Nearly every recruiting conversation I have watched ends in the same place: the split. It is the easiest currency to offer and the only one the brokerage next door can match in ten seconds. A split war is won by whoever is willing to lose the most margin, which is not the same as whoever builds the best brokerage.
Before the value proposition, it is worth looking at the data on an agent’s actual career, because it explains a good deal about why people move.
What the numbers say
According to the member profile NAR published on June 25, 2026, the median experience of a Realtor rose to 13 years, and median tenure with their current firm is 6 years. That second figure deserves a second read: changing houses mid-career is not a betrayal, it is the statistical norm of this industry.
Median gross income from real estate activity was $59,200, up from $58,100 the year before, with median business expenses of $9,530. And it is heavily skewed by experience: those with 16 years or more reported a median of $88,500. The typical agent closed 9 transaction sides in the year; those working on a team, 32 — though that figure is for the whole team, and only 21% of members work on one.
One honest caveat: NAR does not publish why people change firms, so anyone handing you a ranked list of reasons with percentages is either inventing it or quoting their own survey. What the data does sketch is the terrain — a long career, income concentrated in experience, and a change of house every six years on average.
The value proposition that is not the split
With a median income of $59,200 and nearly ten thousand dollars of expenses, five more points of split is a few thousand dollars a year. That sounds good until you compare it with what actually moves the needle: closing two more transactions. That is where a brokerage can compete without bleeding, and it comes down to four things.
- Leads with clear rules. Not "we have leads," but how many, from which source, how they are distributed and what happens when they are not worked. An experienced agent spots vagueness instantly.
- Systems that save administrative time. Every hour not lost to paperwork is an hour of prospecting. This is measurable, and it should be measured before it is promised.
- Real, specific training — not one motivational day a year. What is taught, when, and by whom.
- Personal brand. That the brokerage grows their name, not only its own. An agent who feels they are building something of their own inside your house does not leave over five points.
The interview, in reverse
Most recruiting interviews are a presentation of the brokerage with pauses for questions. You learn far more by asking, and these are the questions that yield the most:
- "Where did your last five closings come from?" It separates the agent with their own engine from the one who depends on leads being handed over. Both answers are valid; they need different brokerages.
- "What part of your week would you like to never do again?" That is your concrete value proposition, stated by the person who will judge it.
- "How do you keep your data today?" If the answer is "on my phone," that is where your onboarding starts.
- "What made you start thinking about a change?" Listen without interrupting. Whatever they say in the first thirty seconds is what you actually have to solve.
- "What would you need to see in ninety days to know you got it right?" It is the only question that sets a shared success criterion before you start.
The thirty days that decide whether they stay
Recruiting does not end at the signature; it ends when the person closes their first deal in your house. The month in between is where people are lost, and almost always through neglect rather than disagreement.
- Week 1: access to everything on day one, their data migrated, and one named person they can ask without feeling stupid. An agent who spends their first week waiting for credentials is already calculating how long until they leave again.
- Week 2: the full flow with a real deal, from lead to file. Not a tutorial: their own client.
- Week 3: first numbers review together. What is coming in, what is stuck, what is missing.
- Week 4: an introduction to the rest of the team with something of theirs — a case, a lesson, an area. Belonging is half of retention.
The three promises not to make
- A lead count you cannot sustain. Promising forty a month and delivering twelve does not cost you an agent: it costs the brokerage its reputation in a market where everybody knows everybody.
- Income. Nobody can promise what another person will earn, and putting it in writing is both unwise and potentially consequential. You show the house’s real numbers — how many agents, what the median produces, how long the last hire took to close — and let people do their own arithmetic.
- That everything will be different. "You will not have the problems you had over there" disproves itself in week one. It is far more credible — and works better — to name your house’s problems and explain how they are handled.
Measuring retention instead of sensing it
Three numbers, reviewed quarterly: how many active agents you have, how many closed at least one transaction in the last ninety days, and the average tenure of each person in the house. The second is the one that matters: a brokerage with thirty agents of whom eight produce does not have a recruiting problem, it has an activation problem — and hiring more people makes it worse.
The seven numbers worth watching every Monday — including activity per agent, which is the early signal of an exit — are in another guide. And if your growth model includes tiered recruiting with overrides, Legacy models that inside the platform and the commission software settles them at closing; but the part that retains people is not solved by any software: it is the Thursday conversation when someone has gone three weeks without closing.
All of it comes down to a line I have heard from more than one broker with a stable team: you recruit with what you promise and you retain with what you deliver. The split is the easiest promise to make and the easiest to match. The rest is not.