A Concise Overview Of Cross-Channel Marketing Models

For several decades, modeling cross-channel marketing impact was synonymous with marketing mix (“top-down”) models. However, with the increase in cross-channel tracking and computational power, new breeds of marketing attribution (“bottom-up”) and other alternative models have appeared in recent years.

analytics-600

(Stock image via Shutterstock.com. Used under license.)

Although these models all seek to quantify the relative impact of different media channels, they are quite different in their approach.

And while there are some articles comparing these different methods, I have yet to find a free unbiased resource that gives an overview of the subject; so, this is my attempt to fill the gap.

This overview intentionally avoids evaluation of the approaches listed. Each method has its own strengths and weaknesses, and the choice of which to use will depend on the particular need of the organization.

Common Approaches to Modeling Marketing Impact

The most frequently applied methods for modeling cross-channel marketing impact today are the following:

  • Econometric top-down approach (e.g., marketing mix models)
  • Algorithmic bottom-up approach (e.g., attribution models)
  • Machine learning approach (e.g., agent-based models)

Below, I will give a high-level summary of what each approach involves and their respective strengths and weaknesses.

mobile-chart-growth-analytics-featuredEconometric Approach

The classical econometric approach to marketing mix has the longest history of practice and is widely applied today. Commonly used for informing high-level budget allocation, this approach typically employs some form of distributed lag regression to estimate the impact of marketing activity on total sales.

Traditionally performed as individual consulting projects, some vendors have been working to automate and productize this method.

Outline of approach:

  1. Compile historical spend and marketing activity data across media channels, along with conversion/sales data, and exogenous (non-media) factors such as economy, pricing changes, etc.
  2. Build a statistical model to assess the effect of these variables on sales or other KPI
  3. Quantify the impact of each marketing channel or campaign

Strength: Data-driven statistical approach allows for high predictive capability.

Weakness: Granularity of insights and what-if scenarios are conditional on granularity of input data.

shutterstock_120614257-cookies

Algorithmic Approach

cookiescookiesThe touchpoint based approach to marketing attribution was made possible only recently via cross-channel tracking. These models are unique in that it looks at cookie-level event data, and attempts to determine the impact of each type of touchpoint from sequence patterns and results.

Over the past several years, a number of vendors have offered up marketing attribution solutions, either the bottom-up approach alone or in conjunction with a top-down approach.

Outline of approach:

  1. Track cookie-level data on marketing touchpoint sequences along with conversion activity
  2. Compare conversion rate for a sequence to similar sequences in order to assess effect of a particular touchpoint being present
  3. Aggregate results across sequences to assign relative value to media or campaign types

Strength: Touchpoint-level calculation allows for very granular attribution and impact assessment for tracked media.

Weakness: More descriptive than predictive; cannot by itself incorporate offline or non-tracked online media.

Note: I will not cover rules-based allocations such as even distribution or U-shaped here. Here are some reads on the rules-based vs. algorithmic attribution.

shutterstock_179490242-machine-learningMachine Learning Approach

Agent-based models treat individuals as “agents” that interact with events or stimuli to exhibit certain behaviors, and utilizes computational simulations to assess impact.

The concept of agent-based modeling first appeared in the 70’s, gaining adoption in social sciences in the 90’s. Application toward the field of marketing has been delayed due to the computational complexity involved, but recent years have seen attempts to apply this approach to marketing impact modeling.

Outline of approach:

  1. Create a simulated population for the business using demographic data and understanding of customer profile
  2. Train and calibrate the simulated population using historical marketing activity, along with conversion/sales data and exogenous factors
  3. Simulate reaction to various marketing activities and changes

Strength: Individual-based models allow for flexible simulations at customer demographic level.

Weakness: Accuracy of results is only as valid as the behavioral assumptions for the business and demographic segment.

Future of Marketing Impact Assessment

As marketing enters an era of accountability, the demand for cross-channel marketing impact assessment will continue to increase. The next decade will see further refinement and combination of these approaches to modeling marketing impact. We already see this trend in the market, with some vendors combining top-down and bottom-up approaches.

In reality, however, there is no such thing as a perfect model. It is impossible to account for all the variables that may impact business or marketing performance, and some – such as a viral marketing success or stock market crash – are virtually impossible to predict.

