Conversion Rates

I don't write about conversion rates too often because often this data isn't available to me from all of my providers. Shopping.com updated their reporting tool to give me conversion rates so I can moniter changes we are making on the site (and indeed my whole collection of comparison shopping sites such as Best Buys Zone. I hit the shopping API between 50,000 and 150,000 times a day. I also cache the results (cache is updated when prices are updated at shopping.com) so this represent roughly 1/3 of my actual traffic. Naturally December provides the highest conversion rates (my conversion rates are well above the contracted target with Shopping.com in any case and I consider that information a trade secret). I am going to mention one day in particular when I saw something so rare it simply must be mentioned. My traffic peaked on 12/11, 12/12, and 12/13 - 12/11 was the highest. This might isn't surprising since 12/13 is roughly the limit people feel comfortable ordering for delivery before Christmas. That's been the case for the last 6 years I have been running shopping comparison sites.

The day I am talking about is different day. This day was 12/21 - not a huge traffic day for me (54,580 queries) but one feature stands out - on that day my conversion was 100%!! On that day every single lead I generated converted to a sale. This is very rare and indeed it's the first time I have seen it in 6 years. I have had other days that were high (40% or so) but this is the first time I have seen a 100% conversion. I think this is outstanding for several reason. Firstly it's because my volume of traffic to the shopping.com api is pretty high. Achieving a 100% conversion rate on 10 customers is pretty doable - doing so when the numbers are higher is much much harder.

Busy working

I have been busy working on improvements to the system and new features. This latest feature is a Flash based version of TagOuts. This now allows you to feature your products and embed them in a myspace profile or a blog. Below is a my PS3 feed in a flash based feed. I need to work on usability a little bit but it works pretty good I think. Naturally clicking on the product takes you to the page with the product.

PS3

Hosting Slowdown

As often happens with new projects of mine I have already outgrown the friggin' budgeted resources on my cluster. This means a new server. Instead of buying one and installing it in the Los Angeles data center I plan on leasing one from Hostway. I really don't want to go through the hassle of shipping/installing linux/installing load balancing/hardening the kernel and then spending a few days installing it on the cluster. UGH. Quick solution here is to get one big enough and revisit it in one year. I am on the phone with the hosting provider to see if there are any other issues that might be gumming up the works (maybe they are undergoing a distributed DOS).

Are people really comparing products via shopping comparison?

One of the core assumptions that shopping comparison engines have made is that people are using the engine to actually compare products before making making a purchase decision. It's one of the reason that Yahoo shopping API home page allows you to reconstruct the compare grid for comparing the various products. The V2 of their API passes the GridImage etc if so requested. (Indeed the shopping grid can be completely recreated, allowing you to compare prices and specifications). Yet a thought a occurred to me - these sorts of comparison grids have been around for a while now since they seem to have been a design decision from early on. Yet I am an avid user of comparison engines such as Shopping.com, Pricewatch and of course my own engine (Early Miser, check out my page at earlymiser.com But I have almost never used the grid features of the various shopping comparison engines and I think no one does except for a few hard core geeks. The reasons for this are multiple but I would like to go into each of them and see if we cannot improve on the shopping experience.

First off the grid approach is designed to compare various technical specifications. Does this unit have more memory? Does this one have a larger hard drive? Yet my own log files indicate that people typically are searching for a SPECIFIC product. I am not alone in this fact. Furthermore the top terms at Yahoo Shopping have a quite a few specific products in them (you can look at them Buzzlog for Shopping.

The other thing that the buzz list at Yahoo has is the keyword "reviews" attached with a product name which I think is fairly revealing how people search. People are making buying decisions about uses - not the specific information that is presented in the product specifications.

They have a use in mind for a product and want to see a review. Reviews never cover the technical specification but how that product performs in real world situations. An ideal review would cover their exact usage and what the review thinks about the use in that specific context. Take a look at the terms in the Yahoo Shopping Buzz

1. Digital Camera Reviews
2. LCD TV Reviews
3. Cell Phone Reviews
4. PS3 Reviews
5. Camcorder Reviews
6. Wii Reviews
7. Plasma TV Reviews
8. GPS Reviews
9. HDTV Reviews
10. Laptop Reviews
11. Zune Review
12. MP3 Player Reviews
13. Kodak C633 Review
14. Samsung D53 Review
15. Olevia 232V Review
16. Xbox 360 Reviews
17. Nokia N73 Reviews
18. Samsung Blackjack Reviews
19. Canon Digital Camera Reviews
20. Mio C310x Review

Notice what they are not searching for - product specifications. Yet almost every shopping comparison engine builds comparison grids for specifications. How can we take user behaviour and improve upon the shopping comparison experience?

By adding peer to peer mediated reviews to our site. Find someone that owns the product in the site or in your extended network (with our upcoming Own Indicator) and then request a specific review from the owner! The question and the review then become part of the editorial content. Over time significant knowledge about the specific uses in a specific situation will be developed. Furthermore this will represent a natural avenue for growing the social network. Most review models are top down or many to one. This enables a many to many model for reviews.

Yahoo Returns!

After an extended 26 hour outage - Yahoo Shopping's web api is returning results again!. Kalu Kala!

Yahoo - Redux

Apparently the Yahoo Shopping API has fallen down and it can get back up. Hopefully this will be fixed tomorrow. On the other hand I kinda doubt it. Funny thing is I originally approached Yahoo Shopping about developing EarlyMiser.com EXCLUSIVELY on their platform as I thought they might like a success story in the social commerce space. They were apparently to busy to take my request and in the end it turned out to be a good thing for me. Their API (I have hit it about 2,000 times today). I will try emailing the list and contacting them again. I am also trying to reach them via their paid syndication program.