Keeping Sellers Informed, Every Week

Client context

The Joan Kagan Team is a Manhattan real estate team at Compass, led by founder Joan Kagan, who has been in the business for 18 years. The team lists condos and co-ops on the Upper East Side and Upper West Side, mostly in the $1.5M to $5M+ range, and most of its business comes from word of mouth. Melissa Monroe, the team’s Director of Operations, runs the back office remotely and has worked with Joan for six years. The team carries anywhere from a few active listings to as many as ten at once.

The challenge

Sellers want to know what is happening with their listing, and the most common complaint sellers have about their brokers is not hearing enough. The team wanted to send a real update every week. Building one meant pulling from at least seven places:

• Compass Insights and StreetEasy listing reports, exported as PDFs
• UrbanDigs market charts for the right neighborhood and unit type
• A weekly open house benchmark email from a Compass colleague
• The team’s own inquiry and showing spreadsheet, which holds far more than a seller should see
• Digital ad and virtual tour reports
• Notes from the agents about what marketing went out that week

Melissa would put a chart on one screen and copy the numbers across by hand: days on market up this much from last year, down that much from last month. Joan reviewed everything before it went out, which made her the bottleneck. By the time a report reached the seller, it was often a week old.

Compass offered automatic analytics reports, and for a while the team sent those instead. They didn’t work. In Joan’s words, “It’s just numbers. It doesn’t really have meaning or message.”

The cost of the problem

A single report took around an hour and a half of combined work from Melissa and Joan, across several tools. Reports often went out late, got cut short, or were skipped. Before the move to Compass the team managed one every two weeks. After it, they leaned on the automated analytics instead, which sellers found too thin.

With ten listings at once, a proper weekly report for every seller was more work than the team could carry. As Joan put it in the first discovery call, “It’s too much to do on a weekly basis.”

The goal

A weekly report that is fast enough to send for every listing, tells the seller a clear story instead of listing numbers, and keeps the team in full control: nothing reaches a seller until Joan has reviewed it, and the AI never invents a figure.

We're taking all the same information sources, but when we upload it, it is giving us something that looks more cohesive and tells more of a story. It's more digestible for a seller.

— Joan Kagan, Founder, The Joan Kagan Team at Compass

The existing workflow

The AI system

A custom web app that works like a guided checklist for one listing’s weekly report.

Sources go in on the left: the team uploads or pastes each week’s material, one step per source. The report builds on the right: every section of the seller’s PDF updates as the sources come in, and each step is color-matched to the part of the report it fills, so the team can see exactly where every input lands.

The flow for a week: sources added → figures read directly from the PDFs, spreadsheet and emails → AI drafts the highlights, marketing summary, online performance story, market context and next steps → the team edits anything on screen → Joan reviews → a branded PDF goes to the seller.

Implementation

The build included:

• Spreadsheet reading for the team’s inquiry tracker, from an upload or a Google Sheets link. It counts inquiries, first showings and return visits, treats future dates as scheduled viewings, and ignores rental and spam inquiries.
• PDF reading for Compass Insights, StreetEasy and digital ad reports, plus an email reader for the weekly virtual tour traffic report.
• Live UrbanDigs charts for the listing’s market segment, with a labeled comparison group so sellers see their listing against true peers.
• A seller-requested view of new versus repeat showings, added after a seller asked how many visits were second looks.
• A choice for buyer feedback on every listing: the team’s exact wording, or an AI summary of common themes (“3 buyers mentioned the entryway”).

It was built in short rounds on live feedback calls with Joan and Melissa between June and October 2026, with most requests shipped within days.

Human oversight and safeguards

• Joan reviews every report before it reaches a seller. Nothing is sent automatically.
• Numbers come from the sources, not the AI. Listing metrics, ad results and tour traffic are read by code, and the AI is blocked from changing them.
• No invented data. The AI works only from the week’s sources and is instructed never to fill a gap with a guess.
• Private information stays private. Buyer and broker names never appear in the report. The inquiry spreadsheet stays with the team; only counts and anonymized feedback reach the seller.
• The team’s words stay theirs. Buyer feedback is never rewritten unless the team asks for a summary, and every AI-drafted line is editable before export.

Before vs. after workflow

Results

Figures are working estimates from a recorded interview with the team in October 2026, not a formal time study.

Business impact

The bigger change is in the seller relationship. One analytical seller, after getting the new report, started asking specific questions about her own data. Joan described it as “a communication tool… it got the dialogue going more.” A new seller also gets the report and, in Joan’s words, “appreciated the detail.”

Quiet weeks no longer look like inactivity. When a listing has no visitors, the report shows the market context around it, so the seller can see it isn’t just their apartment.

The time back goes where the team wants it. Melissa uses it to help the team’s other agents with calls and newsletters. Joan is spending it on herself: a little more sleep and exercise.

Client quote

One of the biggest negatives that sellers would have about their brokers was that they're not getting enough information. When you schedule it regularly, you're giving them the information. They can't go back and say, you're not telling us what you're doing.

— Joan Kagan, Founder, The Joan Kagan Team at Compass

You're proactive, not pushy. You didn't miss any feedback. You're very detail-oriented.

— Joan Kagan, on working with Finlay Systems

What happened next

The tool is in use on the team’s active listings and keeps improving from real weeks. On the same day as the interview, a feedback request from Joan (letting the team choose between its own wording and an AI summary) went live in the app. Joan now wants the team’s other agents building their own reports in it, and the team agreed to a follow-up conversation to scope the next automation: getting showing feedback from agents’ texts and voice notes into the inquiry spreadsheet without retyping it.

What this means for other real estate teams

Every listing team has the same problem: the data sellers want is spread across the brokerage’s analytics, the listing portals, market data tools and the team’s own tracker, and turning it into a weekly story takes longer than anyone has. The pattern here applies to any team that wants consistent seller updates without hiring for it: read the numbers straight from the sources, let AI draft the story, and keep the agent as the final reviewer.