30 May

Notebook Thoughts: Understanding Marketing Attribution

Algorithmic or Probabilistic Attribution uses statistics and machine learning to determine the probability of conversion across marketing touchpoints. In other words, how much of a conversion should be attributed to each channel. In order to keep things simple, I randomly chose a few variables—out of dozens or more—that could go into our model. Let’s go through the rows one at a time.

Touchpoints – This is a mix that includes online as well as offline touchpoints.

Platforms – These are some of the platforms that can be used to collect the data for each touchpoint.

Cost – Cost is one of our most important variables since it helps us determine ROI. Just because something is effective does not meant it is efficient.

Frequency – How many times was our ad/content served to the prospect.

Action – Did the prospect take any action upon viewing our content (i.e. click on it, etc.).

Duration – What is the duration of an engagement? In this example, our prospect spent 6 seconds on landing page after clicking on a mobile ad. Also, we used technology to determine that the prospect looked at an OHH ad for 2 seconds.

Recency – When was the last engagement before conversation. The closer the recency between touch points the higher the weight to the precedent one. Here we see that the prospect conducted a product search within the hour after he/she engaged a social ad/content. Thus, “Social” would get a larger attribution.

Quality score – This a variable that you don’t see often that is extremely important. What is the quality of the ad/content? Was the ad place next to undesirable content? Was the engagement likely from a bot?

Halo – Was there a “halo effect?” That is, did the prospect take a secondary action of value to the brand. In this example, the prospect did a search for a related product and ended up buying that product in addition to the one advertised.

Data Lake / BI – This is the infrastructure needed to process the data and run the machine learning models.

Attribution – Self-explanatory.

28 Mar

Notebook Thoughts: The Marketing Technology Stack

 

  1. DATA – The Data Infrastructure employs a data lake to ensure scalability, speed, and flexibility. Data integration covers all techniques to connect all varieties of structured, semi-structured, and unstructured data.
  2. MODELING – The Modeling layer is the environment to facilitate advanced models and analytics. These include: Marketing Mix Models, Machine learning, AI, Deep Learning, Segmentation, LTV, Attribution modeling, and many more. The key to this layer is to have the analytics environment and tools needed to generate these models (e.g. Python and R).
  3. PROFILING – The profiling layer  is the part of the stack in which we attribute data, metrics and measurable behaviors to build segments and individual profiles (in instances where we collect this data). We can also use our data for matching or adding profiles from 3rd party vendors.
  4. PERSONALIZATION – This Personalization layer is the part of the stack which leverages models and data to deliver content and experiences based on behavior and preference. We use the audience information from our DMP and profile creation to store preferences and deliver against those as well.
23 Feb

Notebook Thoughts: A Data Science Approach to Marketing

  1. Why, how, who and what are we trying to solve for? Are we solving a problem or seizing an opportunity? How does it align to the consumer journey?
  2. Do we have the required data? Are we buying or collecting? Is it structured or unstructured? What about variety, velocity, veracity and volume of data?
  3. What are the appropriate statistical model? Classification, regression, clustering, similarity, etc.? What about marketing models? Segmentation, propensity, etc.?
  4. What is the deliverable and how does it work? Are we delivering through dashboards? Will it need to be real time?
  5. Why should the client move forward with the recommended solution?
  6. What is the investment required?
  7. How are we going to test and optimize our proposed solution?
08 Jul

A Data-Driven Approach to Doing More with Less

In today’s highly competitive landscape, marketers are under pressure to do more with less and make every dollar count. As demand for accountability increases, marketing programs built solely on beautiful creative or humor are unlikely to make it pass the drawing board. CEOs and CFOs are asking marketers to implement programs based on demonstrable ROI models—however, this can only be achieved by leveraging technology and mining data in order to understand and predict purchase behavior.

Contrary to common belief, an ROI driven approach to marketing is no longer a strategy exclusive to large organizations with deep pockets. Increasingly, marketers at smaller companies are overcoming their fear of numbers and technology in order to follow a metrics approach to customer acquisition. This has been made possible by advances in technology and an array of low cost—and sometimes free—data collecting and mining tools.

Making marketing more effective and efficient requires more than simply collecting more data or purchasing data mining software. Having the right mindset and knowing what to do with the data collected are just as important. Marketers should adhere to the following strategies as they shift their marketing approach from one that speaks to a broad market to one that engages, acquires, and retains profitable customers.

