What does your Marketing Department Think?

Traffic Stats work for capacity planning but, what about marketing?

[cyril kearney]

Most tracking software satisfies our needs for capacity planning. They tell us how many visitors saw how many pages.

Advertising is a big portion of an eCommerce site's budget. Advertising expenditures need to be evaluated on the basis of online sales they generate.

Does this mean that eCommerce sites have additonal requirements? If yes, is your current tracking software satisfying these requirements?

Regards,
Cyril

[Macguru]

Yes ecommerce sites have specific requirements. Especially in highly competitive markets. Since a badly positionned ecommerce site wont survive for very long in this jungle, the additional requirement of some ecommerce survivor is to offer more for less.

[ritualcoffee]

oh how timely your post is. I just had to evaluate a whole mess of packages because of our need for more metrics. Percise metrics, precise specific URL metrics...

it was a nightmare. glad the search is over.

[ggrot]

Hehe, if you really wanted to use stats for usability studies and the such, you could do things such as a javascript log of scrolling, mouse movements, etc passed in through a rapidly loaded image on the body unload function. That would give some really neat data, but viewing all that would be nightmarish (is that a word?).

[cyril kearney]

ggrot, let me explain the nightmare from the point of view of the marketing department.

Suppose they bought an advertisement in a newsltter that was being emailed to 200,000 people. Let's say this cost $ 70 a thousnand. or $ 14,000.

Now the media buyer has to determine if that was a good expenditure or not.

Let's look at it in real world terms. The emails went out Sunday night and on Thursday afternoon a decision has to be made on whether to place the advertisement again and pay the $ 14,000 for a second wave.

Could your marketing department get the tracking data they needed to answer this question?

Cyril

[john316]

If you are doing email campaigns, it would seem that you simplify the whole metric issue by sending the recipients a URL that is only specific to the mailing. Very similar to snail mailings where the reply envelope has something like "room 224" on the return address for the reply card/order form to identify the respondents of the mailing and which list they came from.

If you are just sending them to the main page of a site that otherwise generates traffic, you are looking at a lot more work. Most surfers are only going to 3 or 4 pages deep on any site so you could actually build a self contained "mini-site" to track them very easily.

I'm not so sure that the solution is in the tracking software or in the planning of the response and where you send them.

Sunday to Thursday? I can't imagine why you would need that much time.

[cyril kearney]

john316,

That gateway page idea works for some sites. It only solves part of the problem. Once the visitor leaves the gateway page his association with the email is generally lost.

The media buyer knows the number of people that responded to the offer. 200,000 emails were sent and about 16,000 clicked to the page. How many of the 16,000 bought anything?

Now most activity on emails ends after three days. A few response trickle in later than that. If 100 visiitors bought it might be wrong to continue renting the mailing list. If 2000 bought it might be a great list to buy.

Without being able to see this information, the media buyer must make a decision without all the facts. These decisions are as often wrong as right.

The great dot.com fiasco was caused in part by people buying volume and not ensuring that it was profitable.

They lost a little on every sale and tried to make it up on volume.

Cyril

[ggrot]

My response was off topic, just weird thoughts floating through my head. Anyway, what you do is have your gateway page set a cookie on the users browser, then track the number of cookied purchases that are made. This will work even if they leave and come back to your site by typing it in or using a bookmark.

[john316]

<<The great dot.com fiasco was caused in part by people buying volume and not ensuring that it was profitable.>>

The fiasco was caused by underwriters making huge fees bringing dotcoms public, blame it on the Morgan Stanleys' of the world.

The efficiency of the web is apparent when you begin to look at the alternative; a traditional bare bones 200,000 piece mailing will run you close to $80,000.00 when you calculate postage, printing and list rental. At 14 grand, you can email that same list almost 6 times.

Are you selling one product or directing the visitors to a shopping cart?

[skibum]

An agency (if they use DoubleClick) can generate click commands to track clicks from teh message, and then use spotlight tags to track whatever pages are desired, including an order confirmation page. UnityMail, which we use, enables us to insert trackable URL's to track click throughs, and I believe they are implementing a spotlight tag type system, so that everything can be tracked through one system.

[Bolotomus]

My work with an ecommerce system required several changes to my log analysis procedure.

First, I had to "wrap URLs" so that what were really different requests would get processed as the same request. Specifically I took the session ID out of the URL. That way I know how many times a certain product info page was viewed, not the number of times Joe Blow happened to view it.

Then I wrote some custom processing in Perl to get down to the real e-commerce aspects. For example:

#1. The number of times people see a thumbnail vs. the number of times they ask for more info by clicking on it. I call this "interestingness factor." Things with a good interestingness get clicked on a lot, they must be interesting. (Note, low numbers are good. A number of 1.0 would be ultimate.)

#2. (number of times the info pages was seen) / (# of times the product was actually bought) = the saleability-factor. Again, low numbers are good. I find that even the hottest products have a factor over 100 here.

#3. (number of carts for a given period) / (number of orders for a period) = abandonment-rate. When this is high people are loading up their carts and fleeing the site.

Is there software out there that does this automatically? Not for my system. The way I see it, if you author an ecommerce system, you're not really done until you get into custom log analysis which has to be designed specifically for your system.

Bolot