Upon reading Kevin Hillstrom’s post about Proven And Influenced Advertising Channels today, it touched on  a post of mine on multi-channel attribution  3 years ago, but with a more elegant solution (Measuring Multi Channel Marketing – an open question to the Web Metrics Community).

According to Hillstrom, there are two types of advertising channels in the Multichannel Forensics framework.

1. Proven Channels
2. Influenced Channels

Kevin Hillstrom establishes a typical senario, much like I did back in 2006:

Let’s say that you have three advertising channels that resulted in an order:

1. Catalog received on June 1.
2. E-Mail received on June 3.
3. Order on June 5, using Pay-Per-Click.

One could label pay-per-click as the “proven” source. One could also label catalog and e-mail marketing as “influenced” sources — in theory, these advertising channels influenced the order.

By determining a source is “influenced” instead of proven, Kevin Hillstrom puts forward an interesting idea

When catalogs and/or e-mails move from “proven” to “influenced” advertising sources, you don’t have to send as many of these to the customers using them as “influenced sources” in the future — once customers move away from these advertising channels as their primary reason for purchasing, they become less likely to use them in the future.

He doesn’t explain why that’s so, but let me guess, and fill in some applications of how this might work for Influencer Attribution.

Here goes – as your customers evolve from learning about your brand, and your offers - they may move to favor one marketing channel over another, much in the same way that a customer segment evolves from one set of characteristics and needs, to another (measured in Web Analytics as explained to me by Matt Belkin at the first XChange conference, 2 years ago, in Napa Valley).

As we now have much more information and choices than we had even 20 years ago, Direct Marketing, via, say, a catalog or email campaign, might change it’s effectiveness as a customer or receipient digests information and makes choices (evolves).   Static models of Multi Channel Marketing don’t account for people changing and evolving from one segment to another.

What Kevin Hillstrom found, is the effect, I think, of that – if person got used to receiving mail order catalogs – but then ended up using the Web to puchase – the catalog, according to Hillstrom, becauses an Influencer – no longer the primary way the customer purchases, and loses much of it’s importance.   Customer has learnt to connect from another channel and doesn’t need the same stimuli as much.   That much makes sense to me.

The question then becomes – for me – can I chart Influencer Attribution on a continium and is there some example of this that would fit in with the model Hillstrom put forth?  Yes, Rapid Segmentation – which he explains in an earlier post

… Rapid segmentation will be repeated after each subsequent purchase. Customers will move back and forth between segments — usually moving from traditional direct marketing to social shopping, but occasionally going the other way, that’s what makes this fun.

The Hillstrom Influencer Attritubtion model takes care of the Multi Channel question I had 3 years ago – but it also reminds me of something else, one of Paul Krugman‘s blog posts on how Deficits Saved the World:

” …. That’s an interesting way to think about what has happened — and it also suggests a startling conclusion: namely, government deficits, mainly the result of automatic stabilizers rather than discretionary policy, are the only thing that has saved us from a second Great Depression.

The following figure makes the argument:

Here I show the private sector surplus and the public sector deficit, both as functions of GDP; the private sector line is upward-sloping because higher GDP means higher income and more savings, the public-sector line is downward-sloping because higher GDP means higher revenues. In equilibrium the private surplus equals the government deficit (not strictly true for any one country if you add in international capital flows, but think of this as a picture for the world economy). To make the figure cleaner I’ve shown an initial position of balance in both sectors, but this isn’t important.”

Paul Krugman’s chart shows a progression, which is why it stuck in my mine – even if the subject of his post doesn’t appear to relate Influencer Attribution, I think it does relate.   One of the reasons, what Hillstrom has stumbled onto – is that what worked yesterday, might not work tomarrow – that principal is a constant because people and situations change.

In fact, that’s one reason why, last fall, when the FED tried to rescue the economy, which was in free fall, it looked back, as Obama did, to the Great Depression as a model of what to avoid this time, except the situation has changed – people change – they evolve.   Our models have to take in account that people will evolve, more at their own rate, but they may move across segments.

According to this model, it may be possible to predict what is the “Proven” primary channel now – and what a customer will evolve into, over time – and a diagram like that will probably end up looking similar to Paul Krugman’s chart, above.

I’ll leave it to Hillstrom, and anyone else, to fill in the gaps here – or add their own insight, to mine.