If every deal still lands on your desk, your buying criteria aren't written down. They're in your head. Property buying criteria are the rules that decide which deals you chase: where, what, the numbers a deal must hit, the refurb you'll take on and the exit that has to work.
Write each rule so two people would apply it the same way. Label it Stop, Flag or Prefer. Log one line on why you passed on every deal you turn down. Do that and the same document briefs your sourcers, trains your next hire and lets a system do the first pass. You keep the decision, and the same team gets more offers out.
What criteria in one head cost you
The usual sourcing brief is one line: BRRR, North West, under £150k, 20% ROI. Every phrase needs a phone call. Is the ROI on cash in, or after refinance? Which towns, and which streets in them? Is £150k the asking price or the all-in cost?
You know the answers. Your sourcer, your analyst and any software you point at that line do not. So they guess, deals arrive that miss, and you review every one yourself. That isn't carelessness. Give two people the same one-line brief and they will judge the same deal differently, and so will you on a Monday versus a Friday. Researchers call that variability "noise", and it shows up wherever judgement isn't written down.
83 deals reach you in a month. You spend 22 minutes on each before saying no to most. That is 30 hours of founder time a month, close to four working days. If written Stops rejected half of them before you opened them, you would get 15 of those hours back every month.
Stop, Flag, Prefer: rules two people apply the same way
The test for every rule: if two people read it and look at the same deal, do they reach the same answer? If not, rewrite it.
| Vague | Testable |
|---|---|
| Good yield | Gross yield of 9% or more at asking price, using achieved rents for the same bedroom count within half a mile in the last 12 months. |
| Not too much work | Refurb of £30,000 or less per unit and 10 weeks or less on site, priced from our standard schedule of rates. |
| Decent area | Inside the postcode districts on our target list, excluding the streets on our exclusion list. |
| Must refinance well | £10,000 or less left in after refinance at 75% loan-to-value, on a valuation at the lower quartile of sold comparables. |
Then give every rule one of three labels:
- Stop The deal fails. Nobody needs to look at it.
- Flag The deal goes to a person, with the reason attached.
- Prefer Breaks a tie between two deals that both pass.
Next to each rule, write one line on why it exists. "No flats above the fourth floor" is a rule. "Because two lenders on our panel won't lend on them" is the reason. The reason lets your team handle the edge case, and tells you when to change the rule. The rules plus the reasons are your firm's written judgement: the one asset a competitor can't buy, however good their software.
What your criteria should cover
Six blocks. Each one answers a question you already ask of every deal.
| Block | Example rule | Label |
|---|---|---|
| Strategy and exit | BRRR only; refinance at 75% LTV must leave £10,000 or less in | Stop |
| Location | Target postcode districts only. For HMOs, note any Article 4 direction removing the right to convert a house (C3) to a small HMO (C4) | Stop |
| The property | Freehold three-bed houses; no non-standard construction, which some lenders avoid | Stop |
| The numbers | All-in cost (price, SDLT, fees, refurb) of £135,000 or less; stress-tested at today's rate plus 2 points | Stop |
| Refurb | £30,000 or less and 10 weeks or less; any structural work | Flag |
| Deal and counterparty | Sourcer registered with HMRC for anti-money laundering supervision; vendor can complete within eight weeks | Stop / Prefer |
The sourcer check deserves its Stop. HMRC lists property finders among the businesses that must register, and trading without registration is a criminal offence. In one 2024 analysis, 35 sourcing businesses were fined £150,299 in total for anti-money laundering breaches.
Copy this: a one-page buying criteria sheet
One strategy per sheet. Stops first, because that is the order a sourcer or a system will check them in.
Owner: [name] · Last reviewed: [date] · Next review: [date + 3 months]
| Rule | Label | Why |
|---|---|---|
| [Property type and tenure] | Stop | [Lender or tenant reason] |
| [Target postcode districts, minus exclusion list] | Stop | [Where the list comes from] |
| [Maximum all-in cost, with what's included] | Stop | [The money-left-in ceiling it protects] |
| [Minimum return, formula written out] | Stop | [Your cost of capital] |
| [Money left in after refinance, at stated LTV] | Flag | [Capital needed for the next deal] |
| [Refurb budget and weeks on site] | Flag | [Contractor capacity] |
| [Sourcer registration checked] | Stop | Legal requirement; protects your money |
| [Tie-breakers: transport, completion speed] | Prefer | [What makes one passing deal better] |
Log why you passed
Criteria are half the asset. They say what you buy in general. The other half is what you decided on each deal, and why.
