The difference between mixed media modeling and multi-touch attribution

Marketing attribution is the process of assigning credit to different marketing channels and touchpoints for a conversion. This allows businesses to optimize their marketing efforts and allocate budgets more effectively. Mixed media modeling and multi-touch attribution are both useful marketing analytics techniques, but they serve different purposes. While multi-touch attribution is a valuable tool for understanding the customer journey and optimizing individual marketing channels, mixed media modeling provides a more comprehensive view of marketing performance and helps businesses make more informed and effective marketing decisions.

Here are some ways mixed media modeling differs from traditional multi-touch attribution methods:

Holistic view of marketing performance: Multi-touch attribution is limited in its ability to account for cross-channel interactions and the effect of offline touchpoints on online conversions where mixed media modeling considers these and provides a more comprehensive view of the customer journey. For example, it can show how television advertising affects website traffic, or how social media engagement affects search engine rankings. This allows businesses to identify new opportunities for growth and optimize their marketing strategies accordingly.

Use of first-party data: Mixed media modeling allows for the incorporation of first-party data, such as website and customer data, to provide a more detailed and accurate picture of the customer journey. In addition, it can enable more precise audience targeting and personalization of marketing campaigns, as well as provide insights into customer behavior and preferences that can inform product development and improve overall customer experience. By leveraging first-party data in mixed media modeling, businesses can build stronger relationships with their customers and gain a competitive edge in the market.

External factors: Mixed media modeling can provide brands with valuable insights into the impact of external factors on their marketing performance. For example, it can help brands identify how changes in the competitive landscape, such as the entrance of new players or changes in pricing strategies, are affecting their ability to attract and retain customers. It can also help brands understand how shifts in consumer behavior, such as changes in shopping preferences or the adoption of new technologies, are affecting the effectiveness of their marketing efforts. Through these understandings, brands can adjust their marketing strategies accordingly to stay competitive and maintain a strong position in the market.

Use of advanced modeling techniques: Mixed media modeling utilizes advanced modeling techniques, such as machine learning and causal inference, to provide a more thorough understanding of the relationship between different touchpoints and conversions. This results in deeper insights into customer behavior, enabling businesses to make more informed and effective marketing decisions. Additionally, the use of these techniques allows businesses to simulate and test various marketing strategies and scenarios in a virtual environment, allowing for evaluation of potential impact before real-world implementation, thus minimizing the risk of costly mistakes.

Maximize marketing performance

Ultimately, while multi-touch attribution is useful in understanding the customer journey and optimizing individual marketing channels, mixed media modeling offers several advantages that make it a more comprehensive approach to marketing attribution. By providing a more accurate view of marketing performance, mixed media modeling enables businesses to identify new opportunities for growth and optimize their marketing strategies accordingly. 

Additionally, the use of first-party data and advanced modeling techniques like machine learning and causal inference provides deeper insights into customer behavior, allowing businesses to make more informed decisions. Overall, by incorporating mixed media modeling in their marketing attribution strategies, businesses can optimize their marketing efforts, build stronger customer relationships, and gain a competitive edge in the market. 

Want to know more how Keen leverages the power of AI to help marketers stay agile and responsive in this fast-changing digital world?  Go here to learn more.


How to transform marketing strategies with the power of AI

The days of traditional advertising campaigns that rely on guesswork and intuition to make spend decisions are quickly coming to an end. That’s in part because today’s marketers have a powerful tool at their disposal: AI.

AI can help marketers respond quickly to changes in their target audience’s needs by processing large amounts of data and providing insights and recommendations in real-time. This empowers them to create and implement data-driven marketing strategies that are more effective and efficient than traditional methods. With continuously-learning algorithms, marketers can run scenarios instantly and get prescriptive and predictive assistance in their decision-making.

Real-time marketing optimization for better results

Imagine you are a brand leader for a new energy drink company. You have just launched a campaign aimed at millennials, but you’re not seeing the results you were hoping for. With traditional marketing methods, you would have to wait for weeks—in reality, even months—to get a report on the campaign’s performance. And by that time, it’s often too late to make any changes.

Certain AI and machine learning platforms allow daily, real-time data input which, in turn, generates weekly performance reports. That means you can see which ads are resonating with your target audience and which ones are falling flat so you can adjust your messaging, targeting, and creative execution on the fly to optimize your campaign for better performance. Algorithms can also analyze market and societal factors to provide you with insights and recommendations for adapting your marketing mix. So if a new trend emerges among millennials, it can help you identify it and adjust your strategy accordingly. 