After baseline model accuracy and complexity reaches a plateau, these unpredictable or unaccounted-for factors will play a large role in determining how accurate and useful the models are in practice. As a result, the key differentiator is likely to become real-time simulation and prescriptive capabilities:

  1. Support of predictive what-if scenarios at granular cross-sections (e.g., brands, products, markets, segments) and levels (e.g., campaign, publisher, target, creative)
  2. Prescriptive functionality and workflow that allows marketers to take immediate action based on the most recent trends and campaign results

Images via Shutterstock.com, used by permission.

Opinions expressed in the article are those of the guest author and not necessarily Marketing Land.

Related Topics: Analytics | Analytics & Marketing Column | Channel: Analytics

Sponsored


About The Author: leads product marketing at Origami Logic, a cross-channel marketing intelligence solution for modern marketers. With a career of 8 years in marketing and analytics spanning various functions, Kohki's focus has always been on translating data into strategy, simplifying the complex, and bridging the gap between data and organizational silos.



Sign Up To Get This Newsletter Via Email:  


Share

Other ways to share:

Read before commenting! We welcome constructive comments and allow any that meet our common sense criteria. This means being respectful and polite to others. It means providing helpful information that contributes to a story or discussion. It means leaving links only that substantially add further to a discussion. Comments using foul language, being disrespectful to others or otherwise violating what we believe are common sense standards of discussion will be deleted. You can read more about our comments policy here.
  • http://cashwithatrueconscience.com/rbblog Ryan Biddulph

    I dig the idea of qualifying each marketing channel. Serious analysis with a bottom line approach helps you see what’s working and what’s holding you back, channel wise. Then you can rock it out as you trim the channel fat to focus on the prospering strategies.

    I found this post on Kingged.com.

  • Nico N

    With all respect, this comparison is pretty flawed and confuses basic principles of statistics. First of all, I understand the distinction between top-down and bottom-up models, but your choice of “machine learning” for the third group is rather unfortunate. Machine learning is a method and can be used for all kind of models, including the first two mentioned approaches (see random forests). So it’s not a separate group (because that would be like saying a BMW is not a car). BTW: A more traditional way to cluster the first two is to differentiate between a) aggregate data vs. b) individual-level data models.
    Second, please note that both aggregate-data and individual-level models can be descriptive, explanatory, or predictive. This completely depends on the actual type of model, estimation technique, and set up one eventually uses (compare a VAR* with a structural VAR for time-series models). There are also very different advantages between individual-level models and aggregate-data models. E.g., the former are less prone to some crucial inference biases, while the latter typically can distinguish between short- and long-term effects (at least in case of time-series models, which are the typical applications for marketing-mix modelling). Again, depending on model set up.
    *VAR=Vector Autoregression

  • Kohki Yamaguchi

    Hi Nico, to your points on what’s theoretically possible using different methodologies, I agree wholeheartedly. But at the same time, this article is meant to be accessible to a wider audience, so any sort of simplification results in loss of information. :-)

  • Nico N

    Kohki, thanks for your reply. I am completely with you regarding simplification to make the boring and complex world of statistics more appealing to a wider audience. However, the classification suggested above is simply wrong and I am particularly concerned that this makes the rounds (for example, Datalicious already reposted this classification on their blog in Australia). I already see our clients coming to us asking for an “attribution” model because they do not want an “econometric” model.
    Since Marketingland.com has such a large audience, it is particularly important not to mix up basic concepts (in terms of education). To make it really simple, we could say that on the one hand there are
    (1) top-down models (aggregate-data) and
    (2) bottom-up models (individual-level data).
    On the other hand,there are models that use
    a) real measured market data (survey / purchase data) or
    b) simulated market data
    to provide insights into marketing-channel effectiveness. Of course, the different forms can be mixed and combined.

 

Get Our News, Everywhere!

Daily Email:

Follow Marketing Land on Twitter @marketingland Like Marketing Land on Facebook Follow Marketing Land on Google+ Subscribe to Our Feed! Join our LinkedIn Group Check out our Tumblr! See us on Pinterest

 
 

Click to watch SMX conference video

Join us at one of our SMX or MarTech events:

United States

Europe

Australia & China

Learn more about: SMX | MarTech


Free Daily Marketing News!

Marketing Day is a once-per-day newsletter update - sign up below and get the news delivered to you!