Your deal tracker says Passed, or Offer £118,000 accepted. It doesn't say you went £6,000 over your ceiling because the vendor needed a four-week completion and the contractor had already priced the refurb. Venture firm Foundation Capital calls these records decision traces: systems store the outcome, but the reasoning lives in people's heads and message threads, and leaves with them.
The fix is small. Add one column to your tracker: why. Fill it in for every pass, every purchase and every exception, in one sentence. Once a month, read the column. A reason that appears three times is a rule you haven't written down yet, so add it to the sheet.
Recorded reasons are what let a system take work off the founder. One ten-person software company that recorded its decisions for two years turned them into rules and built AI "judges" to apply them. In eight days, the founder reports, the judges decided 7,296 questions, sent 2,213 back and held 1,863 for a human. The founder stepped in 27 times. Their scale isn't yours. The principle is: a decision can only be delegated once its reasons are written down.
Test it on the deals you've already turned down
- Pull your last 30 passes and your last 10 purchases.
- Run each one through the sheet without looking at what you decided.
- Where the sheet says yes and you said no, there's a rule you haven't written. Write it, with its reason.
- Where the sheet says no and you bought anyway, either the rule is wrong or the deal was an exception. Decide which, and log it.
- Give the same 10 deals to two people separately. Where they disagree, the rule is still vague.
Illustrative example. A founder-led BRRR firm buying 38 units a year runs its last 30 passes through a new sheet. Four pass every Stop, yet the founder said no to each. The reasons: two were on main roads and two were more than a 15-minute walk from a station. Both become Flags, with their reasons, the same afternoon.
Put it to work
Your sourcer brief. Send the sheet itself, Stops first. Then measure it: each month, track the share of each sourcer's deals that pass your Stops. A sourcer whose pass rate doesn't rise after they've had the sheet isn't reading it.
Your next acquisitions hire. They learn your judgement from the rules, the reasons and the log, instead of six months of shadowing you.
A first pass on every deal. Once every rule is testable, a system can check each inbound deal against the Stops, attach the Flags and send you only what's worth your time. It doesn't need a complex model. The hard part was always writing the rules down, which is why our method maps that step before anything gets built.
Common mistakes
- Numbers with no reason. A minimum yield nobody can explain gets ignored the first time a deal looks exciting.
- Everything is a Stop. If every rule is a must-have, the team treats none of them as one.
- One person's sheet. If two people decide on deals, both write their rules down and reconcile the differences. Expect some.
- Written once. Review it every quarter, and whenever the why column shows a new pattern.
- Living in a chat thread. One current version, in one place. Sourcers get that version only.
Frequently asked questions
What is a buy box in UK property investment?
A buy box, or buying box, is the set of written criteria a property investor uses to decide which deals to pursue: location, property type, price, minimum returns, refurb limits and exit. For a firm, it works best written as testable rules, each labelled Stop, Flag or Prefer, with a one-line reason next to it.
What should a property sourcing brief include?
Your Stops first: strategy, target postcode districts, property type and tenure, maximum all-in cost and minimum return with its formula written out. Then your Flags, then your Prefers. Before sharing deals or money, ask the sourcer for their HMRC anti-money laundering registration and redress scheme membership.
What is a deal decision log?
A deal decision log records one line on why each deal was passed on, bought, or bought as an exception to your rules. Your criteria say what you buy in general. The log records what you decided on each deal and why. Reviewing it monthly shows the rules you use but have not written down.
Can AI screen property deals against buying criteria?
Yes, if the criteria are written as testable rules. A system can check each inbound deal against your Stops and pass anything flagged to a person with the reason attached. If the rules are vague, AI applies them inconsistently, just as people do.
Founder of AGI Automations. Certified Data Management Professional (CDMP) and Anthropic system integrator. Ben maps acquisition workflows for UK property investment firms and builds the systems that let the same team underwrite more deals.
- Kahneman, Rosenfield, Gandhi and Blaser, "Noise: How to Overcome the High, Hidden Cost of Inconsistent Decision Making", Harvard Business Review, October 2016
- Jaya Gupta and Ashu Garg, "AI's trillion-dollar opportunity: Context graphs", Foundation Capital, December 2025
- Spawn founder, post on running the company with AI judges, X, 2026 (figures self-reported)
- HMRC, Money laundering supervision for estate agency businesses
- NAPSA via PR Fire, analysis of HMRC AML fines on sourcing businesses, September 2024
- Planning Data, Article 4 direction records (example: Tower Hamlets HMO direction)
- Gladfish, How to assess a property investment opportunity in 10 minutes