Evan from Auctionitis sent me the following email in response to my Webwalk column about Trade Me's lack of an API. This discussion has been going on at www.rowansimpson.com. Rowan has just left Trade Me to join on a part time basis, accounting software start-up Xero.
It won't come as any surprise at all that I would fall into the camp of those that think an API would be a useful thing for TradeMe to provide. Some thoughts that I feel are relevent to the idea follow.
The point a number of people made in various guises about providing an API for your sellers is important. Providing an API so sellers can use a tool, or tie their own software to Trademe allows them to reap the efficiency of a user interface optimised towards selling and how they want to do things and away from the standard browser interface which while ubiquitous is a terrible data entry tool. It also reduces the barrier to entry that the current labour intensive model raises for existing businesses that want to use TradeMe as a genuine alternative to their existing sales channels. An API binds these people and businesses more tightly to TradeMe.
As to the dynamic nature of the changes on TradeMe, my assessment would be that the majority of these changes are at the presentation layer rather than the data layer, although I'm open to correction on this point. Even the most basic API would be insulated from these changes in the majority of cases, as the API would bypass the presentation layer. This would probably have the neat side-effect of reducing bandwith thereby indirectly benfitting the other users,most especially the buyers - the real TradeMe audience. When data changes occur it is a simple matter of communication.
If you control the API, you also have the option to impose standards and conditions on it's use - you gain more control not less. eBay allow endorsement of third party products after a QA review - what better way to ensure the quality of what third parties do and therefore the experience of your users ?
There are also unexpected benefits that can accrue from the use of an API might, for example, in the case of Auctionitis the number of pictures stored on the Trademe servers was reduced by (in some cases) a factor of 10. In a couple of cases sellers storing 4,000 or 5,000 pictures on the TradeMe servers reduced that number to 400 or 500. This came about because we provided a mechanism to assist sellers in not having to load the same picture over and over again each time an auction was loaded. They saved bandwidth, they loaded more product, TradeMe reduced storage. It's likely that the reduction was minimal in the scheme of things, but it demonstrates what might happen.
It's pretty reasonable to look for a return on the effort, but the most obvious return would be the number of listings that established businesses could/might throw Trademe's way - especially businesses that currently DON'T sell on TradeMe. At the moment, third party tool makers are largely confined to trying to capture existing TradeMe sellers who understand the effort involved in listing items and are looking for a more efficient system. They largely accept the risk of using an unofficial interface because they are able to quantify it against the effort and expense of entering data through a browser.
Imagine if those third parties were actively recruiting businesses to sell on TradeMe - effectively a free saes force extolling the benefits of TradeMe.
Most of the technical questions are easily answered or solved; I think it's more a case of TradeMe seeing the benefits that could accrue for TradeMe and going after them.