The Hidden Cost of CRE Marketing Handoffs

Most CRE teams do not lose the most time writing the first draft.

They lose it in what happens before and after.

A broker wins the assignment. Property information gets pulled from an old offering memorandum, a rent roll, an email thread, a shared drive, a CRM record, a listing platform, and sometimes a spreadsheet only one person knows exists. Marketing assembles the package. The broker sends changes. Someone updates the OM. Someone else updates the flyer. The website needs the same change. So does the email campaign. A week later, the asking price changes again.

Every step is small. None of it looks like a major problem by itself.

But run that pattern across every active listing and the hidden workflow becomes a capacity constraint.

That is the part of AI in commercial real estate that interests me most. Not whether a model can write a better paragraph or generate a cleaner image. The bigger question is whether AI can remove the handoffs, re-entry, reconciliation, and revision work that consumes capacity across the brokerage.

The visible task versus the hidden workflow

When teams evaluate productivity, they tend to measure the visible artifact: How long did it take to create the OM? How quickly did we produce the flyer? How fast can AI write the property description?

Those are useful measures, but they miss most of the system.

The actual workflow starts earlier. Someone has to find the right source material, identify the current numbers, verify them, decide which version is authoritative, move the information into the next system, and then do it again when something changes.

That distinction matters because task time and flow time are not the same thing.

Microsoft’s 2025 Work Trend Index found that nearly half of employees said work felt chaotic and fragmented, while Microsoft 365 telemetry showed employees being interrupted, on average, every two minutes during core work hours. CRE marketing is a specialized workflow, but the underlying problem is familiar: coordination work expands around the actual work.

The real unit of productivity is therefore not “time to draft.” It is elapsed time and human effort from source information to market-ready execution.

Why revision cycles consume more capacity than first drafts

The first draft gets most of the attention because it is visible.

Revisions are where capacity quietly disappears.

A first draft is usually linear. Gather the information, create the asset, review it, publish it.

A revision is different because it creates dependencies. Change the asking price and the team may need to update the OM, brochure, website, email, social creative, CRM notes, and perhaps a presentation already sent to the client. Change a tenant name, square footage figure, broker headshot, disclaimer, or financial assumption and the same problem appears in a different form.

One field change becomes five file changes.

At that point, a person is no longer doing marketing work. They are acting as the synchronization layer between disconnected systems. That is the coordination tax we described in One OM. Three Revisions.: the edit is one line, and the sequence around it is the cost.

That is why revision speed is a more revealing metric than first-draft speed. A workflow that can generate an OM in ten minutes but takes two hours and three people to propagate an approved change is not a fast workflow.

It is a fast document generator sitting inside a slow operating model.

The cost of duplicated data and disconnected systems

CRE brokerages will always use multiple systems. That is not the problem.

The problem is when those systems do not share state.

If the CRM has one property name, the OM has another, the listing platform has a stale asking price, the marketing folder contains two competing brochures, and the campaign tool cannot tell which version was approved, the organization has created a reconciliation problem.

Deloitte has called out this foundation directly. Its real estate research has repeatedly identified fragmented, nonstandardized data as a constraint on AI, and its 2027 Commercial Real Estate Outlook, published this week, says the jump from AI pilot to production is still being held back by uneven data foundations, legacy processes, governance gaps, and the organizational work required to reimagine workflows.

The point is not that every brokerage needs a giant data transformation program.

It is that the workflow needs a trusted record.

For a listing, that means the property facts, financial assumptions, media, narrative, contacts, approvals, and source documents should resolve to a shared record that downstream work can use. When something changes, the system should know what depends on that change. That is the same architectural bet as treating listing data as the brokerage system of record.

APIs can help. Integrations can help. But moving data from Tool A to Tool B is not the same thing as maintaining a single operational truth across the workflow.

Why point AI tools can speed tasks without speeding the brokerage

AI adoption in CRE is moving quickly. JLL’s 2025 Global Real Estate Technology Survey reported that 92% of corporate real estate teams had started piloting AI or planned to begin that year, while only 5% reported achieving most of their program goals.

That gap is important.

JLL CTO Yao Morin summarized one prerequisite well: “A strong data platform is critical for growth.”

The same lesson is showing up outside CRE. McKinsey’s 2026 research on AI transformation found leaders were 5.3 times more likely to report enterprise value capture when workflows were redesigned than when they were left unchanged: 32% versus 6%.

This is why I think the current obsession with point AI tools can be misleading.

A point tool may write the description faster. Another may resize the image. Another may summarize the rent roll. Another may draft the email. Each can be useful.

But if a person still has to carry the output from one tool to the next, check whether the inputs are current, reconcile conflicting versions, and manually trigger the downstream work, we have automated keystrokes rather than the workflow. That is the difference between digitizing a task and operating from a connected system, which is the argument in From Point Solutions to AI-Native Operating Systems.

The brokerage may have faster tasks and the same capacity ceiling.

