices that product teams make, particularly the choices made during the growth phase when the pressure to compromise is highest, and the consequences of compromising are least immediately visible.

There are several things a PM can do that meaningfully affect the trajectory.

The first is to name the pattern explicitly in strategic conversations. Most product organisations discuss monetisation, growth, and user experience as separate workstreams with their own goals and metrics. The degradation cycle is what happens when those workstreams optimise independently without accounting for their cumulative effect on the core user experience. Making the cumulative effect visible, as a real strategic risk rather than a soft concern, gives the organisation a basis for making different trade-offs.

The second is to treat simplicity as a feature that requires active maintenance rather than a baseline that can be assumed. Simple products do not stay simple by default. Complexity accumulates through hundreds of small decisions, each of which seems reasonable in isolation. Protecting simplicity requires someone who is explicitly accountable for the overall experience, who can say when the cumulative weight of additions is starting to cost more than each individual addition is worth.

The third is to look for the sustainable version of what the product does well, before the business pressure forces an unsustainable version. The products that navigate this cycle best are usually the ones that find a revenue model that aligns with the user experience rather than working against it, before the misalignment becomes load-bearing. That search is harder and less immediately rewarding than the compromises that degrade the experience incrementally. It is also considerably more durable.

The fourth, and perhaps most important, is to stay honest about where the product currently sits on the curve. The instinct when a product is successful is to attribute the success entirely to what is being built and to discount the role played by the gap the product filled when it arrived. That instinct makes it harder to see when the gap is closing, and the product is starting to resemble the thing it replaced. Keeping that possibility genuinely in view, even during periods of strong performance, is what creates the space to respond to it before it becomes a crisis.

The pattern described in this article is not primarily a product failure. It is a structural feature of how markets, business models, and user expectations interact over time. Treating it as a failure tends to produce either despair or denial, neither of which is useful.

The more productive framing is that this cycle is the environment in which product management happens, and the PM's job is to navigate it with more awareness and more intentionality than the default produces. Not to eliminate the pressures that drive the degradation, because those pressures are real and legitimate, but to make the trade-offs consciously, to account for their cumulative effect, and to look for the version of the product that can sustain its core value over a longer arc than the typical cycle allows.

The reset button will keep getting pressed. The question worth asking is not how to prevent it, but how to build something that earns enough genuine user trust, and maintains enough of its original clarity, that the users who experience it remember what it was like when it was good and remain willing to give it the benefit of the doubt when the pressures arrive.

That is a more modest ambition than changing the cycle. It is also, probably, the achievable one.

A question worth sitting with: the product you are currently building, what was the original promise it made to users? And how much of that promise is still intact in the version that exists today?

The gap between those two things is where the most important product thinking usually needs to happen.