So let’s say that there’s a new trend among millennials towards plant-based diets and sustainable living. As a brand leader for a new energy drink company that targets this demographic, you might be wondering how to tap into this trend. By analyzing sales and marketing data, AI can more quickly identify patterns and insights that reveal the motivations, preferences, and needs of this audience segment so you can adjust your marketing mix to align with this trend.

By leveraging the processing power of AI, marketers can stay agile and responsive in a fast-changing digital landscape. With the rise of new technologies, platforms, and consumer behaviors, brand leaders need to be able to adapt quickly and experiment with new approaches to deliver better performance and results. AI-powered marketing optimization allows decision makers to test and refine their strategies in real-time, without having to wait for quarterly or annual reports. This can help them stay ahead of the curve and respond to emerging trends and opportunities before their competitors do.

Keen to know more? Contact us to see real-life examples of prescriptive and predictive assistance derived from the AI and machine learning power of our unique decision optimization engine.


The next evolution in marketing forecasts

Marketing budgets are crucial for businesses of all sizes, and marketers are often tasked with the challenge of maximizing returns on their spend. However, when faced with budget cuts, the situation becomes even more difficult. Marketers must be able to identify areas to optimize and prioritize their spend to ensure that they achieve their business objectives.

One of the biggest obstacles they face is forecasting the future. Whether it’s a global pandemic or some other external event, the future is always uncertain, and no one can predict with complete accuracy what will happen. However, there are ways to prepare for the unknown, and one of the most effective is by using marketing forecast tools. So who says you can’t get a glimpse of what might be around the corner?

Gone are the days of relying on a single forecast for the next quarter

Traditional financial planning has relied on historical performance data to forecast the impact of budget cuts on a business, which fails to account for the nuances of different scenarios. Fortunately, advances in technology have made it possible for businesses to go beyond single marketing forecasts and instead focus on a range of outcomes.

The key is to start with assumptions or hypotheses about what might happen. For example, if there is a new product launch, marketers might assume that it will have a positive impact on sales. However, they need to consider different scenarios that could happen, such as the product launch being delayed or customers not responding as positively as expected.

By running multiple scenarios, marketers can better understand the range of outcomes that could happen. This provides a more complete picture of what might happen in the future, allowing them to make informed decisions based on the most likely outcome. For example, if the best-case scenario is a 20% increase in sales, and the worst-case scenario is a 10% decrease in sales, marketers can plan for both outcomes and develop contingency plans accordingly.

Scenario planning not only helps identify areas to prioritize, but it also helps manage risk and pinpoint places where spend can be cut back without negatively impacting business objectives. If a marketer is faced with a 10% budget cut and is considering pulling out of paid search or TV advertising, they can input different scenarios into a tool to determine the impact on revenue and profitability. This allows them to make data-driven decisions and optimize their remaining budget accordingly.

In today’s fast-paced business environment, marketing departments need to stay ahead of the curve and predict the impact of any proposed changes to their strategy. With the ability to run different scenarios, brand leaders can now have a stronger voice when it comes to those discussions rather than just pointing to a previous study that had been done. 

By effectively demonstrating the correlation between expenditure and business goals, marketers can show other departments and stakeholders the potential impact of their proposals. This enables them to advocate for their department and ensure that the most effective decisions are made to help the company outpace the competition, whether it involves budget cuts, reducing marketing spend, or implementing a new positioning strategy.

Keen to know more? Contact us to see how real-life brands proactively adapt to budgetary, economic, and societal changes in real-time.


Keen’s portfolio view opens a national consumer brand’s umbrella

Keen's portfolio view opens a national consumer brand's umbrella

Allocating your marketing budget across multiple brands doesn’t have to be a daunting task.

Keen helped a leading consumer brand streamline and enhance its marketing strategies across seven product lines. By evaluating the financial contribution of each channel for the brand's portfolio, Keen was able to identify specific shifts to optimize spending by week across products and channels. This optimization produced a 3.1 percentage point improvement over the previous marketing approach, resulting in an additional $24.1M in profitability. The brand successfully implemented this data-driven plan, achieving a payback time of less than 1.5 days net their investment in Keen.

A graph that shows a bayesian MMM with priors.

The benefits of Bayesian marketing mix modeling

Marketing mix modeling (MMM) has been used for decades to help companies understand how their marketing efforts impact sales. Traditional approaches typically use a standard regression analysis, which involves fitting a linear model to the data. However, these models can be limited in their ability to capture the complexity of real-world marketing environments, and may struggle to account for factors like seasonality, non-linear relationships between variables, and interaction effects.

