Predicting Integration Strategy Effectiveness

As consumers evolve and multichannel agnosticism kicks in, identifying which marketing channels are effective for specific campaigns or promotions are tougher than ever. Specially when top executives are more honed in ensuring positive returns for every marketing initiative.

These days, one marketing channel is not enough but it is also futile to select channels you can't sustain. With 'big data' being the most popular topic in this century, more and more ways are being discovered in measuring integrated campaign effectiveness.

Cross-Channel Trinity Predictive Index P(c)

One approach I developed (still pending for further multivariate testing) is to measure effectiveness through a probability of conversion P(c) index. I am calling it the Cross-Channel Trinity Predictive Index. It is a measure of a relative "ad" exposure frequency of consumers and their probability of conversion.

Conversion in this sense is defined as an activated buyer.

I hypothesize that a consumer can only consume X amount of content from X amount of channels per day. I continue to hypothesize that in any given day, a minimum of 3 core channels are used to get a marketing message across that has high probability of conversion.

The more exposed a consumer in these 3 core channels, the higher the correlated conversion rate.  In essence, Relative Ad Exposure Frequency is directly proportional to probability of conversion.


So how do we define Relative Frequency?

Since we are counting exposure of a consumer, we define relative frequency as the total unique consumer who saw your ad over the total Ad impression seen for a specific channel.

For example:

In a digital marketing campaign, a consumer used Facebook as a channel to promote a campaign. It has a corresponding link with tracking codes to identify total unique visitors coming to the final online sales page. Total unique visitors recorded is 100 and Facebook recorded a total ad impression of 1000.

Relative Frequency = 100 / 1000 = .10

So how do we 'predict' cross-channel conversion effectiveness?

The principle behind this approach is based on the random sample spaces and events. That we can isolate a subset of the sample space of a random experiment and describe the new events from a combination of existing events. That by knowing the intersection of two events, we will be able to deduce the possible outcomes that are contained in both events. In this case, the event is the exposure of a single consumer to two channels at the same time.


Going back to my initial hypothesis, in any given day, a consumer is exposed to a minimum of 3 channels. For maximum retention of a message, optimizing these 3 channels will give the highest probability of conversion.

Graphically, the consumer in discussion is intersected in the middle of these 3 channels.

Cross-Channel Trinity Predictive Index P(c)

In order to "predict" the conversion rate among these channels, we apply the concept of Event Intersection using Relative Frequency as the unit of measurement. And based on this statistical concept, computing for the intersection of two events is mainly multiplying the probability per intersection.

Simply said (in the venn diagram shown above):

P(c) = P(RFM1∩RFM2) x P(RFM2∩RFM3) x P(RFM3∩RFM1)

Let's put it into an application:

A marketer decided to use the following 3 channels based on historical data to promote an upcoming Sale. The corresponding count of unique consumers and the ad impression of each channel is displayed below. (assuming the three channels were measured with the same number of days).

CHANNEL                     UNIQUE VISITORS     TOTAL IMPRESSION     RF
1. Facebook                         500                                         1000                        .50

2. Direct Mail Coupon
Reservation via Phone          150                                        200                          .75

3. Youtube Video Ad           950                                        3500                        .27

The Predictive Index for these 3 channels is:

P(c) = P(facebook) x P(Direct Mail) x P(Youtube video ad)
P(c) = P(.5)x P(.75) x P(.27)
P(c) = .10 or 10%

With this volume of exposure, the probability of converting all the consumers who have seen the ad X amount of times is 10%. In this case:

10% of total Unique Visitors = (500 + 150 + 950)10% = 160 converted customers.

The next question is, with this model, is it worth investing in these channels once you've identified your potential return.

Lastly, as you can see, the relative frequency is your key driver. The higher the relative frequency, the higher is the conversion rate.

Caveat Emptor 

This model is still on its testing phase. I've only used this in a minimum sample size in three different countries. Although the results are quite encouraging, this model is still pending for more testing.

I would love people to test this out in their own respective companies and please feel free to message me if you applied it and got the results close to your predicted conversion rate.