The Enshitification of the Real Estate Stack

A term that has been doing the rounds in tech for a while now is enshitification. The basic idea is simple enough. A platform starts out by being useful. Then it gets commercial. Then, once enough people are locked in, it starts leaning harder on both sides of the market to extract more value than it gives back.

The reason I think that is relevant for real estate is that Australia already has a stack that looks very vulnerable to exactly that pattern. The portal layer is the most obvious example, but it is not the only one. Underneath it sits a CRM and workflow layer that has become expensive, fragmented and clunky, and now AI is starting to become another interface between the user and the market, just as monetisation pressure is arriving there too. If you step back and look at the whole thing together, the same shape starts to appear again and again.

The portal layer is the easiest place to start because the pricing has become impossible to ignore. ABC reported last year that one Melbourne seller paid $5,880 in marketing, of which $3,289 went to listing on realestate.com.au. The same report said some agents estimate REA controls more than 80% of the listing market, with premium listings in some areas now costing more than $5,000 and annual increases commonly running at 8 to 10%. REA told the ABC its average listing price increase that year was 7%, but that still leaves the broader point intact... this is a market where the dominant player can keep putting prices up because everyone knows sellers and agents feel they have to be there.

That dynamic had already been described pretty bluntly by former REA chief executive Simon Baker. In The Guardian, he said the vendor-paid model in Australia lets the portal treat agents as an outsourced sales force, while the seller ends up paying the advertising bill. His line was that this creates an ecosystem where the price just keeps going up, and up, and up. Whether you like the language or not, it is hard not to see the underlying structure for what it is.

And the market power behind that structure is now under real scrutiny. Reuters reported in May last year that the ACCC had launched a probe into REA Group, while ABC said the regulator was examining whether the company was using its dominant position to the detriment of consumers and engaging in anti-competitive behaviour. Once a regulator starts looking at the market this way, the story stops being just about expensive listings, it becomes a question of whether the system itself has become too concentrated to discipline pricing properly.

Even the supposed competitive response feels like part of the same broader problem. Industry reporting in March said Domain was capping some price increases at 4% and bundling extra value such as Matterport 3D imaging and drone photography into higher-tier products. Around the same time, one of the better-funded recent challengers, View.com.au, was reported to be closing in June after failing to break the REA-Domain duopoly. In other words, the market is concentrated enough that even the competition story can end up looking like a softer version of the same extraction logic.

Underneath the portal layer sits the software and workflow layer, and I think that part of the stack has its own version of the same disease. A PIPA article from January put it well. PropTech did not exactly fail, but it evolved in isolation. CRMs manage leads but not communication, marketing tools run campaigns but not transactions, reporting systems capture data but rarely interpret it, and the result is a disconnected set of applications that leave agents spending more time managing technology than managing clients. The same piece says the average property professional is spending between eight and twelve hours a week duplicating work across platforms. That is a lot of labour just to hold the machine together.

That is where this starts to feel bigger than just a portal story. In consumer tech, enshitification often looks like worse search results, more ads and more nudging. In real estate software, it often looks like fragmented systems, expensive subscriptions, feature bloat, duplicated entry and workflow that somehow gets heavier the more software you add. The sales pitch is usually about efficiency, but in practice a lot of businesses still end up re-entering the same information, moving it between systems, and paying multiple vendors while doing it. The stack becomes more commercial, but not necessarily more intelligent.

Now AI is starting to become another layer in that same stack. REA launched a realestate.com.au app inside ChatGPT in February, which means users can now search live listings directly in ChatGPT using natural language. That may well be genuinely useful, and I suspect it will be, but it is also a sign that the front door to property search is shifting again, and that the next layer of the stack is already being built.

At the same time, OpenAI announced in February that it was testing ads in ChatGPT, and on 26 March said it would expand that pilot beyond the US into Canada, Australia and New Zealand in the coming weeks. Reuters then reported that the US pilot had already crossed a $100 million annualised revenue run rate within six weeks. I am not making a moral argument here about whether a company should monetise. The point is that once the interface becomes commercially important enough, the pressure to monetise tends to increase, and the user experience rarely stays untouched forever. That is the pattern people are worried about in consumer AI, and it is hard not to wonder how quickly the same logic could start shaping AI-powered property search, lead routing and recommendation layers as well.

That, for me, is what makes this a useful real estate story rather than just a borrowed tech word. The portal layer is already showing what happens when a two-sided market becomes concentrated and the tollbooth realises it has pricing power. The CRM and workflow layer is already showing what happens when software becomes fragmented, overbuilt and labour-intensive to operate. And the AI layer is arriving just as monetisation is becoming explicit there too. Put all three together and the risk is not simply that one part of the stack gets worse, it is that the whole thing slowly becomes optimised to extract more value from agents, vendors and users without actually making the work much cleaner.

I think that is the more interesting question now. Not whether one portal is a bit cheaper than the other, or whether one CRM has a few more features on the pricing page, or whether one AI interface feels a bit slicker this month. The deeper question is whether the real estate stack is gradually being shaped around extraction rather than simplification. Because if it is, then the opportunity is not just for a better tool, it is for a better architecture altogether.