Defining success

The cornerstone of a solid ROI-driven marketing strategy is a clearly defined set of goals or measures of success.

• Organizational alignment—Key players in sales, marketing, finance, IT, and HR should be aligned behind a common definition of success and a set of business goals. Goals must be specific, measurable, achievable, realistic, and time-bound (S.M.A.R.T.).

• Strategic alignment—All marketing tactics should support specific strategies and all strategies should drive defined business goals.

• Measurement—Define a starting point by benchmarking business goals. Establish milestones to ensure that strategies are on track, and progress is being made towards achieving the desired outcome.

Closing the loop

It is essential for companies to establish a measurable link between marketing initiatives and financial results.

• Targeting—Marketers should start by combining marketing strategies with data-gathering and analysis techniques to develop highly targeted campaigns based on customer insights and behavior.

• Close the funnel—Campaign results should then be tied to demand generation, sales conversion, and brand experience. This closed-loop funnel system creates a common view of the demand-to-sale-to-experience continuum across marketing, sales, and customers.

• Optimize—Closed-loop marketing allows companies to innovate value propositions and fine-tune marketing initiatives in order to continuously and efficiently acquire and retain valuable customers.

Collecting the data

Before companies start collecting data they should determine what, where, and how products are being sold, and who is buying them. Data gathering can be a daunting task. But costs and effort can be kept to a minimum, if companies follow a systematic and targeted approach.

• Profiling—Start by developing profiles that paint a demographic, geographic, psychographic, and behavioral picture for each customer—and assigns a monetary value to him or her. If individual profiles are not possible, break down your segmentation into as many groups as is feasible.

• Secondary data sources—Secondary data and information can be obtained through government reports, academic research, data brokers, and libraries.

• Primary data sources—Primary data and information can be obtained through interviews, focus groups, surveys, and customer transaction data.

• Become data-driven—From the Web all the way to the cash register, every customer point-of-touch should have a data collection component.

Eliciting a response

Companies should shun marketing strategies that rely solely on mass advertising. Rather, they should embrace a customer- and data-centric marketing approach that delivers the right value proposition, to the right customer, and at the right time.

• Call-to-action—If possible, all marketing initiatives should include a call-to-action that directs customers to purchase, visit a Web site, call a toll-free number, or otherwise interact with a brand’s many points of touch.

• Marketing mix—Call-to-action messaging combined with tracking technologies allows companies to determine which marketing initiatives are most effective in delivering the desired results—and thus, allocate resources accordingly.

Developing actionable insights

Mountains of data alone cannot guarantee business success. Often, those who “own” the data within an organization are either unable or unwilling to share and mine the data. Data collected must be mined and synthesized, in a timely manner, into relevant actionable insights. These insights should be put in the hands of decision-makers across the organization. Information should be grouped into four categories: customer, brand, effectiveness, and efficiency.

• Customer—Data collected should be used to determine which customers present the highest lifetime monetary value. Insights should be developed to identify how to best acquire and retain these customers.

• Brand—It is essential for a brand equity index to include measures that go beyond brand associations. Understanding how employees embody the brand is as important. Also, an effort should be made to quantify the impact of brand assets, such as patents or proprietary processes.

• Efficiency—Data and analytics should be leveraged to help companies stretch their marketing dollars. A key measure of efficiency is cost-per-acquisition (CPA). Marketers should track CPA and related metrics to ensure that the return marketing investment is maximized.

• Effectiveness—Just because a program is low-cost and efficient does not mean that it is effective. Metrics such as click-through rates and impressions are meaningless unless they can be linked to specific business goals. Rather, marketers should focus on measures that show that effective initiatives are resulting in growing sales and building long-term financial value.

Choosing the tools

Data-centric marketing strategies do not necessarily require heavy investments in hardware or software. Today, the right set of tools exists for companies both large and small.

• Database—Volume and scale allow larger companies to purchase enterprise-level databases and customer relationship management (CRM) systems from IBM and Oracle. Smaller companies can rely on relatively inexpensive, simpler tools. They can start by creating a database using Microsoft Access or MySQL, for instance.

• Lead management—For basic lead management, a product from Act! might suffice. Mid-size companies might choose to move one step up and purchase one of the many products offered by Salesforce.com.

• Web metrics—For most companies, Google Analytics is all they need to track and analyze Web traffic. There are a number of more advanced Web analytics tools in the market, with Omniture being the most dominant.

