Spent $18,400 on a land lead site. 41 leads, one closing.
I finished shutting this down last month so the numbers are final.
The idea was narrow and I still think it was reasonable. Rural land owners who inherited a parcel and don't want it, in three states where I've bought before. A site, a form, some content pages about how to sell inherited land, and paid traffic. Sell the leads to two land buyers I know at $300 each, keep anything I wanted for myself.
Spend: $12,600 to a development shop for the site and a basic admin panel where buyers could log in and see leads. $4,100 in paid traffic over five months. $1,700 in content writing. $18,400 total.
Result: 41 form fills in five months. Of those, 22 were real people with a real parcel. My two buyers took 9 of them at $300, so $2,700 in revenue. One closed. I took two for myself and one of those went to contract and then died over an access easement that nobody had looked at.
Where it actually went wrong was step one, and it wasn't the traffic. I paid for the platform before I knew whether the traffic would produce anything. The $12,600 site with logins and a lead status pipeline was built for a volume of leads I never got close to. 41 leads over five months is a spreadsheet. My buyers used the login twice each and then asked me to just email them.
The second thing is that I never checked what a click cost in this niche before I committed. Inherited land searches are thin and expensive. I was paying real money for a handful of clicks a day, and at that volume you can't learn anything, because 6 fills a month tells you nothing about whether your page converts.
One buyer said something at the end that I keep thinking about. He said he doesn't need more land leads, he needs better ones, and $300 was already more than the same lead costs him from a mail campaign he already runs. So my price was set against nothing.
What I'd do differently: spend the first $2,000 on traffic to a free one page form and a phone number, with no platform at all, and find out the cost per real conversation before writing a check to a developer. If the number works, build. If it doesn't, I'm out two grand instead of eighteen.