My craziest 24 hours as a YE
The Background
Despite what I wrote the other day about the 3 keys to SportsLizard's success, there's more to it. You and I both know that it'll be tough to be profitable relying solely on advertising. I've sort of glazed over it previously, but that plan is just a stall until we can launch an online price guide (which will have a free and premium version). That would make advertising a secondary revenue stream, something I'm very comfortable with.
Anyone in the collectibles industry knows that the incumbent way of pricing is bad. Not only is the data from an extremely small and old sample, it tends to be artificially inflated either intentionally or unintentionally - you NEVER get book value for an item - and price guides pretty much serve as a barometer and are not nearly as exact as they could be. Online price guides are essentially print versions of their counterparts.
The plan with SL was to increase the number of sellers and items listed until they grew from 10k to at least 200k and then use that 200k to create a live price guide. There are several problems with this:
With that, I decided earlier this week to try to figure out a way to get my hands on some more data. I had a few different options, but I came up with the idea of crawling all of the collectibles sites on the web and having the data dumped into the database...similar to how Google crawls the web. I found Web Scraper Plus, a PHENOMENAL piece of software that allows you to crawl sites and insert their data into your database just as I was looking for.
After toying with the software for a few days, I got it to do exactly what I wanted it to do....except there were a ton of problems:
My Night
I went to bed a bit deflated. All of my sites have been successful sites, but not so successful businesses. With the exception of SEO work I haven't had a hell of a lot of luck making money as an entrepreneur. I couldn't bear to have another realization that I should switch directions. I know the collectibles industry better than I know anything else, and I think the plan we had this time around was so solid that it was killing me to think it wouldn't even be feasible.
My "Morning"
Back pain from my still sore lower back woke me up around 2:45 AM. As I drifted in and out of sleep, I thought of a possible way to make the price guide work and solve all of the aforementioned problems. I decided to get up and give it a shot. 6 hours later I had a working demo for my partners to wake up to!
I used Google's API to get the data I needed. Beckett, eBay, and many other large collectible sites list their items on Froogle, and that data is therefore made available. When the user submits a search on my demo, it queries Froogle's data and returns a feed with the results to me, which I then parse and can do whatever I want with. We don't need to store millions of products on our server, and we shouldn't need to worry about queries crashing our server - the heavy work is done on Google's end.
Over the next few days I'll develop an algorithm to incorporate SportsLizard's search and purchase data, previous price guide queries, and Yahoo's search data to measure supply and demand of a collectible. The "report" a user gets will be both simple and complex - they can just take our "final price" and run with it, or see all of the factors and how they contribute. And oh yea, it will be free to run 3 queries/day and cost $4.99/month to be able to run unlimited queries and save "my collection" among other things.
It's now 11:20 AM and I'm still wide awake...partly from the Red Bull's...but mostly from the elation of getting this to work. From the low of all lows and the feeling that I was a failure to the high of all highs and the feeling that SL is primed to explode. God I love what I do. Look for the price guide to launch in a week or two. I'm going to go relax and get some well deserved rest.
Despite what I wrote the other day about the 3 keys to SportsLizard's success, there's more to it. You and I both know that it'll be tough to be profitable relying solely on advertising. I've sort of glazed over it previously, but that plan is just a stall until we can launch an online price guide (which will have a free and premium version). That would make advertising a secondary revenue stream, something I'm very comfortable with.
Anyone in the collectibles industry knows that the incumbent way of pricing is bad. Not only is the data from an extremely small and old sample, it tends to be artificially inflated either intentionally or unintentionally - you NEVER get book value for an item - and price guides pretty much serve as a barometer and are not nearly as exact as they could be. Online price guides are essentially print versions of their counterparts.
The plan with SL was to increase the number of sellers and items listed until they grew from 10k to at least 200k and then use that 200k to create a live price guide. There are several problems with this:
- 200k is too small of a sampling of the 10-20 million collectibles available online. A stats dork like me is obsessed with sample size and will do anything he can to get it. It just won't be accurate with such a minuscule sampling.
- I realized that no matter how hard I try I'll only get 10% of the small/mid-size stores to participate. That's great for the marketplace (that'll be a shit ton of items) but not so good for accurate pricing
- It neglects eBay, Beckett, and Naxcom - three sites that in my estimation account for ~70% of the sales of cards and collectibles online
With that, I decided earlier this week to try to figure out a way to get my hands on some more data. I had a few different options, but I came up with the idea of crawling all of the collectibles sites on the web and having the data dumped into the database...similar to how Google crawls the web. I found Web Scraper Plus, a PHENOMENAL piece of software that allows you to crawl sites and insert their data into your database just as I was looking for.
After toying with the software for a few days, I got it to do exactly what I wanted it to do....except there were a ton of problems:
- It would probably take my desktop a few days to do a crawl of these immense sites
- Our database would fill up with millions of items...pushing storage space and possibly killing the server when running queries
- There would need to be a lot of quality checks for all this data. If eBay made one change to their site, for example, our crawling might get messed up and the data be invalid.
My Night
I went to bed a bit deflated. All of my sites have been successful sites, but not so successful businesses. With the exception of SEO work I haven't had a hell of a lot of luck making money as an entrepreneur. I couldn't bear to have another realization that I should switch directions. I know the collectibles industry better than I know anything else, and I think the plan we had this time around was so solid that it was killing me to think it wouldn't even be feasible.
My "Morning"
Back pain from my still sore lower back woke me up around 2:45 AM. As I drifted in and out of sleep, I thought of a possible way to make the price guide work and solve all of the aforementioned problems. I decided to get up and give it a shot. 6 hours later I had a working demo for my partners to wake up to!
I used Google's API to get the data I needed. Beckett, eBay, and many other large collectible sites list their items on Froogle, and that data is therefore made available. When the user submits a search on my demo, it queries Froogle's data and returns a feed with the results to me, which I then parse and can do whatever I want with. We don't need to store millions of products on our server, and we shouldn't need to worry about queries crashing our server - the heavy work is done on Google's end.
Over the next few days I'll develop an algorithm to incorporate SportsLizard's search and purchase data, previous price guide queries, and Yahoo's search data to measure supply and demand of a collectible. The "report" a user gets will be both simple and complex - they can just take our "final price" and run with it, or see all of the factors and how they contribute. And oh yea, it will be free to run 3 queries/day and cost $4.99/month to be able to run unlimited queries and save "my collection" among other things.
It's now 11:20 AM and I'm still wide awake...partly from the Red Bull's...but mostly from the elation of getting this to work. From the low of all lows and the feeling that I was a failure to the high of all highs and the feeling that SL is primed to explode. God I love what I do. Look for the price guide to launch in a week or two. I'm going to go relax and get some well deserved rest.
posted by Adam McFarland at
4/13/2007 10:37:00 AM
Links to this post:
Create a Link
<< Home