If Airbnb were really “taking over London,” you’d expect to see it everywhere. That’s not what I saw when I tried to recommend a place near where I live in Islington. I zoomed in. Nothing. Zoomed out. Still nothing. Then suddenly, dozens of listings, clustered a few postcodes away. Not gradual. Abrupt.

That pattern mattered. Because what looked like “Airbnb taking over” wasn’t sprawl - it was pressure pooling in very specific places.

So I did what I usually do when something feels off: I pulled the data. I scraped every listing, mapped the coordinates, and let the spatial patterns speak.

What emerged wasn’t a story about Airbnb acting as a market-maker (as I’d previously found with Google Maps). It was something quieter and more uncomfortable.

The takeaway, upfront: Airbnb didn’t overwhelm London. It concentrated in places where housing scarcity, wage pressure, and planning gaps already made short-term letting the rational response. This isn’t a defence of Airbnb though. It’s a claim about causality: regulate the interface all you want, but if you ignore the underlying housing pressure, the problem will simply reappear in another form.

The platform didn’t create the problem. It found the gaps. At the individual level, Airbnb often does increase efficiency (call me an economist, just for this take…). Spare rooms get used. Travel gets cheaper. Households patch income gaps without moving. The conflict isn’t efficiency. It’s distribution - who bears the cost when individually rational behaviour aggregates into neighbourhood-level harm.

To see this, I built a dashboard map. It answers one question: *where does short-term Airbnb letting cluster? *In other words, where is the pressure of Airbnb felt the most?

Not to moralise. To diagnose. Because you can’t intervene in a system you can’t see - and right now, policymakers are regulating blind. And because individual choices still matter, I added an escape hatch: a layer that surfaces listings in low housing-pressure areas. Places where Airbnb still looks like housing, not infrastructure. High ratings. Reasonable prices. Outside the tourist-trap clusters.

Call it the guilt-free filter - or, perhaps more accurately, the here can you book to “not make things worse” layer.

Here’s what the map actual does and reveals - and why it changes the question from “How do we stop Airbnb?”

Before we go further: London - like most major cities - does regulate short-term lets. This entire debate goes wrong because London regulates Airbnb as a temporal problem when its impacts are fundamentally spatial.

Since 2015, Londoners can rent out their entire home for up to 90 nights a year without planning permission. The intent was reasonable: let people monetise a flat while on holiday, ease mortgage pressure, add flexibility to the tourism market.

Beyond 90 nights, you’re no longer operating a dwelling - you’re running "temporary sleeping accommodation," which requires explicit borough permission. In high-pressure areas, boroughs like Camden and Westminster, and Hackney routinely refuse these applications, with rejection rates around 60-80%. And Airbnb does enforce this. Since 2017, entire-home listings auto-lock at 90 booked nights. On paper, this is tidy. Cities love tidy.

But the system is porous: it tracks bookings, not actual occupancy. It can’t catch hosts listing across multiple platforms (Airbnb + Booking.com + Vrbo = 270 nights, technically legal), creating duplicate accounts, or false claims about permissions. And while Westminster has dedicated enforcement officers, most boroughs don’t.

You can have a street where every flat is listed for exactly 89 nights. Every host is compliant. And yet the street has functionally become a hotel corridor. Rolling suitcases. Keypad lockboxes. A constant churn of strangers. The law sees individual compliance; residents experience collective transformation.

This is the governance vacuum: the metric being regulated (time) isn’t where the disruption is happening (space and density). It’s why seeing and correcting the map matters more than counting the nights. The city is enforcing the law perfectly - it’s just enforcing the wrong law.

The public Airbnb map is fine if you want to book a weekend and ignore context. It shows availability, not impact. It tells you where listings exist, but nothing about why they concentrate - or what that concentration means for a neighbourhood.

So I built an adjusted version: the Airbnb pressure Map - showing where Airbnb pressure is high. The question this map asks is: where does that perfectly rational, legal behaviour aggregate into something that starts behaving like hotel infrastructure?

What the map does and why it matters:

It normalises density by housing stock. Raw listing counts lie. Shoreditch with 500 listings and Barking with 500 listings are not the same problem. One is a dense, already-saturated neighbourhood; the other is a sprawling borough where 500 is merely a rounding error. The map weights listings against actual housing units so you’re comparing pressure, not pins.

It discovers high Airbnb density clusters without assuming them. Rather than drawing boundaries by borough or branding (“creative quarter”, “vibrant hub”), the map lets the data speak. I use DBSCAN to create clusters which is non-parametric and indifferent to administrative lines. Parameters: eps = 0.003 (~330m at London’s latitude), min_samples = 5: roughly the scale at which repeated short-term lets change how a street feels to residents. Change the parameters and the labels move; the underlying pattern does not. Clusters render as convex-hull polygons with centroid bubbles sized by listing count - readable at city scale without drowning in ~80,000 individual pins. For texture, I sample up to 120 points per cluster and aggregate the rest. This all allows me to reveal the real reasons why clusters form.

It surfaces low-pressure listings. Finally, I added an escape hatch from the dense clusters: a pressure score combining ratings, price, and distance from high-intensity clusters highlights places where Airbnb pressure is low and still looks like housing, not infrastructure. The top ~1,500 appear as a toggleable layer when you zoom in. If Airbnb is efficient, this is where that efficiency stays local - without tipping neighbourhoods into churn. Worth your money, without the guilt. As a friend of mine says: no matter where you are in London, your destination is always 45 minutes away. Might as well stay somewhere that isn’t making things worse, e.g., Tower Hamlets perhaps?

How to read it: Darker shading = more pressure. Bigger bubbles = larger clusters. Toggle “low pressure listings” and zoom in to find the escape hatch.