The Bayesian MMM approach solves for the limitations of traditional marketing mix modeling. By incorporating prior estimates of tactic elasticity, and updating those prior beliefs with new data, the model can adapt and learn over time. This is a more powerful and flexible approach, since it can handle a wider range of data types, incorporate prior knowledge, and provide a probabilistic framework for modeling marketing data.

Here are five benefits to choosing a Bayesian marketing mix modeling.

1. ROI calculation

Bayesian MMM helps companies optimize their marketing spend by accurately estimating the ROI of each marketing channel, reflecting the impact of each on overall sales, and accounting for the interplay between channels. This allows marketing leaders to make data-driven decisions about how to allocate their budgets and optimize their resources. For example, a company may find that their ROI is higher for a particular channel than previously realized due to previously unestimated halo effects. Alternatively, if the ROI is determined to be lower than previously believed for a particular channel, they could decide to shift resources away to focus on more profitable tactics. By quantifying the relationship between marketing spend and sales, companies can gain a more nuanced understanding of how their marketing efforts are affecting their bottom line. 

2. Better decision making

Using data to understand the relative impact of different channels means companies can be confident that they are getting the most value from their marketing budgets and drive growth and profitability in a more targeted and effective way. Data-driven decision-making can lead to better business outcomes in a number of ways. By using a Bayesian approach to optimize their marketing spend, companies have more sophisticated insights, and can therefore drive more sales and generate a higher ROI. Additionally, data-driven decision-making helps companies identify areas for improvement and fine-tune their strategies over time. By continually analyzing data and adapting their approach, brands stay ahead of the competition and remain responsive to changing market conditions. 

3. Flexibility

Bayesian MMM is a flexible approach that models a wide range of marketing variables, such as advertising spend, pricing, promotions, and other factors that impact sales. This is a key advantage because it allows companies to adapt to changing market conditions and consumer behavior. By modeling a wide range of variables, companies can identify which marketing tactics are most effective in different situations, and adjust their strategies accordingly. For example, a company might find that their social media campaigns are having a much larger impact on sales during certain seasons or in the presence of other channels, and could adjust their marketing spend accordingly. By using a flexible approach, companies stay agile and responsive, and continue to drive growth and profitability in a rapidly evolving business environment.

4. Robustness

Bayesian MMM can handle a wide range of data types, including continuous and categorical variables, as well as data with missing values. This is because the Bayesian estimation process utilizes prior estimates, which allows the model to fill in any historical data gaps. This robustness provides a more accurate picture of the true impact of marketing variables on business outcomes. For example, if a company is missing data for a particular channel, a Bayesian estimation can estimate the tactic performance, and still provide accurate estimates of its impact on sales. This robustness also helps companies avoid making decisions based on incomplete or misleading data, which ultimately lead to better business outcomes.

5. Bayesian MMM approach

Bayesian models use probability theory to estimate the likelihood of different outcomes and the degree of uncertainty associated with those estimates. This framework enables the model to incorporate prior knowledge about the relationships between marketing variables and business outcomes, which can improve the accuracy of the model. Additionally, by addressing uncertainty with a monte carlo simulation, Bayesian MMM provides a range of possible outcomes with associated probabilities, rather than a single point estimate. For example, if a company is deciding between two marketing tactics, a Bayesian marketing mix model can estimate the probability of success for each tactic and offer several potential scenarios, which leads to more informed decisions about how to allocate marketing resources. 

Choose power and flexibility

Overall, a next-generation Bayesian MMM is a powerful and flexible approach that accurately models complex relationships between marketing variables and business outcomes using prior knowledge. It provides more accurate ROI estimations than traditional methods, more accurate sales and revenue forecasting, and a more comprehensive understanding of the impact of all marketing tactics.

In a complex and competitive marketplace, data-driven decisions are crucial for optimizing marketing spend and achieving better business outcomes. A Bayesian approach offers a valuable tool for gaining a comprehensive understanding of the impact of marketing tactics and making informed decisions about resource allocation.

Keen to learn more? Request a demo model

cookieless attribution - an image of a cookie on a green background

Navigating a cookieless world: challenges and opportunities for marketers

Cookies were created to improve the user experience of websites by allowing them to remember certain information about a user as they navigate the site. While cookies have provided many benefits for both users and marketers, concerns have been raised about the privacy implications of cookie tracking. In recent years, web browsers and regulators have taken steps to limit their use and protect user privacy, which has led to the development of alternative methods for collecting data and delivering personalized experiences.

