What My Sales Data Actually Said: Turning Notes Into Strategy With AI

Every week, I read through our leasing reports. I want to know what prospective renters loved, what made them hesitate, which competitors they’re visiting, who was ready to sign, and who still needs time.

It’s incredibly helpful in shaping our strategy and tactics.

I try to take it in objectively, like a scientist. But in reality, if I’ve heard the same objection three weeks in a row, my brain likely starts weighing it heavier than maybe it deserves. That’s human. Recency bias exists.

And while I might catch patterns over a month, I know I’m not wired to catch every insight that’s buried in a full quarter or year’s worth of notes.

Luckily, AI is. 🙂

What If Our Notes Could Talk Back?

In 2025, one property had nearly 750 tours, and the rockstar sales team documented notes in 100% of the records.

First and foremost – kudos to this team for their incredible consistency. That’s impressive.

I knew there were impactful, actionable insights just sitting in this data if we could work through it, so I partnered with AI to see what I could learn.

Through the analysis, I created a specific leasing guide with case-by-case scenarios for our leasing team and a living example of what anyone can do with thoughtful data, good questions, and the right kind of AI partnership.

You don’t need to be in multifamily real estate to apply this. Replace “tour notes” with sales calls, discovery sessions, onboarding forms, or any qualitative info you’ve been sitting on. This is about using AI to think with you, not for you.

Here’s How You Do It:

1. Determine the details.

Pull a report with the data you want to “talk to,” along with any relevant surrounding details you’d like as context for your conversation.

For example, I pulled a report with notes from every single tour we had in 2025, along with lead sources, statuses, and key milestone dates. Then I cleaned the data to ensure it was structured in a way the AI could easily understand. Everything I needed to answer my questions was included, and nothing more.

Be sure to remove any personally identifiable information and cut out fields that aren’t relevant. The more you include, the more the AI has to process. And the more noise you introduce, the higher the risk of confusion. If a column doesn’t add value, cut it.

2. Test what the AI already knows.

Share your data without explanation or context and ask, “What do you understand about this data without me providing any additional information?”

It helps reveal which fields might need clarification and which pieces of industry lingo you’ve been assuming everyone (including the AI) already understands.

Then switch your position and ask what questions the AI has about the data. There’s almost always something you can explain to improve the results.

3. Give it a minute. Literally.

Use “thinking mode” (slower processing). You can select this by clicking on the arrow next the ChatGPT 5.2 in ChatGPT. This isn’t a race. The model took about 7 minutes to review, analyze, and apply reasoning to my data. This pause makes the outputs noticeably smarter.

4. Set the tone: no yes-men (or women) allowed.

I’ve trained my AI to challenge me. I don’t want flattery. I want friction.

This is built into my AI’s personality through the personality settings, but you can also get there manually by prompting it to:

“Push back if this doesn’t make sense.”
“Tell me if there’s a flaw in my logic.”
“What am I not asking that I should be?”

If you want to change the personality permanently, click on your name at the bottom left of your screen > click settings > select personalization > add instructions in the “custom instructions” section.

5. Tell it who the output is for.

This changes everything.

Before asking your AI thought partner to analyze and report back, tell it who will read the output:

The sales team
A board member
The operations team

That frames the style, tone, level of detail, and angle of the output.

6. Collaborate on better questions.

Lead with your curiosity. Say what you’re genuinely trying to figure out. Maybe it’s: “What percent of tours mention a competitor?” or “Which ones show up most often?” or “Does saying X in the notes usually lead to a signed lease?”

Then let your AI play editor. Ask it to rewrite those questions in a way that’ll get you sharper, more actionable answers.

And finally, ask what questions you should be asking, but haven’t.

7. Don’t settle for shallow insights.

The first result? Too vague.

“Some prospects cited location as a concern.”

Nope. I need specifics:

“What percentage of prospects cited location, and what was the close rate among them?”

Now we’re getting somewhere.

Specificity and substance matter. Don’t settle for surface-level responses. Dig deeper.

8. Check its homework.

YOU are the expert. If something doesn’t track with your experience, question it.

For example, in this exercise, the AI said our top competitor wasn’t showing up often.

I knew that wasn’t right.

Turns out, the AI missed that people wrote the competitor’s name various ways. Sometimes in shorthand, or by building number, or with a typo.

When I pushed back, the AI looked through the data again and explained it was using “exact name matching” rather than “pattern matching reasoning.” Once I corrected that context, the output shifted.

9. Format matters.

Make your report output what you want to see. Sometimes you want a narrative. Other times, succinct bullets do the trick. Or maybe you need a combination of both.

Just ask for it the way you want.

AI’s great at editing… if you tell it how.

10. Add psychology to the mix.

Once we saw the patterns, I asked AI to suggest sales strategies based on behaviors:

“If someone mentions pricing twice, what approach might help the team close?”

“If a prospect brings up location as a concern but still books a tour, what message would help reinforce value?”

“When someone asks detailed questions about amenities early on, does that signal a ready-to-lease mindset, or are they just browsing?”

This is where it gets really good.

The Outcome

I walked away with a report that helps our leasing team instantly understand where a prospect might be in their decision cycle. Whether they’re warm, cold, comparison-shopping, or ready to sign today.

And not just based on gut feeling, but on patterns from 748 real interactions.

What This Means For You

You probably already have the data.

Meeting notes. Call summaries. CRM entries.

What you might not have is a habit of listening to that data with help from a critical, curious, AI-powered partner.

You don’t need a custom GPT (although I absolutely turned my chat into one, so I can do this as often as I want).

You just need to:

Ask better questions
Give your AI context
Push it to think, not just respond

Your Turn

What’s one piece of data you’ve been sitting on that deserves a second look?

Open that spreadsheet. Drop it into AI, and ask:

“What might I be missing here?”

If you try it, I want to hear what surprised you.

This is how the work gets better.

reneemcintyre.com