The Brokerage CMA Software Buyer's Guide: 12 Questions to Ask Before You Sign

CompsAgent Team · August 21, 2026 · 11 min read

This is the procurement companion to CMA software for brokerages, not solo agents, which covers what to evaluate on. This one covers what to actually ask, out loud, before money moves.

Before the demo

1. What is our real monthly report volume?

Ask yourself, not the vendor. Count listing appointments from last quarter, not opportunities or leads. Almost every brokerage overestimates this by a factor of two or three, and the estimate is what per-seat pricing gets sold against.

If you can't answer this, you're not ready to price anything. Everything downstream — per-seat vs. per-report, tier selection, ROI — depends on it.

2. What specifically is broken right now?

"Our CMAs could be better" isn't a spec. Narrow it to one of: they take too long, they look inconsistent across agents, the data is wrong or thin, or they don't win appointments. Different problems, different tools — and a vendor will happily sell you a solution to whichever one their product handles.

Data and output

3. Where do the comps come from, and does the report say so?

The most important question on the list. Ask directly: MLS feed, public records, aggregated listing portals, or a proprietary model?

Then ask the follow-up that matters: for each comp, is it a confirmed closed sale, a list price, or an estimate — and does the client-facing report disclose which? A report that presents an estimate with the same visual weight as a closed sale is a liability at a listing appointment where a seller has done their own homework. Background in MLS data vs. public listing aggregators.

4. How does it behave on a hard property?

Don't let them pick the demo address. Bring three of yours:

• A straightforward tract home with plenty of comps — the easy case
• Something unusual: acreage, heavy custom work, a non-conforming layout
• Something in a thin-comp pocket where you know the honest answer is "there isn't much"

The third one is the real test. A tool that confidently produces five comps where you know only two exist is telling you it will stretch, and it will stretch in front of your clients too.

5. Can I see two different agents' output side by side?

Consistency is a brokerage criterion with no solo-agent equivalent, and it's easy to check. If two agents' reports on the same property look meaningfully different, you don't have a brokerage tool — you have thirty personal ones with a shared invoice.

6. Who controls the branding, and at what level?

Specifically: can you set the firm-level template centrally, and can an agent override it? Both answers matter. Full agent freedom kills consistency; zero agent personalization kills adoption. You want firm-controlled frame, agent-controlled name and photo.

Adoption

7. Can a skeptic beat their current method on the first try?

The single best predictor of whether this purchase works.

Have your least tech-enthusiastic producer build one report, cold, with no training and nobody helping. Time it. If it beats their current process on attempt one, adoption takes care of itself. If it needs a session and two weeks to reach parity, most of your roster will quietly keep doing what they did before — and you'll pay for seats nobody opens.

A pilot staffed with volunteers tells you what your enthusiasts think. Enthusiasts were never the adoption risk. Put two skeptics in the pilot or don't run one.

8. What does onboarding actually require of each agent?

Get it in hours, not adjectives. "Quick and easy" means nothing. A required 90-minute training session across 30 agents is 45 hours of selling time before you see a single report — a real cost that never appears on the quote.

Commercials

9. Can I reduce seats mid-term?

Most annual contracts let you add seats instantly and remove them only at renewal. Rosters shrink as well as grow. If you sign for 30 and finish the year at 24, you want to know now whether you're paying for six empty seats for eight months.

10. What's the renewal price, and is the increase capped?

Ask for the cap in writing. An attractive first-year number followed by an uncapped renewal is the oldest structure in software sales, and it works because year-two you is a captive customer with agents who've built habits.

Term	What to ask for	Why
Auto-renewal	Notice window, in days	30-day windows are easy to miss and roll you into another year
Seat reduction	Mid-term, with notice	Rosters move both directions
Price escalation	Capped % at renewal	Uncapped renewal is where the margin lives
Data export	On termination, in a usable format	Otherwise leaving costs you your history
Pilot terms	Paid pilot creditable against year one	Vendors will usually agree; almost nobody asks

11. What happens when an agent leaves?

Three sub-questions, and vendors are rarely asked any of them:

• Does the seat transfer to their replacement, or do you buy a new one?
• Do their past reports and client history stay with the brokerage, or leave with the login?
• Can the departing agent keep using reports carrying your firm's branding?

Given replacement costs of $15,000 to $50,000 per departing agent, losing their client history on top is a compounding own goal. And it's not hypothetical — turnover is a routine event, not an exception.

12. What does it cost to be wrong?

The question that should govern the whole decision. If this is the wrong tool in month four, what happens — contractually, and in migration work?

A per-report tool with no contract costs you nothing to abandon. An annual enterprise platform costs you the remaining term plus the effort of moving everyone back. That difference should be priced into the comparison, and almost never is.

The short version

If you only ask three, ask these:

1. "Show me a comp on a property in a thin-comp area, and tell me whether it's a confirmed sale." Tests data honesty.
2. "My most skeptical agent will build one now, untrained, while we watch." Tests adoption.
3. "What does cancelling in month four cost me?" Tests everything else.

The bottom line

Brokerage software purchases rarely fail because the software was bad. They fail because the buying process only examined the parts the vendor chose to show, and the failure surfaced in month five as an adoption problem, a data embarrassment in front of a client, or a renewal invoice nobody budgeted for.

All three are visible in advance. They're just not visible from inside a demo.

Frequently asked questions

What should a brokerage ask before buying CMA software?

Establish four things the demo will not cover on its own: where the comp data actually comes from and whether the report discloses it, what happens to reports and client history when an agent leaves, what the contract costs to exit early, and whether an unenthusiastic agent can beat their current method on the first attempt with no training.

Should a brokerage run a pilot before rolling out CMA software?

Yes, and the pilot group should include skeptics rather than only enthusiasts. A tool that only the early adopters like will show excellent pilot metrics and then stall at roster-wide rollout, which is the most common way these purchases fail.

What contract terms matter most in real estate software?

Auto-renewal windows, mid-term seat reduction rights, price escalation caps at renewal, and data export on termination. Seat reduction matters most for brokerages, because rosters shrink as well as grow and most annual contracts only let you add.

How do you evaluate CMA data quality?

Run three addresses you know well — one straightforward, one unusual property, one in a thin-comp area — and check the comps against what you know actually sold. Ask whether each comp is a confirmed closed sale, a list price, or an estimate, and whether the report tells the client which.

Related reading

• CMA Software for Brokerages, Not Solo Agents
• MoxiWorks Alternatives for CMA and Listing Presentations
• Automated CMA vs. Manual Comps
• MLS Data vs. Public Listing Aggregators