In part one of this two-part editorial series, I argued Real’s $880 million REMAX deal is a bet on the agent at the moment the consumer is moving away — and that Tamir Poleg almost had to make the deal anyway, because the alternative was getting stuck as a mature eXp in a winner-take-most market.
What happens to the commission when the consumer finally knows what they’re paying for?
Real with the REMAX deal, and Compass with the Anywhere deal, didn’t buy a brokerage so much as the apparatus around it — the franchise fees, the splits, the agent caps, the mortgage and title attach. All of it sits on top of the commission.
TAKE THE INMAN INTEL INDEX SURVEY
But AI is doing to the commission what it’s doing to every other bundled service: pulling it apart into pieces consumers can price.
The work to sell a $400,000 house and a $1.5 million house is roughly the same. The commission is wildly different — 3 percent of $400,000 is $12,000; 3 percent of $1.5 million is $45,000 for the same listing entry, the same MLS coordination, the same showings.
That mismatch has been true for 40 years and has held up because no consumer ever had a way to question it at the moment they were signing the listing agreement.
That changed in November 2022 with the launch of ChatGPT.
A seller now sits down with an AI before they sit down with an agent. They ask what a 6 percent commission actually buys. They get a list. Listing entry, comp pull, marketing copy, MLS coordination, scheduling, paperwork — work AI is already doing, free. They ask which parts of the bundle still require a human.
They get a shorter list. Negotiation, on-site judgment, getting the deal closed, legal liability. They ask what those parts are worth on their own.
They get a number, and the number is not 6 percent (or even 5.44 percent).
Then they ask the agent to itemize. They negotiate the listing-side commission. They ask why they are also paying the buyer’s agent.
The seller in that conversation is not behaving the way sellers behaved when REMAX wrote its franchise agreements, when Compass set its splits or when Real built its commission cap.
None of those models were designed for a seller who shows up knowing the math
Realtor.com’s October 2025 survey found 82 percent of buyers and sellers using AI for housing decisions — not housing search, but actual decisions about what to pay, what to offer, what to negotiate. ChatGPT (67 percent) and Gemini (54 percent) lead. The most important finding in that survey is buried under the headline: Respondents already rate AI a 61 on the “makes me smarter about the market” question, against agents at 62.
Three years after ChatGPT launched, the consumer’s confidence in AI as a market intelligence source has effectively caught the agent.
The 5.44 percent national average commission does not disappear tomorrow. But within five to seven years — possibly sooner — it stops being the default expectation.
Flat-fee and à la carte structures take a meaningful share of transactions, mostly because consumers stop paying for services they no longer need. Discount brokerages get a second look. And because the full-service agent stops being the only path, every brokerage’s revenue per transaction goes down.
There’s no version of this where it doesn’t.
The companies positioned for that come out fine. The companies whose entire stack is built to attach to the existing commission do not.
Real-REMAX is in the second category. So is Compass-Anywhere. Both deals make complete sense if the existing commission structure holds. Both deals have a much harder math problem if it doesn’t.
Rocket is the only deal of the three that does not need the commission to hold, because Rocket bought a customer relationship that survives no matter what happens to the legacy agent-brokerage model.
A homeowner with a Mr. Cooper-serviced loan is going to keep being a homeowner with a Mr. Cooper-serviced loan, whether the listing commission is 6 percent, 4 percent or a $3,000 flat fee. A buyer using Rocket Mortgage to finance is going to use Rocket Mortgage whether their agent took the listing for 3 percent or 1.5 percent. The relationship is upstream of the commission. That is why Rocket’s bet works in any of the futures, and why Real’s only works in one of them.
This is the harder argument the industry doesn’t want to have. Consolidation in residential real estate right now is being driven by two forces at once.
The first is that scale is starting to matter again — agent count, brand recognition, technology investment, all the things bigger gets you. The second is that the per-transaction economics are about to compress, and the only way to survive lower margins is more agents. Both forces push every operator toward the same playbook. Buy more agents. Cut more cost. Ride out the recalibration.
The problem is the playbook works only if you accept that the brokerage as we know it is the unit of the future.
That is the assumption Compass-Anywhere and Real-REMAX both make
Rocket made a different one, betting roughly $16 billion that the customer relationship is the unit of the future and the brokerage is one piece of a longer arc.
Tamir Poleg is one of the better operators in this industry, and he had to do this deal. Wall Street’s growth math gave him no other choice that didn’t end with Real hitting the same wall eXp hit. Done right, the synergies will help his margins for years.
The harder question is whether his $880 million is pointed at the customer relationship that survives the next decade.
Rocket bought the homeowner — the person who lives in the house, pays the mortgage, refinances, takes a HELOC, eventually sells, eventually buys again. Real bought the agent — the person who shows up for the transaction, gets paid out of the commission and goes home.
Only one of those bets gets stronger as the consumer’s relationship with the transaction keeps changing. And the consumer’s relationship is changing fast.
Amit Kulkarni is co-founder of Alloy Advisors, and currently serves as Interim CEO at Homes for Heroes. Connect with him on LinkedIn.