What an AI-native workflow requires

An AI-native workflow starts with a different assumption: AI is not a feature beside the work. It participates in the work.

That requires several things.

Shared data layer

The system needs a canonical property record rather than a collection of documents that each contain their own version of the property.

Provenance

When AI extracts a rent figure, square footage, lease expiration, or operating expense, a human should be able to see where it came from. Trust comes from traceability, not from confidence scores alone.

Persistent context

The system should know the brokerage’s templates, terminology, brand standards, preferred layouts, approval patterns, and prior decisions. Otherwise every interaction starts from zero.

Connected outputs

The OM, flyer, listing page, social content, campaign, and follow-up should not behave like unrelated projects. They are different expressions of the same listing.

Role-aware human control

Brokers, marketing teams, coordinators, and leadership do not make the same decisions. AI should accelerate the mechanical work while preserving clear review and approval where judgment matters.

Learning from decisions

The workflow should learn from decisions over time. Which template did the team choose? Which copy did the broker rewrite? Which assumptions changed before launch? Which buyers engaged? That decision exhaust becomes part of the firm’s operating knowledge.

This is where AI becomes more valuable than generic content generation. It begins to understand how the brokerage actually works.

What brokerage teams should measure

The easiest way to miss the value of workflow AI is to measure the wrong thing. Tokens generated do not matter. The number of AI prompts does not matter. Even “documents created” can be misleading.

I would start with seven operating metrics:

1. Time to market. Elapsed time from receiving source material to a market-ready campaign.
2. Human hours per listing. Not calendar time, but actual labor consumed across brokers, marketing, coordinators, and outside resources.
3. Revision latency. How long it takes an approved change to be reflected everywhere it needs to appear.
4. Number of handoffs. How many times work changes owners before completion.
5. Duplicate data entry. How many times the same property fact is manually entered or copied.
6. Rework and stale-data defects. How often the team has to correct an asset because the wrong version or value was used.
7. Capacity per team. How many active listings the organization can support without adding equivalent headcount.

These measures expose a very different picture from “AI saved me ten minutes writing copy.” They show whether the brokerage itself is getting faster.

A practical evaluation checklist

When evaluating an AI or CRE technology platform, I would ask a few questions before looking at the demo polish:

1. Does the system create and maintain a shared property record, or does every output become another isolated file?
2. Can it ingest the source material the team already receives and show the provenance of extracted facts?
3. When one approved value changes, can the system identify and update the downstream assets affected by that change?
4. Does it ingest and preserve brokerage templates, brand standards, preferences, and prior decisions?
5. Can brokers and marketing teams review and approve work at the right points without becoming the manual routing engine?
6. Does the workflow continue from pre-listing through listing, marketing, distribution, and follow-up, or does the AI stop after generating an artifact?
7. Can the platform work alongside the CRM, data, and listing systems the brokerage already depends on without creating another reconciliation layer?
8. Can the vendor show operating metrics such as hours saved, revision time, or capacity returned—not just examples of generated content?

If the answer to most of those questions is “no,” the tool may still be useful. But it is probably optimizing a task, not redesigning the workflow.

Where Antela fits

This is the problem we built Antela around.

Our goal is not to replace every system a brokerage uses. It is to create a connected operating layer around the property workflow so information can be reused from pre-listing through marketing and follow-up.

A representative listing workflow we mapped consumed about 31 hours across multiple activities and systems. In Antela, that same workflow was reduced to roughly 35 minutes of active production time. Separately, one brokerage accumulated 335 measured hours returned across 10 active users in its first months on the platform—33.5 hours per active user. Results will vary by team, activity level, and workflow.

The number matters, but the mechanism matters more.

Those hours did not come from asking people to type faster. They came from reducing repeated collection, re-entry, formatting, handoffs, and revisions across the same property workflow.

The listing record becomes reusable context. The source documents remain traceable. Marketing assets are created from the same underlying information. Human approvals stay in the loop. Follow-up can continue from the work that happened upstream instead of starting from scratch in another system.

That is what we mean when we describe Antela as an AI operating system for CRE brokerages.

The opportunity is not another faster tool

The most interesting question in CRE AI is no longer, “Can AI create this?”

In most cases, the answer is already yes.

The better question is: “What still has to happen after AI creates it?”

Who checks the number? Who moves the file? Who tells marketing the price changed? Who updates the website? Who regenerates the brochure? Who records the decision? Who follows up with the people who engaged?

If the answer is still a chain of people moving information between disconnected systems, the brokerage has not removed the constraint. It has only accelerated one step inside it.

The fastest AI tool is not necessarily the fastest brokerage.

The firms that create real capacity will be the ones that reduce the number of times people have to find, move, reconcile, and re-create the same information.

A simple test: the next time one fact changes on a live listing, count how many people, files, and systems have to touch it before everything is right again.

That number tells you more about your AI readiness than almost any demo.