Over the past few years, I have tried to challenge my thinking on how to best represent how a target audience responds to a marketing stimulus.
At a high level, if you investment money as a company in marketing it is because you believe that it will eventually drive sales for your organization that will deliver and sustain an ongoing level of profit margin.
From a distance, this is fairly easy to understand. However, if you want to understand the underlying dynamics of delivering profit margin through an integrated set of marketing channels, it is a different ball game.
Most phenomenon in nature are not linear, they actually are more likely to be quadratic with a significant level of statistical noise that makes any modeling attempt quite tricky. Nonetheless, human nature likes to believe that business can fairly be represented in a linear way.
So, to keep it simple, if you keep the range of your marketing investment within the bounds of reasonable, there is a way to linearize marketing.
Let's start with a simple example. Let's assume that you invest $x in a direct marketing channel like paid search (Google, Bing or other) to promote a new book called "The Marketing Matrix". You'd set up your campaign and would use your $x on a cost-per-click basis when consumers click on your ad displayed in the search engines:
From this investment, you would deliver $y in sales (hopefully $y > $x at least). A high level way to determine the effectiveness of your investement is to look at your return on marketing investment a.k.a. ROMI=y/x
This linear metric represents what you supposedly got from paid search conversions for having invested in paid search. In this case, I am going to call the ROMI of paid search via a paid search investment r(ps,ps).
Real marketing life is more complicated than that though. Most of the time, people naturally tend to look at their marketing investment in isolation, believing that each investement in a given channel returns revenue from this same channel only.
This is not true.
As always, the art of business blurs the lines of channel isolation and value overlaps among all channels...
Let's assume that you are investing $Xtv in a TV campaign to drive sales through your call center and your online assets and that your have $Xps to run your paid search campaigns at the same time (to promote the same product/service).
If we stuck to our one-point linear vision of marketing, $Yps representing the revenue generated by the PS campaign, we'd be inclined to believe $Yps = r(ps,ps)*$Xps
This wouldn't make much sense because the first thing that your target consumers do when seeing your TV commercial is to look up your core keywords from the spot in the search engines, and often times, click on the related paid search ads.
What this means is that a portion of your total paid search budget is specifically going to be used to capture some of the traffic generated online by your TV campaign. Assuming that your TV landing page or microsite or website has had time to be optimized from an search engine optimization standpoint and properly referenced, you will also organically capture some of the online visits generated by TV without paying a dime.
Well... we need to introduce new value drivers here:
- Xps|ps the portion of your total paid search budget (Xps) that is truly used by your target consumer looking for information on the search engine and clicking on your ad
- Xps|tv the portion of your total paid search budget (Xps) that is used because people saw your TV ad first which caused them to go online to look up core keywords and click on your pay-per-click ad
All in all, in this example:
Xps = Xps|ps + Xps|tv
Your apparent total revenue from "paid search" will therefore be:
Yps= r(ps,ps)*Xps|ps + r(ps,tv)*Xps|tv
Reciprocity applies here. Let's assume that the consumer saw and clicked on your paid search ad first, withtout converting, then saw your TV which sold him on your product/service and called your call center to buy. This is much less likely but can occur.
In that case, using the same notations:
Ytv=r(tv,tv)*Xtv|tv + r(tv,ps)*Xtv|ps
Where Xtv=Xtv|tv + Xtv|ps and good luck in real-life to accurately determine Xtv|tv -the portion of your TV budget that directly converts in to sale from TV exposure , and Xtv|ps the portion of your TV budget that closed the deal for your product/service on people who had seen your paid search without converting first. LOL How is that for art as long as we are not able to track what's going on in people's minds?
With this example, if we stick to the 2 channels aforementioned (TV and Paid Search) we see that our vision of marketing is a 2*4 matrix from the R2*R4 vectorial space. If we look at our revenue as a vector from R2 and at our marketing investment as a vectror from R4
Y = RX
Where:
Y= |Ytv
| Yps
X= |Xtv,tv
|Xtv,ps
|Xps,ps
|Xps,tv
and R=|r(tv,tv) r(tv,ps) 0 0 |
| 0 0 r(ps,ps) r(ps,tv)|
This is the easy part, if we throw a 3rd channel in the mix like Direct Mail, it becomes a tad more complicated and it becomes almost impossible to model if you assume that someone can receive your Direct Mail, see your ad on TV that promotes the offer that was on your Direct Mail, and decide to go on the web to click on your paid search ad to finally buy your product/service...
I could have some fun modelling this in N dimensions where N is the number of marketing channels where you are investing your marketing dollars, I am not so sure that it would be of tremendous value. The point is that the marketing matrix is a mysterious one where hops and bounds among channels eventually result into bottom line results.
Therefore, I believe that it is important to feel the matrix "like a splinter in your mind"
and understand that the only way to determine if marketing campaigns are successful is to look at the overall power of integration. The continuous hops and bounds of your target audience among all channels weave a solid fabric that holds the keys to the success of your marketing campaign.
If you don't create this fabric or pokes holes in it by isolating channels, your success will slip through the holes of lower returns.
In my next post, I will value feedback on how we can apply digital signal processing techniques to better understand the marketing value drivers behind noisy mass marketing campaigns such as DRTV. We will also try to determine if it is possible to make media placements calls based on facts rather than subjective impressions on these highly volatile random marketing signals.
Happy turkey to all!
No comments:
Post a Comment