• Ad serving— Given the complexity and scope of ad placements made by large companies, they are likely to continue to rely on media agencies to handle ad planning and placement. Smaller companies can rely on a variety of tools to place and manage their media assets. Google’s Adword allows companies to place and manage both print and online media. A service from Spotrunner provides thousands of customizable TV ads, which can be bought and placed at a fraction of the price charged by advertising agencies.

• Data mining—For consumer companies with millions of customers, mining data collected might require advanced analytics tools such as SAS or MatLab—but these are fairly expensive and require special training and programming skills. Open-source business intelligence tools such as Pentaho provide a low-cost alternative, but require advanced programming skills. For companies with a small database, Microsoft Excel might be all they need. A popular alternative would be to outsource the data mining overseas.

As you can see, company size and budgets are no longer a roadblock to implementing data-centric customer engagement strategies. Start small, and scale up as business growth justifies larger investments in data systems and analytics. Remember, though, that being data-centric is not about implementing systems or tools; being data-centric is a mindset and business philosophy that marketers at all levels should embrace.

05 Jun

The Eight Major Hurdles to Marketing Measurement

Whether you are trying to determine customer lifetime value or put in place a media mix model, marketing measurement can often be a complex and daunting task.  There are, however, some steps marketers can take to improve the efficiency and effectiveness of their marketing metrics programs

Alignment – In my experience, this is one of the biggest challenges to putting in place an effective measured approach to marketing. That is, lack of alignment behind the need to weave metrics into marketing initiatives. Furthermore, lack of alignment among key departments—finance, technology, sales and marketing—on how measurement initiatives should be deployed and who is responsible for what.

Defining Success – How do we define success? What are the right benchmarks? Should we look at short-term gains in sales or an increase in customer lifetime value? Do we confuse efficiency with effectiveness, and vise-versa? In our industry, talk of ROI or ROMI is ubiquitous. However, marketers manipulate and redefining this term to demonstrate success where none exists.

Data – Data gathering can undoubtedly be a daunting task—tracking codes, DNIs, unique 800 numbers, surveys, POS results, etc. Even if there is a mechanism is place to gather both customer and marketing data, there must be a sufficient sample and enough observations. Furthermore, in order to draw actionable insights, data must be collected, cleansed and mined in a timely manner.

Technology – Even if the will to embrace metrics is there, must companies are constrained by their technology infrastructure. This is specially the case for SMBs. Technology consulting, as it relates to marketing measurement, could be one of the services offered by the new agency.

Attribution – This is a challenge that we faced when working with Dell and other companies. Today’s multi-channel marketing initiatives can be fairly complex—often reaching consumers simultaneously through many points-of-touch. How exactly do we know which ad pushed a consumer from consideration to purchase? We used a variety of methods to deal with the problem of attribution: from the often overused fairness approach to the last touch-point approach to developing RFM models that helped weight results.

Lag – Arguably, most marketing vehicles have a short-term effect. However, there is a residual value that should be taken into account. Often consumers get sold on products and services through long-term brand building initiatives that can be hard to track and measure. How do we account for this lag? As challenging as it may be, it cannot be ignored. Metrics strategies must include long-term equity building initiatives into ROI models.

Methodology – Marketers tend to apply simplistic models to complex measurement challenges. Statistical methods will vary depending on goals defined and data available. For example, media mix models will often involve some type of multivariate regression analysis—linear or logistic. These multivariate analysis should look at a number of dependent and independent variables. For instance, for certain initiatives it might make sense to go beyond the obvious  look at the weather or search engine volume or the consumer confidence index. Furthermore, cost of creative production should be included in these models–but often is not. Ultimately, methodologies used are a factor of data available and measurement goals.

Staff – In my experience, the staff needed to provide marketing metrics services to clients cannot have a background on data analytics alone. It is important to also hire staff with an in-depth understanding of how business models work (i.e. MBAs, etc). Furthermore, it is important for staff to be familiar with account planning techniques as well as advertising models.

22 May

Notebook Thoughts: Marketing and the DCF Trap

The need for publicly traded companies to deliver quarterly results—and therefore take mostly a short-term view to marketing investment—often leads to them falling into a discounted cash-flow trap. That is, the believe that cutting marketing investment will not affect projected cash inflows. In fact, the opposite is true. The more they cut the lower the future cash-flow. The leads to them cutting even further; falling into cash-flow trap that spirals out of control.