Skip to main content
Light-brick building with tinted windows, between two red-brick facades.
← Blog

user-journeydiscoverytrust

Real estate advisors facing automation: a matter of accountability

A tool executes a task, it does not answer for the result. How does a profession's accountability draw the boundary of automation? Notes from a user-journey mapping conducted with real estate advisors.

In real estate transactions, execution tasks automate well: publishing listings, following up, formatting, collecting documents. As the imminent end in France of la pige (prospecting private sellers by phone from listing data) signals the loss of an acquisition channel, adopting the right tools becomes vital. Yet resistance persists at the moments when the advisor states something specific about a property, a price or a situation. They will have to answer for it before the seller and the buyer, in a local market where a reputation takes years to build.

A tool can execute a task, but it does not answer for the result.

Advisors draw this line themselves, in the way they work. The observation comes out of a user-journey mapping and interviews conducted with advisors from several organizations, whether agency-based or working as independent agents.

The advisor, the trusted intermediary

Travel agents carried a similar fear about the automation of their trade, and about its impact on their clients' trust. Any profession that lives off its expertise has an interest in describing it as impossible to automate. It is a sector under strong pressure from disintermediation, meaning that suppliers bypass intermediaries to sell directly to the consumer, and so capture part of the agencies' added value.

If the argument holds, the comparison with real estate still needs nuance. A trip is booked in a few hours and can be forgotten on return, while a real estate transaction plays out over months, commits significant amounts for years and sits within a strict legal framework.

Mapping the user journey does not dispel that worry, but it does locate where the friction sits today.

The journey starts in self-service. The buyer searches and compares on listing portals, long before having a file ready. The seller asks around, gets exposed to ads, forms an idea of the price. The client prefers this autonomous phase and the advisor does not compete with it: they are its endpoint, the one people end up contacting through a platform or on recommendation.

For most clients, a real estate transaction is a rare event and the amounts at stake forbid mistakes. They look for someone who commits to what they claim and who brings mastery of their territory. The advisor and the client then build a relationship of trust, which starts with a mandate (the French listing agreement) and can open onto property management or advisory services.

A price has to be unlearned

In the seller journey, the valuation is the entry point. The seller arrives with a price in mind, built on more or less rational elements: the purchase price, the renovation work done, properties seen online, the online estimate, the amount needed for the next project, old memories. The advisor wins the mandate, even the exclusive listing when they negotiate well, by working the gap between that anchor and the reality of the market. A gap that moves with the economic conditions of the moment.

Automated valuation follows a logic of its own. It sets a first price, potentially flattering, before the advisor is even involved. The models produce decent results on standard properties, and can miss what makes an exceptional one. But the reluctance persists, because price touches what founds the advisor's legitimacy and expertise, and because they will have to argue against a machine, already deployed, that never has to justify itself.

The machine saw sunlight in the land registry

On the surface, writing listings is an ideal task to automate: known format, repetitive text, structured information, market data and field characteristics, everything is ready to feed a language model.

Yet it is an area where reservation remains, because a listing is a public statement about the property, published under the signature of the advisor and their agency. It engages the agency's liability, from mandatory disclosures such as the energy performance certificate (DPE) to the accuracy of the stated surface areas.

The tool works on data, the advisor knows the property, and the gap shows in the texts produced. On one side, generic descriptions, dressed with superlatives and other verbal tics, interchangeable from one property to the next. On the other, invented but plausible qualities: a bright living room deduced from the orientation, or sunlight inferred from the location. Nothing crude, nothing you catch without rereading everything. The time saved on writing is lost on rereading and validating the positioning of the listing.

The vendors' product approach thus collides with the reality of the field. The tools are designed for a functional benefit, when what an advisor needs is business: generating qualified leads and winning the mandate, with exclusivity. A generic automation, which smooths their message and makes it interchangeable with the agency across the street, strips them of their arguments at the very moment they need to convince.

Fix the sky, do not move the walls

Property marketing shifts the question. Photography, video, drone, virtual tours: technology and production, calibrated to the economics of the mandate. A professional capture session can cost several hundred euros, while a virtual tour service costs several thousand euros of equipment plus a monthly subscription for the agency. Meanwhile everyone has a smartphone in their pocket and can shoot and edit images quickly.

These investments are hard to justify on a simple mandate, but can become profitable on an exclusive listing, and are part of the expected service at the high end.

A good photo raises the perceived value of the property without creating a promise the visit will not keep. It influences the click on the portal, the property that stands out from the rest, the spark, then the first contact.

Generative AI finds its place, in virtual staging, as long as what is shown stays attached to reality: a corrected sky or brightness, an empty room furnished to help people project themselves. Except the drift is one click away, enlarged rooms, facing buildings erased, enough to set up a disappointment at the visit, carrying the same share of responsibility as a misleading description.

Terrace before and after virtual home staging: bare on the left, furnished and planted on the right.
Before and after of a virtual home staging on a terrace.

Advisors do not reject AI as such. Many assemble their own toolbox from general-purpose services, to retouch an image or rough out a text. What they discard is the generic module that produces the same listing for every agency. The difference lies in control: they choose the tool, provide the material and the context, and validate what gets shared.

Beyond the gadget: back to the professional tool

If the limit runs through responsibility, the need for automation lies elsewhere, where advisors ask for it and do not find it.

The vendor market, for its part, concentrates on integrated suites, consumer-facing estimators backed by price models, editorial AI modules inside CRMs, in a logic of supply more than needs. But no integrated suite closes the gap between what the tool knows and what the advisor observes on site. The opportunity space splits today into two axes:

  • Operational efficiency (the back office)

Retrieving diagnostics, building and tracking files, compliance, access to the local zoning plan (PLU), follow-ups, and so on. All of this falls within the scope of the CRMs agencies already pay for and use at a fraction of their capacity, because the right tool can be buried in a rich, dense interface.

  • Acquisition and presence (the front office)

Visibility is shifting from call volume to a continuous presence, local and online. Being the one people recommend before they even have a project means existing in the self-service phase, the one that decides who gets contacted.

Most advisors do not necessarily have the time or the method to build that presence, especially independent agents (mandataires). Large organizations have marketing teams. The independent, meanwhile, faces this need for personal branding. Yet part of the material already exists. The content produced for listings disappears with them.

The last summer of la pige

In France, law no. 2025-594 of June 30, 2025 rewrites article L.223-1 of the Consumer Code. From August 11, 2026:

It is prohibited to canvass by telephone, directly or through a third party acting on its behalf, a consumer who has not previously expressed consent to being the subject of commercial prospecting by this means.

Consent must be free, specific, informed, unequivocal and revocable, given by a clear affirmative act, and the burden of proving it falls on the professional.

A dedicated checkbox, unchecked by default, can be enough to collect it on a real estate platform, provided that it clearly states which professional may use the number and for what purposes, and that it offers a simple way to revoke it.

La pige calls private individuals who asked for nothing: it falls within the scope of the ban, subcontracting included.

Add to that a budget paradox: an agency can spend several thousand euros a month between advertising and listing distribution. That budget rents audience from the portals but builds no visibility of its own, no brand. Those platforms keep capturing the demand and distribute the contact details of those who consented, while agencies have to invest in what makes them recognizable, worth recommending and reachable on those platforms and beyond.


From August 11, 2026, prospecting requires tracking, while seeking to be less intrusive.

Advisors need tools they choose and trust, to grow while remaining accountable for their image and for what they stand behind, without being stripped of what makes their value.

Sources and references

Photos by @objectifbenb - Article co-produced with AI.