Forum Moderators: buckworks
But I think I should be at least as interested in tracking keyphrases to conversions at the product level. I want to know product sales by keyphrase, don't I? . In other words, isn't knowing which keywords are typed by a customer who is destined to buy widget 'X' ultimately more actionable than knowing key-word-combo 'Y' has a ROI of 3:1?
I'm brand new to the eCommerce game and I'd love to hear your comments.
I will probably get shot down for saying that as a lot of people advise against it, becuase it's letting the PPC engine know what keywords are converting and which one aren't.
Look at it this way at lease you'll have the previous data stored and recorded if you use their systems whilst you're experimenting and finding the one that works best for you.
I'm still wondering though, whether others find matching the products sold to the PPC keywords to be a compelling bit of information. Maybe it's so obvious that it doesn't need a solution? Or maybe it's just the income from the sale that counts?
We sell products that are highly suitable for a number of different, though related, market segments.
We have 50 or so products that we manufacture and sell on our website that are suitable for a number of different, but highly related, market segments. (e.g. snow boarding and snow skiing afficianados would wear many of the same items of apparel, but given 50 products, each group might rank the items very differently in terms of desirability)
Our Google adwords campaigns are very structured with rigorous naming conventions that are all tied to specific CPA numbers. Bids are raised or lowered to meet our goals and at that level we don't really care what particular items are bought and sold.
However, at the website level, we definitely do care. We want to show the customer items that they are most likely going to be interested in. (There are actually some pretty cool companies out there that are trying to get smarter about how to calculate this, by looking at user data from thousands of visitor paths through a site. Last time I looked at them though, they were more "drawingboard" technology than something proven to work. The technology is getting there though.)
So, each night, our site looks over sales data for the recent past and decides what products are going to be shown in what categories and in what order.
Back to the snowboarder/snow skier example, let's assume that we sell snow pants. A site could be structured as:
All Pants - lists all 50 pair of pants in order of recent sales volume
Downhill Ski Pants - lists top 15 pants purchased by people coming from the "downhill ski pants" ad group by recent sales volume.
Snow Boarding Pants - lists top 15 pants purchased by people coming from the "snow boarding pants" ad group by recent sales volume.
Ice Fishing Pants - lists top 15 pants purchased by people coming from the "ice fishing pants" ad group by recent sales volume.
Cross Country Ski Pants - lists top 15 pants purchased by people coming from the "cross country ski pants" ad group by recent sales volume.
The part of the system that builds out the categories is automated. (Its also nice because what sells best for us is greatly affected by time of year. Everything automatically adjusts.)
So, our system automatically adjusts itself regularly based on the buying patterns of our customers at the keyword level.
They had hundreds of ppc keywords and were spending 50k a month.
They gave me access to their ppc accounts and their site conversion data.
It took 5 hours of my time to find out that there was one group of keywords that produced profits and another group that was a loss.
They sold a main product line and an accessory.
When prospects came in on the main product keyword, they often purchased the main product plus the accessory.
When prospects came in on the accessory keyword, they were less likely to purchase the main product.
My suggestion was to kill all the accessory keywords.
They added 10k to their bottom line that month.
Firstly it is all about finding what groups do well, individual keywords are important but secondary.
Please note: I NO longer do ppc analyzation.