Once you stop looking at raw listings and start looking at pressure, the story changes.

Airbnb is not "everywhere." The disruption is sharply concentrated. Most of London remains relatively stable; this is not a citywide takeover.

Hotspots don’t track regulation. If the 90-day rule were meaningfully shaping behaviour, density would be lowest where enforcement is tightest: Camden, Westminster, Hackney - boroughs that are explicitly restrictive and vocal about it. The map shows clusters there anyway. It also shows clusters where enforcement is fragmented and rules exist mostly on paper. Strict or lax, the outcome is the same. Regulation isn’t absent. It’s aimed at the wrong thing.

**Time-based rules fail in a density-driven world. **London caps nights per property. What actually transforms a street is how many units operate as short-term lets at once. You can have perfect individual compliance - every host stopping at 89 nights - and still produce a hotel corridor. The law sees atoms; residents live the aggregate collective transformation emerging from behaviour that is entirely “legal” in isolation.

Airbnb is filling an economic gap; it doesn’t create one. Overlay income-rent pressure and the clusters line up with places where short-term letting works as a patch: for households servicing mortgages that wages no longer cover, and for councils who’d rather tolerate visitor spend than fund enforcement. A listing in a borough where median wages cover median rent is usually a lifestyle choice. A listing where they don’t is more likely a survival strategy - someone turning their only income-scalable asset into cash. That doesn’t make the platform benign, but it does make it endogenous: ban it and the pressure finds another outlet, possibly a worse one. The pressure didn’t start with Airbnb. Airbnb is the shape it took. Platforms don’t have morals. They have optimisation functions.

The real problem isn’t rule-breaking. It’s that the rules describe a London that no longer exists.

Run the story backwards and Airbnb isn’t the first mover here. The gaps already existed; the platform just figured out how to monetise them efficiently.

Very roughly, the mechanism looks like this: Stagnant wages made housing the only income-scalable asset most households have. Austerity-era budgets made councils quietly dependent on tourism spend. Broken planning and low housebuilding baked in scarcity.

Airbnb then behaves like any well-designed platform: it routes demand through the path of least resistance. If you want to respond with something more serious than “ban Airbnb,” policy has to line up with how the system actually behaves. One thing I like to always emphasize: bans are emotionally satisfying. They are not a theory of change. If you want to respond seriously:

**Regulate the right dimension. **If the disruption is about density and clustering, regulating nights per property is necessary but not sufficient. Instead, introduce street- or block-level caps: beyond a certain share of dwellings in a micro-area, no new short-term lets. And tie permissions to local saturation: a borough with 0.2% Airbnb share is not the same as one at 3%.

**Stop flying blind on data. **Right now, councils are outgunned. Platforms have real-time operational data; and regulators only have freedom of information requests (or fragments via audits, complaints). (Guess who moves faster…) Instead, require platform-level reporting of short-term let activity to a central register (this is coming at national level, but only slowly…). Also, make aggregated, anonymised data public so independent researchers (hi) can stress-test the effects. If you can’t see the problem in something with minimal lag, you can’t manage it.

**Align incentives instead of pretending they don’t exist. **If residents are using Airbnb as a wage patch and councils are quietly enjoying visitor spend, bans alone are not serious policy. Some options: (i) hypothecate a portion of short-term let tax/fee income to local housing funds in the same borough (hint: perhaps lower the council tax in Islington?). (ii) Offer fast-track planning or tax relief for landlords who convert short-term units back into long-term rentals in high-pressure zones. In short, don’t just fight the platform at the front door while cashing the receipts at the back.

**Fix the underlying housing and wage story (yes, the hard part). **None of this replaces fundamentals. Sorry. Build more genuinely affordable housing (this is the abundance agenda 101: scarcity in housing isn’t natural) and reduce reliance on housing assets as the primary middle-class safety net.

None of this is radical. What is radical is continuing to regulate a spatial problem with a temporal ruler. I’m not naïve enough to think a Substack post changes planning policy. But I am stubborn enough to think that making the gap visible is the first step and cheaper than another consultation.

In short, Airbnb isn’t blanketing London. It’s pooling in the gaps - places where the housing system was already failing before a single listing went live. The platform didn’t create the problem. It just gave the problem an interface.

But is this a London story or a universal one?

Florence - my previous home - just banned all new short-term lets in the UNESCO historic centre, prohibited keyboxes entirely, and now requires in-person guest check-in. The national registration system wiped out 20-30% of listings overnight; many turned out to be offices and warehouses masquerading as apartments. Aggressive intervention, real results. But Florence has ~10,500 listings in a city of 360,000 - the problem is concentration in a tiny walkable core.

Amsterdam - the closest big city South of where I grew up - went even harder: 30 nights maximum (not 90), mandatory permits, pre-notification of every booking, and a ban on short-term lets in parts of the old town since 2019. The centre and De Pijp may drop to 15 nights from April 2026. Yet despite all this, long-term rents have still risen 37% since restrictions began. The pressure underneath - housing scarcity, capital flows, tourism-first urbanism - doesn’t care which rules you write.

Three cities. Three regulatory regimes. One underlying dynamic. If yelling at the interface worked, Amsterdam would have solved this by now. It hasn’t. The system underneath - the one that made platform income perfectly rational - is what needs redesigning. My map won’t do that. But it makes the gap visible. And once you see the gap, the question changes: not “How do we stop Airbnb?” but “What is the city no longer doing for itself - and what did platforms step in to replace?”

*If you like data-driven stories that go beyond yelling at apps, do share this, subscribe and consider Buying me a coffee to keep the project alive:) If you are interested in the all the code and data of this Substack post, I’ve made it available in my Github here. *

Share

No posts