In the past, cookies were a valuable tool for tracking users across devices and browsers, and for attributing conversions to specific marketing campaigns. However, with the rise of privacy concerns and the increasing use of ad blockers, cookies are becoming less effective and in some cases, unavailable. As the world moves towards a cookieless future, marketers are facing new challenges in understanding and reaching their target audiences.

The problem of cookieless attribution

One of the main challenges that marketers are facing in a cookieless world is attribution. Attribution refers to the process of identifying which marketing channels or touchpoints led to a conversion or a desired action by a user. Without cookies, it may be more difficult to track users across devices and browsers, and to attribute conversions to specific marketing campaigns or efforts. This makes it harder for marketers to optimize their strategies and understand which channels are most effective for reaching and converting customers.

In the absence of cookies, it becomes much harder to track a user’s behavior across multiple devices. For example, a user may browse a website on their laptop and then make a purchase on their mobile phone. Without cookies, it is difficult to attribute the purchase to the advertising campaign that led the user to the website in the first place.

Another challenge of attribution in a cookieless world is the loss of granular data. Cookies allow marketers to track user behavior at a very granular level, such as which pages the user visited, how long they spent on each page, and which actions they took. The loss of cookies makes it more difficult to track user behavior in such detail. This can make it harder to identify which touchpoints or marketing channels are driving conversions.

The loss of third-party data

Third-party data refers to data that is collected and shared by companies other than the one it is originally collected from. In recent years, there have been several high-profile breaches and privacy scandals that have led to increased scrutiny of this practice. As a result, many companies are now facing challenges related to third-party data loss.

One inherent challenges is the loss of valuable insights and intelligence. Third-party data is often used by companies to enrich and build a more complete understanding of their customers and target audience. This data can be used to personalize marketing campaigns, improve customer engagement, and drive revenue growth. The loss of third-party data can also have legal and compliance implications because many companies rely on it for compliance with regulations such as GDPR and CCPA. 

Another challenge of third-party data loss is the impact on advertising and marketing campaigns. Third-party data is often used to target advertising and marketing campaigns to specific audiences. For example, a retailer may use it to target ads to users who have recently searched for a specific product or who have demonstrated an interest in a particular category of products. The loss of this data can make it more difficult for companies to effectively target their advertising and marketing campaigns, which can lead to reduced effectiveness and ROI.

Effective alternative methods for cookieless attribution

With the disappearance of cookies, marketers will have to rely on other methods of identifying users and their behaviors. These include using first-party data, IP addresses, device fingerprints, and probabilistic models, which can be less accurate than cookies. In order to overcome these challenges, marketers will need to adopt new strategies and technologies. One of the most promising solutions is the use of deterministic identity solutions, which rely on user-provided information, such as email addresses or phone numbers, to identify and track users. 

Companies can focus on building their own first-party data by encouraging customers to opt-in to data collection and by offering personalized experiences and incentives. By collecting data directly from customers, marketers can gain a better understanding of their preferences and behaviors, and can target ads more effectively. They can also work to build more transparent and ethical data practices to ensure compliance with regulations and build trust with customers.

While the transition to a cookieless world may be a challenging time for marketers, it also presents an opportunity to reassess their marketing strategies and discover innovative ways to connect with customers. By embracing alternative methods and technologies, as well as building a more transparent and ethical approach to data collection, companies can continue to deliver personalized experiences and effectively target their advertising and marketing campaigns, while also prioritizing user privacy and compliance with regulations. With the right approach, marketers can successfully navigate the challenges of a cookieless world and thrive in the new era of digital marketing.

Want to hear how Keen can help?  Contact us today.


Optimize your marketing spend with Keen

Marketing can be a challenging and often unpredictable field, which is why having access to reliable data and analytics is so important. That’s where Keen can help with our patent-pending Marketing Elasticity Engine that consists of informed priors by category, proprietary response curves, and short- and long-term decay rates. By combining our knowledge estate with your brand’s time-series data and prior ROI and lift studies, Keen is able to provide both a comprehensive historical analysis and future-looking marketing planning scenarios.

To achieve this, we take the comprehensive dataset and run it through our Bayesian regression model, which generates an output that includes model statistics and transparent priors. The use of Bayesian regression allows us to estimate the relationship between variables by incorporating prior knowledge or beliefs into the analysis.

