On improving ad spend effectiveness and efficiency

by Robert Palumbo

What's the vision behind the work Sean Billings and I are doing at OutPoint, and why do we think it is so important?

This year, global ad spend will jump 10.2% to a record $651 Billion (with a B) in 2021 after falling 4.1% in 2020. But behind the record growth numbers lies a persistent problem around wasted spend: according to data aggregated from 70 studies, the median estimate is that 40% or $260 Billion of that spend will go to waste.

Wasted spend is, clearly, a huge economic problem for advertisers. Neither businesses or consumers benefit, creating a significant deadweight loss that only benefits a few large incumbents content with maintaining the status quo:

A deadweight loss is a cost to society as a whole that is generated by an economically inefficient allocation of resources within the market. A drop in the efficiency of resource allocation matters because it results in a reduction in welfare throughout society. This is highly significant because welfare, in economic terms, refers to a society’s living standards and overall prosperity. A deadweight loss is typically measured in terms of income or GDP (gross domestic product). - The Intelligent Economist

Our in-house definition of wasted spend is media dollars that do not map to business performance (including brand building) or are well past the point of diminishing and negative returns. To reduce waste, it is optimal to make ad spend decisions based on an understanding of incremental value, cross-channel impacts, and marginal returns. The causal effect of showing an ad to a potential audience or not, commonly referred to as “incrementality,” is a central question related to advertising effectiveness, reducing wasted spending, and budgeting for optimal growth

Suppose we can work to reduce the deadweight loss by re-allocating inefficient dollars to a more effective channel. In that case, the revenue of the businesses we serve increases, creating more potential for economic growth, job creation, customer relationships, and opportunity.

Our mission is thus to grow the size of the digital economy through marketing science. Stripe and Shopify are two inspirations for the OutPoint vision: these companies articulate missions around enabling GDP growth and "arming the rebels" against the status quo.

Alternatives to fix the problem

Growth-oriented companies are not blind to the multi-billion dollar 'wasted spend problem' and deploy tremendous resources trying to fix it, in search of higher return on ad spend (ROAS).

It is now possible to leverage data to model scenarios to reduce waste and improve effectiveness of ad spend, but it is challenging, complicated work.

 Here's what the world looks like today:

  • Status quo: This is the default option. Decision makers will continue to make media allocation decisions based on on gut feel, Excel sheets, biased reporting from leading ad platforms, and legacy analytics tools that are fine at historical measurement, but lack forward-looking decision-making rigor. For example, many ad spend decisions are made based on reported Average CAC vs. an understanding of incrementality and Marginal Returns. (See: Marginal vs. Average CAC)

google analytics
Google Analytics Channel Reporting

  • In-house data science teams: Many top-tier growth operations, such as Stitch Fix and Wayfair, have opted to build tools and systems in-house to optimize spend allocations. This is a great option for scaled-up organizations, but prohibitively expensive for most earlier-stage companies. In addition to the cost, Data Science talent is scarce.

For an example, check out this job description for Sr. Data Scientist, Growth Algorithms at Wealthsimple to see how a leading brand defines the marketing science opportunity:

Wealthsimple Job description

  • External media mix modeling (MMM) consultancies: an expensive, low-frequency, project-based option (See: Nielsen MMM). These can be very well-done, depending on the vendor, but are typically one-offs vs. a tool that is useful for ongoing decision making.
As a modern marketer, you’re leveraging a wider array of channels and tactics to reach your customers than ever before. The question is, how do you know which activities are actually driving revenue and profit, and where you should be spending to maximize impact? With Nielsen’s Marketing Mix Modeling solution, you can assess the impact of your investments, see what’s working and optimize your marketing and spend accordingly.
  • nielsen
    Outsource media buying decisions to an agency: many agencies do great work, and we are proud to partner with them, but even so, the decision making may suffer some of the same flaws of the status quo option. If the decision maker is not making decisions based on incremental value and marginal returns, there may be wasted spend. Further, agencies may charge a fee of 10-15%+ of ad spend, which could be prohibitively expensive compared to bringing this capability in-house.

We think the problem is best solved with data science and technology, paired with an understanding of marketing fundamentals.

With OutPoint, you can simulate optimal paid media mix across current and new paid channels. OutPoint's models apply media mix modeling principles in real-time, using machine learning and causal inference. Our solution is a forward-looking decision-making system that aids marketers across the following areas:

  • Understanding true marketing incrementality and where the efficient frontier for marginal returns is for each channel 
  • Channel-level predicted outcomes at various spend amounts, taking into account how each channel interacts with the others
  • Future budget and channel allocation recommendations to achieve higher revenues, lower costs, or both

In conclusion, we think there is a better way forward to improve effectiveness, reduce waste, and help companies scale paid growth. This path forward relies on data science and software.

Does this sound like an interesting and valuable problem to you? We're hiring for a Full-stack Engineer and expect to have many more roles available across data and engineering in the months to come.