Step 1 of 5 · 2009–18

Agents recruited agents for revenue share

An agent grab paid from future commissions

Glenn Sanford, a former Keller Williams top producer, built eXp without offices because in the downturn he could not afford them AG9 AG6. Recruiting was built into the pay plan. A sponsor, an eXp agent who brings in another productive agent, earns revenue share: a portion of eXp's own cut of that recruit's commissions, paid on several tiers of recruits below the sponsor AG1 AG4. Agents also earned company stock for their first closing and when an agent they sponsored closed a first deal AG9. eXp spent no cash on recruiting; its agents recruited, paid from commissions not yet earned AG9.

The choice made growth fast and cheap. In 2018 eXp went from just over 6,000 agents in January to 11,000 in May, the month it listed on Nasdaq AG9. Sanford saw a race for agent count and did not claim to attract only the best: "We are like everyone else in the industry," he said AG9. A sponsor was paid for any recruit who met a transaction threshold, so the plan rewarded numbers of agents, not the size of one agent's client list AG1. Recruiting paid off only if the added agents produced and stayed, and the payoff went to a few: eXp's income disclosure shows a median revenue-share payment of zero AG3.

Rivals Compass did the reverse: after its rentals business failed it hired sales agents whose clients followed them by name, starting with Leonard Steinberg from Douglas Elliman in 2014, and later ran recruiting through an enterprise sales team CPX-13 CO2.

Step 2 of 5 · 2018–23

Low overhead instead of a war chest

The office choice dates from the founding; it became eXp's answer once Compass began spending venture money on agents in 2018. eXp had no leases, office insurance, utilities, office salaries or maintenance to pay AG9. Training, agent support and meetings ran online, from 2024 in FrameVR, a web-based 3D workspace AG1. The savings went to agents as a lower price. eXp took 20% of each commission until an agent had paid it $16,000 in a year, plus a small transaction fee, after which the agent kept 100% AG9.

The price was funded from commissions, not investors. Sanford accepted falling margins to build the agent base and pointed out that, unlike Compass with its venture capital, eXp was cash-flow positive AG9. That kept eXp solvent but capped what it could offer any one agent: a top producer got the same capped terms as everyone else, while Compass could bid more for the agents it most wanted. In the 2023 downturn Mike DelPrete found that low-fee brokerages, with "significantly lower fixed costs", were operating much more profitably than legacy firms, yet Compass kept its revenue lead and returned to agent growth AG10.

Rivals Compass raised more than $1.1 billion by 2019 and spent it on splits near 90/10, bonuses, marketing funds and $70 million of agent equity, clawed back in full from agents who left within two or three years CPX-10 CPX-13 CMX-20. Realogy sued it in 2019 for overpaying agents with investor money CPX-2.

Trade-offs & activity fitPenetration Pricing

Step 3 of 5 · 2022–25

Kept agents with pay, not software

When Compass ended cash and stock sign-on bonuses in August 2022 and leaned on its agent software instead, eXp answered with more of its own currency CMX-19. In 2024 it added the ICON Incentive Program and a revenue-share capping incentive, and launched REVenue Share 2.0, which let agents draw revenue-share payments instantly for a small fee; its filing says the programs were meant to hold agents through a market contraction and rising attrition AG1. The rest was support, not a proprietary system: FrameVR, multilingual agent care and online courses AG1.

Pay kept the door open in both directions. eXp's agent count fell 5% in 2024, from 87,515 to 82,980, and was still 5% lower a year later in mid-2025 AG1 CMX-17. eXp reports attrition rather than a retention rate CMX-16. The same portability brought teams over from Compass. In December 2024 Kris Caldwell moved his 30-agent, $200 million Colorado team, citing tools, training and "revenue-sharing opportunities" AG11. In 2025 the Denman Group, first in Massachusetts by volume, followed for branding freedom, training and revenue share CMX-4. Agents carried their clients either way, and neither brokerage held them with software.

Rivals Compass reported quarterly retention of its principal agents near 97% after the bonuses ended and recruited more than 2,000 agents without them by early 2024; in its own survey more than 80% of recruits named the platform as the main reason CMX-19 CMX-18.

Step 4 of 5 · 2018–26

Grew only by recruiting until 2026

This step spans the years of steps 2 and 3 because Compass bought brokerages with the same capital it used to recruit. eXp grew one agent at a time through sponsorship, paying recruiters in revenue share and stock rather than buying firms AG9 AG1. Compass took about half of the 8,000 agents it added in 2018 and early 2019 from brokerages it bought, gaining a market's agents, offices and local brand at once, at an estimated $56,000 per agent CPX-10. Sponsorship could not add a market's leading firm overnight.

eXp changed course only in 2026. In May, four months after Compass closed its purchase of Anywhere Real Estate, it bought NextHome, a national franchise, and in June it renamed the parent AGNT, a holding company for eXp Realty, NextHome, FrameVR.io and SUCCESS Enterprises AG8 AG2. Its agent count rose in the first half of 2026 mainly because of the NextHome agents AG8. The parent now includes a franchise model, so its results after mid-2026 no longer measure the virtual brokerage alone AG8 AG2.

Rivals Compass bought Pacific Union in 2018, the Christie's International Real Estate brand in 2025 and Anywhere, formerly Realogy, in January 2026, going from 36,990 agents to 84,187 brokerage agents in three months CPX-13 CO1 CPX-8. eXp Realty's CEO Leo Pareja answered that scale alone does not guarantee success CPX-5.

Step 5 of 5 · 2025–

More agents, less revenue

The agent count that sponsorship built in step 1 did not become a revenue lead. eXp ended 2025 with 83,060 agents, 60 more than a year earlier after heavy attrition, and $4.77 billion of revenue, up 4%, with a net loss of $22.7 million CPX-12. Compass, with 36,990 agents before Anywhere, reported $7.0 billion CPX-8 CO1. Both figures are gross commissions, much of them paid on to agents, so the gap measures how much business each agent brought in: what Compass recruited for and eXp's plan did not select on.

The low-cost model still works as a business. In the second quarter of 2026 eXp's cost per transaction fell 15% while its transactions rose 12%, helped by lower staff and marketing costs AG8. Agent satisfaction, measured by eXp's own agent Net Promoter Score, fell to 69 from 77 a year earlier AG8. eXp remains a large, low-cost brokerage with more than 80,000 agents, now one brand inside a multi-model parent CPX-12 AG2. It lost the contest to be the largest by business to the company that paid for productive agents.

Rivals Compass became the largest U.S. brokerage, with about 84,000 brokerage agents after Anywhere and $4.3 billion of revenue in the second quarter of 2026 alone CPX-8 CPX-7.

Trade-offs & activity fit