spikey - 12:54 am on Sep 25, 2006 (gmt 0)

When you say conversion was higher, do you mean on that individual product or sales overall?

Sales overall. The page being tested was the first page of the cart. I showed related products to 50% of my users and not to the other 50%. The 50% I didn't show the related products to where more likley to complete their order (at a statistically significant level).

I can see in theory how it could increase conversion, but our system did it in an automated way. It was basically "customers who bought this product also bought" (i.e. we would show products that had historically shown up the most frequently in carts that had the same products as the current cart).

I think to get it to work correctly (at least in my domain) you'd need a more complex selection algorithm and I also think you need to explain to users why you're recommending the product. I think one of the biggest failings I see in ecomm is the idea that you can put up products under a heading like "you may also be interested in" and people will buy them.

It's easy enough (through online marketing) to sell a product to someone who is looking to buy it. Because of that, I think we seriously under-estimate how hard it is to sell a product that the person is not looking for. I'm not saying it's impossible, but it's a whole other level of difficulty in marketing. If anyone's had a different experience I'd love to hear about it.

We resell products for a variety of vendors with known brand names. We often have people (either other companies or internal people) saying "here's a great new brand, let's sell a bunch of it too". I then go into my speach about the difference between riding on the coattails of existing brand recognition and building a brand name. How it's not just an order of magnitude (e.g. sell 10% of the big boys), how marketing sucess won't be about ROI but about the cost of customer acquisition (over years) and how much it costs to get enough critical mass to have a brand identity. Again, I'm not saying it can't be done, I just think the marketing costs and success metrics are in two different worlds. Maybe it's just because I do really well with the former and not so well with the later.

Again, I'd love to hear other's experience with this, especially if you've found differenlty. i.e. when you're about to get a customer to buy something how do you sell them a little more?