Online shopping and analytics service Mpire announced its first two sponsored buying guides today, a home and garden catalogue sponsored by Zillow and a baby and toddlers guide sponsored by Judy’s Book. Mpire mines data from transactions on eBay, Craigslist, Yahoo and Overstock.com to display price trends and quantities of items searched for. The buying guides are sponsored, stylized displays of that data in particular shopping niches. Unfortunately, I don’t think they work very well.
Mpire is a 16 person company based in Seattle that just launched this month. The team is interesting. Matt Hulett, Chairman & CEO, used to be the President of the corporate travel division at Expedia. J. Richard Rock, former Senior Director of Business Development at eBay, is a primary investor. Philip Rosedale, Founder and CEO of Linden Lab (the home of Second Life) is an advisor. With a team like that I can imagine that great things are possible.
I like the price trend graphs that Mpire offers and if I refine my searches enough to exclude accessories related to my search terms then the data looks helpful. These new sponsored buying guides don’t do that though. One highlighted item, for example, is a Weber grill. When I see it in the Mpire catalogue there’s a nice low average price listed, but when I click through I see eBay listings that include grill covers, replacement parts and other things that include the product name in the search results. This makes the average price of the grill itself appear much lower than it actually is.
The company says that Mpire provides helpful data on real market value for items bought online. That may be true for some items, but I don’t think it works with these buying guides. In my conversation with the company, Styx concert t-shirts were a favorite example. I imagine that the real market value of obscure items are well delivered by Mpire, but I’m not so sure about the kind of very general search terms highlighted in these sponsored buying guides.
Search results are also dominated by eBay auctions because low prices are what’s prioritized. Craigslist results aren’t integrated in the page, they just have a simple interface for performing your search in Craigslist.
The idea here is a good one and it is useful if I really roll up my sleeves and refine my searches a number of times or if I’m searching for an item that’s relatively straightforward. Trying to fashion a catalogue out of search results is a tough thing to do though. Keeping the interface simple relative to some complicated data is a real challenge as well, I imagine. Overall I like the idea but I think the implementation leaves a fair bit to be desired.
I believe you meant zillow.com.
Thanks Dave. You win the prize.
Site is not very usable, tried to search electronics and it keeps bringing me back to the main page.
I’ve been using mpire for over a year as their core business was/is an eBay seller tool.
khang toh: Same here… I even used one of their suggested items and just got the main page over and over again…
My name is Dave and I’m one of the co-founders at Mpire. Marshall thanks for the write up and everyone else for their comments. Couple of things:
1. khang.toh/Jimmy, We updated one of the servers this morning and it caused a hiccup. We knocked out the problem. Should be working as advertised now. Sorry about that.
2. Marshall, agreed that trying to fashion a catalogue out of search results is a challenge. You have to admit though, our graphics guy rocks! Anyway, one of things we’ll be doing immediately is putting exclude words and other settings on the search strings within the pricing guides. That simple act goes a long way towards getting the results you want; keeping things apples vs. apples. As we talked about, we’re also spending a lot of time working on ways to help the user refine their search, so that they get exactly what they’re looking for, as quickly as they can. And once that search is nailed, the historical data we provide can be very powerful in helping figure how much you should pay. Its early days in our efforts and stay tuned for improvements in this area.
i’m getting a ‘503 service unavailable’ error…
But what about Qunu?
How is mpire getting the data? Do they have the rights to use the data from the above names web sites? Are they “screen scraping,” which is usually against the TOS of major web sites?
Dave here, one of the co-founders at Mpire. There are two parts to answering your question:
1. Initially, the core ingredient in our historical pricing engine is eBay Marketplace Data. We literally parse through millions of rows of historical data about what has bought and sold on eBay, and who is purchasing it to help you understand how much to pay. We get the raw data directly from eBay. We have some pretty smart folks who work with the data to make it digestible.
2. The marketplaces like what we’re doing. We’re doing a combination of API level access for some marketplaces and what we call “Flash Crawling” for the others. Flash crawling is a little different that your traditional crawling, but necessary for providing real-time display of items that have an end time. When a user initiates a search through Mpire, we pass the request through to the appropriate site and crawl the return results in real-time.
This has a few big benefits: 1) we get real-time feedback which is critical for products that are time sensitive 2) we don’t have to write crawlers that pound our partner sites walking through their data and utilizing resources; they’re serving us results just as they would a user. It’s consistent with their terms of service and they’re supportive of our approach.
Dave
Hi,
I was just using another site that does similar things. http://corp.andale.com/research for ebay data. They claim, and it seems to be correct that they can auto filter everything out. So the concern you had is auto taken care of. If you search for a camera… they remove all bags and it gives awesome prices.
Ridhima