I built itrackmine.com (http://mashable.com/2008/12/26/itrackmine/). And its killer recommendation engine (books, movies, and music) -- we knew what you owned from all stores, not just what you //bought// at Amazon (or single store) so ours were extremely accurate. Along with a "user A is this similar to user B" system...and the whole tracking, sharing, mobile app, barcode-scan, manage-your-stuff-package.
Made $10 over the ~8 years it was up...from one donation. Yey.
I'm very passionate about the collection management aspect but haven't found one that does it right... Tech has changed quite a bit and graph databases (which IMO are key to this) are much more accessible now.
I agree with your assessment of the potential and hope to work on this idea. If you are interested in focusing on the collector, let me know.
Neither my business partner nor I are any good at marketing. And we didn't like how most sites put so much pressure on their users to buy things, click ads, share with friends, send out marketing emails, etc...so the things we DID put in were kind of halfhearted, and always out of the way.
We tried selling the data (amazing data if you think about it!!). Anonymous of course! And/or selling the recommendations (easy enough since everything was organized by UPC/ISBN). But that's something it seems like you need to know people -- CEO's/CTO's of bigger companies. We just didn't have the connections, nor the personalities to kiss that much ass. But we tried for several years. I think they're all of the mindset that they've got "good enough" data on their own...or two guys in their garage are too small peanuts to trust. It's funny that just this week Amazon had a post here about it's 20th year of recommendation engine, ~yey~ -- and they still can't get it right (sour grapes? yeah...).
If we were to do it again, we'd probably focus on core functions (the site did A LOT). And seriously work on marketing aspects. Not dark patterns (https://darkpatterns.org/) but more organized and focused efforts.
Honestly though, if we did it again, it would not be for the money - it would be for a cool product and for our own fun...maybe integrate with Kodi...
We'd be smarter on hosting. We were at Rackspace first for $30k/mo...then realized we were being hoodwinked and found PEER1 for $1200/mo...then realized we could get away with a $300/mo DO droplet by converting everything to open source and optimizing -- granted over the course of the 8-10 years, tech improved. But, no more MSSQL, no paid search engine (we ended up on Solr -- AMAZING!), no Windows and just all open-source. I ran our recommendation engine on my computers at home and uploaded results from there. As with hosting, I think the open source stack has improved a lot.
Smarter on who we hire for PR - I got all our serious PR (Mashable, LifeHacker, Makeuseof, etc). We spend about $30k on snake-oil PR companies. Our first PR company took $20k and literally did nothing (not even a press release blast or PDF) - but suing would cost us way too much time/money so we ate it and went to another referral - who did the job, but it ended up more or less a pointless exercise.
After all that time, going into personal debt -- with creditors calling at all hours, no income in sight, thousands of hours devoted to it, alienating friends and family borrowing from them to survive, we finally decided to call it quits.
Made $10 over the ~8 years it was up...from one donation. Yey.