“Should we hold or reduce?”

If you’ve worked in agency for more than a week you’ve been asked it. I’ve been answering it since 1998, mostly from experience, instinct and whatever happened on the last few listings.

This time I answered it with evidence. Over the last week AI and I worked through every sales listing created at Preston Baker since 2023, 2,606 of them, along with their viewings, all 7,212 price changes and the buyers’ own price feedback, all held in Greenhouse OS.

The practical bit comes first. The evidence follows.

What to Tell Your Sellers

1. The first two weeks matter most. If there have been no viewings in the first 14 days, that’s when we talk about price, not at two months. Homes with no early viewings were roughly three times as likely to be withdrawn.
2. Waiting rarely works once it’s gone quiet. Of homes that held their price at day 60 with little interest, about one in seven sold in the next two months and nearly half came off the market.
3. A small reduction is a wasted reduction. Cuts under 2% made no difference to viewings because the portals don’t alert buyers below that. The cut that drops you under a Rightmove price point is the one buyers notice.
4. Reducing doesn’t open the door to more haggling. You’ll probably accept less than your original asking price, but reduced homes still agreed at over 99% of their new asking price.
5. The buyers are telling us something. We ask every viewer what they think the home is worth, and if that keeps coming in under 90% of asking it’s an early warning worth acting on.

What to Tell Your Team

1. Day 14 is the first price review. No viewings by then means a price conversation, not another fortnight of hoping.
2. Read the interest, not the calendar. Two or more viewings in the last four weeks and holding is reasonable. Fewer than that and holding rarely works, and the odds fall every month.
3. Reduce while there’s still interest. A home with some interest that cut across a Rightmove point sold within 90 days 76% of the time, against 40% to 50% for holding. Leave it until the home has gone quiet and the same cut only moves the odds from about 20% to 25%, which suggests price may not be the only thing to fix by then.
4. Never recommend a cut under 2%. Start with the cut that crosses the nearest Rightmove point.
5. Get a price opinion from every viewer. It’s the earliest warning we have.
6. Use the evidence, not your opinion. “Of homes like yours that held at this point, about one in seven sold in the next two months” is a far easier conversation than “I think we should reduce”.

Working With AI on Our Own Data

Until recently AI was something I used to draft a social post or tidy up an email. Useful, but on the edge of the business.

That has changed. Because it can now read everything we hold in Greenhouse OS, I can ask a question in plain English and it goes and does the digging. I push back on an answer and it goes back and checks. Where the sample is too small to trust, it says so.

Doing this by hand would have meant exporting from several places, cleaning the data and building a spreadsheet over a few weeks. I suspect, like most agency owners, I’d have started it and never finished.

None of this works unless the data is in one place and you control it. Your data is the asset. AI is just the first thing that’s made it easy to use.

The Evidence

Viewings in the first 14 days, across 1,713 sales listings:

Viewings in first 14 days	Listings	Sold	Withdrawn	Sold within 60 days	Median days to sale agreed
0	411	57%	29%	37%	60
1-2	603	63%	29%	44%	38
3-5	381	74%	19%	59%	22
6+	318	87%	10%	77%	17

Homes that were still unsold and held their price, split by interest in the previous four weeks (quiet means 0 or 1 viewings, some interest means 2 or more):

Unsold at day	Quiet: sold in next 60 days	Quiet: eventually withdrawn	Some interest: sold in next 60 days	Some interest: eventually withdrawn
45	15%	45%	48%	25%
60	15%	45%	45%	24%
90	12%	47%	35%	35%
120	8%	51%	33%	36%

What happened after a price cut, by size of cut:

Size of cut	Cuts	Viewings 4 weeks before	Viewings 4 weeks after	Sold within 90 days	Eventually withdrawn
Under 2%	12	1.1	1.1	33%	42%
2-3%	46	0.8	1.9	39%	22%
3-5%	74	1.9	2.3	46%	22%
5-8%	63	1.3	2.4	29%	32%
8%+	39	0.9	4.4	31%	15%

Bigger cuts bring more viewings but not proportionally more sales, mainly because they tend to be made later, on weaker homes. Most of the benefit comes from clearing 2% and crossing a Rightmove point.

What the seller achieved:

Price history	Sales	Agreed vs original asking	Agreed vs final asking
Never reduced	1,380	100%	100%
Reduced once	92	96.1%	99.4%
Reduced twice or more	42	91.6%	99.3%

And the buyers’ view. Where viewers’ median price opinion was under 90% of asking, 57% of homes sold and 25% were withdrawn. Where it was 95% or more, 75% to 78% sold and only 11% to 14% were withdrawn. Homes that buyers valued at or above asking agreed at a median 101.7%.

What Goes Into the App

We’re building this straight into our agent app:

1. A live listing with no viewings by day 14 gets a “Price review” next move, with the odds from the first table.
2. The price calculator shows what similar homes did when they held versus cut, based on that listing’s days on the market and recent viewings.
3. A cut under 2% is flagged as unlikely to change anything, and the cut that crosses the nearest Rightmove point is suggested first.
4. The buyers’ price view sits on the listing, flagged when it drops under 90% of asking.
5. Agents can choose to add a short hold-or-reduce paragraph to the seller’s weekly report.

A Note on the Data

These are Preston Baker’s own figures. They move as new sales come through and are correct at the time of writing. A few honest limitations:

• The sample of price cuts is small, 165 in total, so read those figures as direction rather than precise odds.
• Only 460 of 1,226 completed listings have a recorded live date, so for the rest the first viewing marks the start of marketing. That makes the “0 viewings in 14 days” group smaller than it really is.
• “Sold” means a sale agreed and not later withdrawn.
• Agreed prices often appear to be recorded at the asking price, which may flatter the never-reduced homes.
• Portal enquiries and Rightmove click-through data aren’t in this analysis yet.

Final Thought

None of this will shock an experienced agent. Most of us knew quiet homes rarely sell by waiting.

The difference is that I can now put real evidence in front of a seller, from homes like theirs in our own patch, rather than asking them to take my word for it. That makes for a more honest conversation, and I think a kinder one, because nobody wants to watch their home sit on the market for six months and then come off.

This is what I mean by High Skill / High Tech. The AI does the analysis I’d never have found time for. The valuer still has to have the conversation, and now has a much better one.

If you’d like to talk about what your own data could tell you, feel free to drop me a WhatsApp. And if you’d like these articles by email each week, subscribe below.