6 Reasons To Stop Outsourcing Your Attribution

By Greg Silverman, Dec 06, 2017

CMOs have long viewed attribution as an anchoring component for devising a good market strategy. Developing an attribution model is an outsourced activity.  Consultants and agencies are the go-to option for most brands. Outsourcing the solution established an industry that takes months to produce results, requires specialized data, and has high costs for one-time deliverables. The emergence of software applications that allow brands to bring the attribution capability in-house makes the outsourcing model a bad decision. Here are 6 reasons why:

Outsiders are too slow

A typical project with a consultant answers an attribution question for a single planning period.  If they are fast, they get it done in 2 months.  With the average consultant, it takes 4 months. The whole process is longer if they have sold you on a data management project as well (see below). The report comes in. The consultant answers your follow-up questions. Then the project is over.  Off the marketer goes using their insights. Perhaps the consultant has even left a tool for channel planning with some of the team. A static system is in place and the tool aids decisions based on the past. The data is months old when you go live. Let’s compare their system with the market rhythm, the cadence that the brand is operating at.  The team makes their plans that are adjusted throughout the year.  New media options are emerging.  Competitors make announcements.  Retailers shift focus. The government issues a ruling.  The product is ready. Brand tracking research flows in.  In-market sales results are used to update the model and the attribution percentages. The market often requires you to make decisions quickly as conditions change.

The market often requires you to make decisions quickly as conditions change."

Software solutions empower you to update plans dynamically and to forecast “what-if” scenarios based on continuous information.  No longer do you need a statement of work for a new project.  You walk down the hall, ask your team and they give you an answer in minutes or hours.

Your data is good enough

One of the most prevalent marketing campaigns in the past decade has been promoting the notion that Big Data and Analytics are a necessity. It was a great idea that brought many marketers into a new era.  It established a category.  However, at some point, it became as much toxin as antidote. Those selling data have made BIG a monolithic idea that is the only solution to all problems, both BIG and small. Techniques that grew up around BIG data need big data. This lock between solution and data has kept simple and transparent systems out.

Attribution and marketing mix modeling have become part of the BIG data connection lock-out. This position has made CMOs believe that they must have an expert to manage their data. Brands cannot be good enough at data management to handle a model. The beliefs and perceptions are reinforced to help big data providers keep hold of market share. Meanwhile, CMOs and their teams have all the data they need.  They manage a business day-to-day using it.  New marketing technology should and does fit into their current data world. The optimal systems include more than sales and media.  They start to help understand the brand and how it works as well. Breaking through for the CMO requires a shift in attitude toward data but also a shift in POV on costs.

You can save a lot of money

If you are a CMO, a CFO, a CEO, you are looking to lower your marketing spend without compromising results. If you work with a procurement group they will insist on it. One area where immediate cost savings are possible is to bring your attribution modeling in-house.  Let’s take a look at the business case for software for a brand, a portfolio of brands, and for the unplanned follow-up questions.

If you were to contract with a consultant a typical project for one brand would cost on average $300k. This would likely cover data collection, modeling, a report, and a follow-up question occurring near the end of the project. To make the project work you probably need to assign an FTE equivalent for at least a quarter of the time of the project, which lasts 120 days.  Assume the total loaded cost for the FTE is $150k per year, for a blend of various skill levels on the team. 30 days for this FTE equates to $12,500. So your total project cost would be $300k + 12.5K = $312.5k.  The software alternative would cost $60k for the license and 2 months of the FTE time for $25k.  The total is $85k, a roughly 75% reduction for a cost savings of $227.5k.  For a portfolio of brands (10), the savings can reach over $2 million.

What is even more compelling with software is the follow-on work.  With a model set-up and your team trained, using the model does not require any further outside costs.  So let’s say you assigned a team that was an equal of one FTE for the year and had them update the model quarterly.  The total cost for the license and FTE for the year would be $210K.  And let’s say the consultant was willing to update the model at 50% of their original cost four times.  This cost would be $50k for your team and $600k for the consultant, a total of $650k.  The consultant costs minus the software costs would equal a $440k savings.

Your insights are your advantage

While cost benefits are nice for the short-term, owning your own insights may provide an even greater long-term value.  An unintended consequence of data proliferation is the ubiquity of similar information.  In virtually any category, you can now find syndicated brand research, total industry sales, and current trends in word-of-mouth through secondary or third party research in a day.  Your industry shares insights and data.  And with each passing day of data collection, the value of any commonly available data drops.

Once data becomes benchmarked, proprietary insights are the only thing left to get an edge. Attribution and marketing mix insights show how a brand performs. These tools are one of the few specific things a brand can know. Using an outsourced model bleeds IP back into the market.  What is specific to a brand becomes clay for the potter’s wheel when the consultant goes to the next project down the street. The reality is that consultants dilute proprietary IP. 

I want to be clear. I have never seen a consultant disclose proprietary client insights.  Rather they blend what they learn from you with what they learn from others.  When data was still widely asymmetrical that was an advantage to using a consultant.  When data become proprietary IP, sharing it, even indirectly, erodes competitive advantage.

"When data was still widely asymmetrical that was an advantage to using a consultant. When data become proprietary IP, sharing it, even indirectly, erodes competitive advantage."

It's the better way to learn

“It’s a black box is a common phrase used to describe a model.  For a modeler, those words always come as a shock.  As you build the model, you are constantly documenting assumptions, clarifying settings, and calibrating  the model to the data set you are using for validation.  There are no secrets in the model, no black holes.  Everything is open, transparent, and authentic. You know your work must pass a test – the model must align with what is happening in the real world.  Comparing the model to the actuals produces confidence. Expertise improves because you are learning what works and what does not every day. You learn because you are the user.

Ending outsourcing enables this learning to occur for the brand, not the consultant.  Bringing a system in-house enables a continuous improvement capability for the team.  Data, plans, forecasts, and actual results that define the plan become understood by a core team.  The black box fades away. Variances become learning opportunities. The team learns which creative worked, what data could have been better, or what assumptions were wrong..  Both aggregate and detailed insights come faster, allowing the team to adjust sooner.

It's the way to survive in your C-suite

CMO tenure is short and maybe even declining. CMOs face unending pressure for ROI clarity and proof of the value of marketing.  Today the tools available to them are often in silos or partial parts of the whole picture.  In a recent survey, Winterberry reported that 75% of brands use 5 to 30 tools to understand their marketing. The new attribution systems create a planning platform that unifies many of these tools and creates one common body of knowledge. Consolidation of insight helps explain the case of how marketing is working. A system delineates how the next set of plans is likely to deliver results and improve the value of the brand.

At this point, you may be asking, who are the purveyors of these tools? Where can I find them?  Since the category is early in its development, neither Forrester nor Gartner have evaluated the software only marketplace yet.  So we have made a list of key vendors that we see moving to the self-service model:

Bizible – B2B attribution modeling with technical support http://www.bizible.com/

Concentric – B2B and B2C marketing strategy with automated attribution https://www.concentricmarket.com

ScanmarQED – B2C Marketing mix model software with data services https://www.scanmarqed.com/

 

On-demand-Webinar-Blog-Post-CTA-1000x552-1

 

Cognitive Dissonance: Why It's Hurting Your Analytics
What is Cognitive Dissonance in Business?
Simulation for Strategic Decision-making in Times of Uncertainty