What buyer-side number goes in a model for a 2027 purchase

Question

I am rebuilding acquisition cost assumptions out to 2027 and the buyer-side fee is the one input I cannot defend with data. Redfin has averages near 2.4 percent in early 2025, essentially flat against pre-settlement. A Federal Reserve note said commissions may have slipped some while staying high, and credited that partly to sellers' agents sharing compensation outside the MLS. One industry survey has combined commissions ticking up through 2025. Against all of that sits the forecast of fees falling as much as 30 percent over time with 1.6 million agents affected, which has not shown up anywhere in the near-term numbers.

So the honest range for 2027 is something like 1.7 to 2.5 percent, and that spread is 24 basis points of purchase price on a 300k deal, roughly 2,400. On a five-deal pipeline it is real money in the reserve line.

The argument for holding 2.4 is that the mechanism holding it up is structural. Buyers are out of cash after down payment and closing costs, so sellers keep covering the fee to stay competitive, and no rule change fixes buyer liquidity. The argument for modeling lower is that transparency compounds. Every buyer who signs a number now knows what the number is, and agent headcount pressure points to consolidation.

I have my own answer and I want to see how far off consensus it is.

megan_doyle15 · Accepted Answer

Dumb question, is 2.4 percent the buyer side alone or both sides added together? I keep mixing those two up.

paulgrant17 · Answer

Rigging, the 2.4 figure is the buyer side by itself. Combined with the listing side you land somewhere near 5, which is why the survey Sextant mentions about combined commissions is a different number from the Redfin one. Sextant, I would model 2.4 and hold a separate line for the gap you might have to fund yourself, because the risk on a single deal is not the average drifting, it is one seller who offers 1.5 when you signed 2.5.

hannah_west29 · Answer

Your range is doing something odd. You are treating 1.7 and 2.5 as symmetric possibilities and they are not. The distribution of what sellers actually offer is clumpy, it sits at 2.5, 2.4 and 2.0 with very little between, because those are the numbers brokerages have printed on forms for thirty years. Averages drift smoothly, individual deals do not. So for a five-deal pipeline I would model at 2.5 and treat anything under it as upside rather than run the mean and get surprised on the specific house. The other thing your model probably hides is that the fee is a percentage of a number you negotiate. Getting 8k off the price does more for you than 20 basis points off the fee, and the fee argument eats attention out of proportion to its size.

karen_whitfield · Answer

On land the whole framing shifts. Fewer comps, fewer agents who know the parcel, and I have paid flat fees above 2.4 equivalent happily because the work was survey and access questions rather than showings.

gregpowell66 · Answer

Ledger's point about the clumpy distribution matches what I see from the construction side, everybody quotes the same number until one guy breaks ranks and then a whole market resets in a season. That is how sub trades went in my county. If it happens to buyer-side fees it will not be a slow slide to 1.7, it will be one large brokerage in a metro publishing a flat fee and the rest matching within a year. Sextant, that argues for modeling the current number and keeping the flexibility to reprice, since a step change you can respond to hurts less than an average you guessed wrong.

What buyer-side number goes in a model for a 2027 purchase

5 replies