I’m careful about the media I follow, not because I’m trying to be clever, but because social platforms don’t show you the world. They show you a feed.
What you see is shaped by who you follow, what you’ve engaged with before, and what the platform thinks fits your profile. So when a clickbait video dropped into my feed titled “Why AI is a load of rubbish for PropTech Suppliers”, I did what everyone does. I clicked it.
It was posted by a property journalist I follow and generally respect, which is why it hit a nerve. The guest was a referencing supplier. To be fair, they did concede that AI is already supporting parts of their process. But the framing was the thing. That headline captures a mood I see constantly in UK property: a mix of fear, frustration and misunderstanding about what AI is, where it works, and what it’s for.
I spend my time applying technology inside estate and letting agencies so the people who generate income spend more time doing exactly that. Less admin. Faster follow-up. More conversations. More instructions. That’s the job. Not shiny tools. Not buzzwords. Just getting good people back to high-value work.
So let me give you one example that every agency owner understands immediately.
There isn’t an agency in the country that doesn’t monitor competitor instructions. Everyone does it. It’s essential. We operate in a marketplace where sellers and landlords invite more than one agent out. And in plenty of cases, the home ends up being sold or let by the second instructed agent. That means if you’re not watching, you’re not competing.
But the moment you see a competitor listing, the same problem hits you in the face.
You can see it. You can’t reliably identify the full address.
You’ll have your own systems. Some agencies subscribe to software. Some rely on portal email alerts. Some still use old-school board spotting, though that’s dropping for obvious reasons. However you detect the listing, the next step is the same. You need the full address so you can do something with it.
And that’s where the industry loses time.
The portals don’t make it easy. Rightmove, Zoopla and OnTheMarket typically show the full postcode and a partial address line. You’ll see the road, maybe a locality, but not the full property identity. That’s deliberate. Their terms restrict scraping, and the data is presented in a way that makes it awkward to extract cleanly. Competitors don’t want you targeting their instructions. The platforms don’t want you doing it either. So you end up stuck between what you can see and what you need to know.
Which is why agencies turn address identification into a weird mix of detective work and local knowledge.
You already know the list. You’ve probably done all of these.
You’ve visited the property before because you were invited out. You’ve spotted it on the way to another appointment. You’ve tried to match the front photo to Street View. You’ve pulled the EPC and tried to match the rating. You’ve checked old adverts. You’ve looked for recent planning applications and matched features. You’ve used browser extensions. You’ve asked someone who “might know”. And yes, some people still try to get the address by questionable means, which I’m not interested in.
I’m also not pretending I’m above it. I’ve found thousands of competitor addresses over the years. I can do it quickly in my local area because I’ve built the muscle memory for it.
Which brings me back to the image above.
Look at it properly. A good property person can often infer the full address from tiny signals. A neighbour’s number plate. The bins. The way a street is numbered, odds one side and evens the other. The housing style and the position on the road. It’s not magic. It’s repetition.
But the real question is the one agency owners should care about.
How do you turn that into a workflow your business can run without relying on one person’s experience?
Because even if you’ve got one “address wizard” in the office, that doesn’t scale. It also doesn’t belong on a lister’s desk when they should be speaking to vendors and landlords.
Now, if you try to solve this like a human, you hit three problems immediately.
First, the portals and many websites hide the full address and only give limited information. Second, their terms prohibit scraping and they put basic measures in place to stop you extracting the full address in a simple way. Third, computers struggle to interpret images the way humans do. A person can spot a number on a neighbour’s door or infer the likely property number from tiny visual signals. A typical system can’t. Not reliably. Not at scale.
So you need a different approach.
A workflow is simply a repeatable process that starts with a trigger and ends with a reliable outcome. In this case, the trigger is a competitor listing appearing. The outcome is the full address, with confidence.
That’s what The Shadow does.
And to be clear, we do not hack portals. We do not screen scrape in breach of terms. We take digital responsibility seriously.
The Shadow monitors competitor activity using legitimate signals. That might be an alert your agency already receives. It might be a change detected on a competitor’s public website. Once it has that signal, it records what it can: postcode (where present), partial address line, price, date first marketed, agent name, property type, bedrooms, plus a unique identifier so it can recognise that listing later and avoid duplicates.
From there, it does what a human does, but without wasting your team’s time.
It doesn’t rely on one “single point of truth”. It builds the possible address permutations and then starts eliminating them using structured data.
If the data suggests there are 26 possible properties within that postcode segment, fine. The Shadow will work from 26. It will pull EPC data and prioritise candidates that actually have an EPC. If the listing shows an EPC rating, those candidates rise. If property type and bedroom count align, they rise again. It turns “a messy list of possibilities” into “a ranked shortlist with confidence”.
At that stage, it’s doing what good operators do manually, except it’s consistent, repeatable and it doesn’t get bored after the fiftieth listing of the week.
Where it gets interesting is in the edge cases, because edge cases are where most tools quietly give up.
New builds where a postcode is missing or fresh. Listings where the EPC is absent, wrong, or not displayed. Flats and blocks where partial address data creates too many plausible matches. Listings with deliberate errors. All the awkward stuff.
The Shadow uses AI-assisted checks to spot when the listing data looks wrong, to decide which validation route makes sense next, and to extract additional clues when the structured trail runs thin. It’s not trying to be a person. It’s trying to be useful inside a process.
And when it still can’t validate confidently, it escalates to a human. Because accuracy matters. If you’re going to act on an address, you need to trust it.
The other point I want to make is this.
The Shadow isn’t “an address finder” for the sake of it. An address doesn’t win instructions. Speed and follow-up do.
So the workflow doesn’t stop at identification. Where your CRM allows it, The Shadow can check whether you’ve already got history with that address. It can surface relevant context to the right person, at the right moment. It can pull Land Registry title information where that’s useful. It can trigger approved outreach letters and messages at different points in a property’s journey. The goal is to remove the manual friction that slows your team down and causes opportunities to slip.
That’s why the clickbait video annoyed me.
The industry keeps arguing about whether “AI is good” instead of asking the more important question. What parts of your agency’s week are repeatable, time-draining, and directly linked to revenue if you fix them?
Competitor monitoring is one of the clearest examples because every owner understands the value, and almost every owner has watched their team lose time turning a listing into an address.
If you want to see what The Shadow does, and whether it fits how your agency generates instructions, visit the landing page here.
