I’m not looking for the flashy stuff or the exciting wins.
I’m talking about the unsexy, repeatable, sometimes annoying things you do that quietly move the needle the most.
Examples could be:
What’s the boring thing you’d never want to give up now, even though it’s not fun?
Curious to see if any patterns show up in the responses. 
I used to not follow up with people at all. For a long time, it wasn’t that vital of a practice, because you could always send more marketing to new places and find the bottom-of-the-barrel, most motivated sellers.
Now, we follow up until we find a big problem with the property, or until the owner tells us to stop. There are a lot of people who just need to be contacted four, five, six, sometimes ten times before they’ll wake up and start talking to you.
Following the script when underwriting land deals. There are no shortcuts. No one is above the process.
Land is inherently difficult to value and most folks overestimate their capabilities (or they just accept the word of an ‘expert’ without doublechecking the data).
Find the extra comp. Make the extra call for utility availability. Ask the extra question. Spend the extra hour (or ten) reviewing.
Your assumptions are your downfall. Getting attached to deals kills your business.
Once the wire goes out, there are no take-backs.
Underwriting over everything.
====
Secondarily, building AI workflows with the appropriate amount of context. AI is only as good as the context you provide it. Proper context can take hours and hours of focused work and constant tweaks.
@chrisduff, those are great. Well said!
Have you used Deep Research from ChatGPT, Perplexity, or Gemini on any properties yet? I tried it on a potential development deal in Colorado last week. I fed it plenty of information on the prompt (property coordinates, parcel map, and literally everything I had researched about it) and let it go to work. It brought back a TON of information about the local economic outlook, surrounding competitors, possible alternative uses, and which ones were optimal for the location… stuff I hadn’t even thought to research myself.
I was glad to have the additional information and narrative about the property so I could say ‘NO!’ with more confidence, haha.
That’s a great anecdote, and a route that we need to experiment with more internally, particularly as subdivisions become a larger part of our portfolio, and once the initial comp set is promising (something out-of-the-box LLM models are still very limited at). Thanks!