Based on this output, Keen delivers your historical performance data, just like you would get from a traditional mix provider. However, Keen’s offering doesn’t stop there. We also run our machine learning predictive algorithm to build optimized marketing spend plans that consider your historic performance and projections on outside factors, such as distribution, category movement, price trends, and more. On top of each plan is a revenue forecast, which typically estimates within a 4% margin of error.

One of the strengths of Keen’s approach is its ability to pull levers to wargame scenarios based on your business objectives. This capacity empowers you to optimize current and future spend decisions. The always-on model means that the engine is available in the cloud, so you can run what-if scenarios whenever you want. 

As you execute your plan, Keen updates your knowledge estate to include the most recent data, which informs the prior in our model and offers real-time insights. This means you’re always up-to-date on the latest data and trends, giving you the necessary information to make data-driven decisions.

Overall, Keen offers a powerful and easy-to-use solution for companies looking to optimize their marketing spend. By combining your data with Keen’s knowledge estate and running it through our Bayesian regression model, you can gain a better understanding of the relationship between different variables and make informed decisions that help achieve your business objectives.

To learn more, request a demo model.



How to optimize your future marketing spend with past and present data

In a competitive business and branding environment, one wrong marketing decision can sink a company. Or at least a stock price. In a perfect world, the better product should always win. But that’s not always the case. All things equal, the difference between a market leader and a follower is the marketing mix. No pressure.

Typically, as a brand marketer, you begin the planning process by taking a look at performance metrics from the previous quarter or year. Don’t get us wrong, that historical data is valuable and informative in its own right. But it’s just the tip of the iceberg for your marketing spend. What lies below the surface is a new dimension of marketing insight optimized through advances in AI technology and machine learning.

Marketing spend optimization in the face of unforeseen competitive and societal shifts

Let’s take a look at two fictional, high-end cooler companies. Both coolers look the same, keep drinks cold for the same amount of time, and cost the same. But, only one marketing team is able to respond to a real-time challenge. In this case, the government passes a strict tariff on raw materials essential to manufacturing. Margins collapse, Suits are sweating, and their departments are tapped to cut expenses.

First up, Old School Coolers. Marketing is legislated by the C-suite. They live in the past, desperately trying to predict future performance trends by accessing limited data on a quarterly basis. Their decisions are informed by what they did in their last crisis. And they scramble to identify which contracts they can cancel the fastest and with the lowest penalties. They have to wait at least another quarter for their data partner to give them new insights, Essentially, they’re alone.

Future Forward Coolers takes a different approach. Leadership believes marketers should not be blindfolded when planning for a crisis. By leveraging Keen, this team rose to the challenge. While planning earlier in the year, historical data was merged with outside data sources—trends and tactical performance metrics—to most effectively predict performance and profitability measures. With this foundational model in tow, periodic marketing mix checks were made throughout the year benefitting from Keen’s Elasticity Engine which continues to strengthen through the ingestion and analysis of contemporary performance and profitability data. So when the crisis hit, Future Forward’s team changed its parameters, ran new what-if scenarios, and delivered the adjustments needed to ensure their ROI remained consistent. The right questions were answered and the marketing spend was optimized.

What sets the two marketing teams apart is the use of advanced analytics to optimize performance and profitability; bridging the gap between marketing goals and financial data. Having access to a system of aggregated user data while simultaneously running future-powered and timely scenarios, can turn a good product into an unbeatable brand.  

Keen to know more? Contact us to see how real-life brands leverage data outside of their own to open up a new world of planning capabilities. Whether it’s once-a-year, once-a-quarter, or a once-in-a-lifetime crisis, we can plan for it.


World’s third largest confectioner scores sweet results

World's third-largest confectioner pumps fresh air into Airheads and Mentos brands

Discover marketing measurement that does more than measure your success.

Keen worked with the world’s third-largest confectioner Perfetti Van Melle to pump some fresh air into their TV strategy for iconic brands Airheads and Mentos. Industry measurement guru Bill Mackison leveraged Keen’s future-focused models to de-risk what appeared to be a risky TV investment decision: breaking the 40-GRP rule of thumb. Learn how the confectioner sweetened its results with a different TV strategy, backed by insights from Keen.


From shut down to rebound (Bob’s Discount Furniture)

From shut down to rebound (Bob's Discount Furniture)

What you’ll learn from this case study:

- How one furniture retailer's agile strategy propped up their back-half sales in the midst of COVID-19 store shutdowns

- How the team navigated lost foot traffic and re-opened their stores with their strongest marketing performance to-date and healthy marketing spend levels

- How to increase your agility so you can pivot quickly in response to the unexpected in 2022 and beyond