Tuesday, November 08, 2011

Multi Channel Attribution

So, in this blog I try to make sure I only say things that Adobe cant do me for. In many cases its about Intellectual Property (i.e. things I've created whilst under their tenure).

Fortunately, most of what I come across is something that has been addressed elsewhere so whilst I may worry about a legal suit from SPSS their focus is elsewhere. Talking about Multi Channel attribution does not leave me worried about the legal giants at Adobe.

So here's the lowdown.

1) Have you got a multi-channel (online) problem. e.g. does PPC clash with SEO and affliates. Find out and size it. You really may not have a problem at all. No, really. Do this before you buy something you regret.

2) Be all means try a bit of time series. Have a look at consumption of channels over time and see if that correlates with anything that happens on your website. From that think of something that you can actively change as a result. Good luck with that.

3) Dont even think about some crazy weighting system that allots some figure based on time or the number of interactions between it and something else like a purchase. Its really naff. BTW how are you going to pick your weights?

4) Very simply, work out for a given time frame, what is the likelihood of any channel/campaign whatever to lead to a conversion if its the only thing a customer sees/clicks on. You then look at how this compares to the figures when an customer clicks on another campaign/channel etc. Now you've got to spot/calculate whether any stand out, i.e. the likelihood to convert increases if certain things occur together rather than if they were alone.

And that aint Adobe IP. We did that at SPSS across "true" multichannel.

Oh, and once you've done all that and found a relationship, think about what that actually means in terms of "action". What can you actually influence in terms acquisition for customers that you know something about? Some things are possible but not as easy as you think...

Thursday, January 13, 2011

When people pay you to tell them they are wrong

Being a consultant can be tricky, particularly when you work in areas like analytics, rather than in a strict technical capacity.

Customers want you to come in and tell them how to improve their business, how to make those incremental uplifts in their business to make them look good to the boss.

What they don't like is when you point out a glaring issue that has been overlooked or perhaps misunderstood.

I'm talking about those people who believe high correlations between key KPIs mean they should break out the champers. (people should look at how things are correlated and what the connection means)

I'm talking about online marketers doing a great job spending loads of money on online marketing only to find 98% of purchases are from customers dropping directly on the site or who perhaps are existing customers that would have come to the site regardless of the "lookup" in the google phonebook.

I'm talking about the person who hasn't noticed the big errors firing all over the site when a customer tries to make a purchase.

I'm talking about the designer who seems to actively want you to start the purchase process every single time the customer enters something wrong in the application form.

I'm talking about the people who for the past 10 years have read the wrong number on a chart.

I get paid to explain where they might be going wrong.

I'm not always asked back.

Footer

Add to Technorati Favorites