What Is Actually the Hardest Problem in CRE Brokerage?

For three years it was the correct answer.

Marcus & Millichap is the cleanest public window into the private-client brokerage economy, and its 2023 was carnage. Revenue fell 50.4 percent to $645.9 million from $1.30 billion, and the firm swung from $104.2 million of net income to a $34.0 million net loss. Brokerage transactions dropped from 9,111 to 5,475. Every broker who lived through that has a rate story, and the story is true.

Brokers kept the answer after the conditions changed.

No, and that is the whole argument.

The recovery is unambiguous. CBRE reported Q2 2026 revenue up 16 percent, with global and US leasing each up 24 percent and US office leasing up 29 percent. JLL reported Q2 2026 Leasing Advisory revenue of $836.9 million, up 23.7 percent. Colliers reported leasing and capital markets each up 23 percent. Three companies, same quarter, same direction.

Now run the test on the producer side. Marcus & Millichap's 2025 revenue was $755.2 million across 1,808 investment sales and financing professionals. In 2022 it was $1.30 billion across 1,904. Revenue per professional fell from roughly $683,000 to roughly $418,000, down about 39 percent, in a year the firm called an improving trading environment. That division is mine, but every input sits in its filings.

The market healed and per-producer economics did not. If rates were the binding constraint, the recovery would have closed that gap.

The hardest problem is not the market. It is finding out too late.

Decision-window blindness is the gap between when a tenant starts deciding and when you learn a decision was underway. Here is the arithmetic. CBRE's leasing research puts average office lease terms at 107 months in prime buildings and 86 months in non-prime. Colliers advises occupiers to begin market exploration no later than 12 months before expiration, and 18 to 36 months for anything complex or built to suit. Put those together and roughly one tenant in seven to one in eight sits inside a live decision window in a given year. The rest of your list spends that year doing something else.

Prospecting is a timing problem wearing a persistence costume. Calling an out-of-window tenant with a perfect pitch produces the same revenue as not calling, and it costs a real hour.

Then there is where those decisions land. CBRE found renewals made up 42 percent of US office lease transactions in H1 2024, against a 31 percent pre-pandemic baseline. A renewal is a deal you never got to compete for. The tenant weighed the friction of moving, called the landlord, and closed the window before it showed up as market activity.

The tenants cannot see much either. JLL's Global Occupancy Planning Benchmark Report 2026, covering 84 organizations and 716 million square feet, found only 7 percent rate their own data capabilities as excellent. CBRE's 2026 Americas Office Occupier Sentiment Survey found 66 percent of companies plan to hold or grow their footprint over the next three years, with tech-sector expansion intent jumping to 64 percent from 41 percent. You cannot buy a database of a decision nobody has made.

Not on current evidence, which is uncomfortable for anyone selling software.

JLL's 2025 Global Real Estate Technology Survey of more than 1,500 senior decision-makers found 88 percent of investors and 92 percent of occupiers piloting AI, while only 5 percent of occupiers hit all their program goals. First American Data & Analytics and DealGround surveyed 255 CRE professionals in spring 2026 and found 66 percent using AI weekly or daily, 5 percent trusting it to inform a deal decision, and 34 percent saying they do not know which tools to use.

The tooling that exists points at the wrong layer. CoStar Tenant, factually, offers 7 million-plus commercial tenant records with locations, contacts, and lease expiration data. That is a real and useful product. My opinion is that expiration dates are a lagging proxy for intent, because the expiration is public and the deliberation is not, so every competitor calls the same list on the same week.

Data you can buy is data your competitor already bought.

Because most brokers have no mechanism that survives a quiet week.

The 2025 DNA of CRE Broker Report from Buildout and theBrokerList, a vendor-run survey of 183 brokers fielded in Q4 2024, found 78 percent of opportunities come from referrals from past clients, with relationship building ranked the most effective business-development method at 74 percent. Referral flow is weather. You can dress for it, you cannot schedule it, and it thins out exactly when volume does.

The same vendor's 2026 survey reports brokers spending 46 percent of their time on administrative and manual tasks. Treat that as directional, then hold it next to the last detailed prospecting breakdown I could verify: in the 2020 DNA of CRE survey of 730 brokers, 48.8 percent spent 1 to 5 hours per week prospecting and another 30.9 percent spent 6 to 10. Nearly eight in ten were under two hours a day looking for the next deal.

That is the operating model. Referral-fed, admin-heavy, and unable to notice a decision window that opens on a Tuesday inside a company nobody called.

Four things, none of which carry a price.

Sort by window, not by logo. Rank the target list on months-to-expiration. Past 18 months is a relationship call. Inside 15 months is a live pursuit. County records, permit filings, and the tenant's hiring page cost nothing.

Call the renewal before it becomes one. If 42 percent of transactions are renewals, the highest-yield conversation happens before the tenant decides moving is too hard. That call lands 15 to 18 months out.

Write down where every deal came from. Origination source for your last twenty pursuits, one page. If more than 70 percent traces to referral, you have an inbox rather than a pipeline.

Protect two hours. Not four, not a whole morning. Two hours a day, same time, aimed only at in-window names. That alone puts you in the top fifth of the 2020 prospecting distribution.

Three objections, and the first is good enough to hurt.

Systems do not create demand. A disciplined broker in 2023 still earned less than a sloppy broker in 2021, because commission income is levered to transaction volume no process can conjure. This argument explains relative performance inside a market, not absolute income across cycles.

Second, my revenue-per-producer math rests on one firm. Marcus & Millichap does private-client investment sales, not Seattle tenant rep, and its headcount mix shifted through the downturn. Treat it as an indicator rather than a proof. I would drop the claim tomorrow if the same ratio at a tenant-rep-weighted firm ran the other way.

Third, relationship-led brokerage works. Seventy-eight percent referral origination is not automatically a failure state, and for a broker twenty years into a Seattle book it is the right model. The fee pressure people keep predicting has also not arrived. A May 2025 Federal Reserve note by Banerjee and Paciorek found residential buyer-agent commissions drifted only from about 3 percent in the late 1990s to about 2.7 percent, with buyer representation agreements having negligible effect on advertised rates. NAR has said the Sitzer/Burnett settlement addresses residential transactions, and CCIM's coverage frames the commercial spillover as more formal representation agreements and clearer fee disclosure rather than compression. Commission structure is what everyone worries about instead of this.

I would treat the decision window as the product and everything else as overhead.

Seattle makes the case for me. Kidder Mathews put Seattle Close-In office vacancy at 28.2 percent in Q2 2026, the CBD at 34.9 percent, and regional net absorption still negative at -46,836 square feet. The temptation in a market like that is to work harder on the same list. That is exactly wrong. When 28 percent of the space sits empty, landlords will talk to anyone, and space becomes the abundant input. The scarce input is a tenant who has started deciding.

So I would rebuild the week around three questions and refuse to answer anything else. Who is inside 15 months. What changed at that company in the last 90 days. Who else already knows. If a name does not clear all three, it goes to the relationship list for a quarterly touch, not an hour.

I would also get honest about renewal risk on my own accounts before getting clever about someone else's. Every tenant you represent is a renewal candidate for the landlord's broker, who knows that building better than you. The defense is calendar discipline, not charm.

Most brokers are losing to a version of themselves that found out six weeks late. Rates recover. Six weeks late does not.

One disclosure, because the argument is self-serving. BrokerHQ is building the decision-window layer I just claimed is missing, so I am not neutral here. The falsifiable test does not involve us: run window-sorted prospecting for two quarters, and if the referral share of your originations does not fall, my thesis is wrong and no software fixes it.