How AOL’s Approach To Online/Offline Attribution Is Set To Evolve

It's been almost a year-and-a-half since Verizon acquired AOL for $4.4 billion. It's also been a year since AOL itself bought mobile ad platform Millennial Media and revamped the MapQuest brand. Amy Mitchell explains how far the internet pioneer has come and where it's going in bridging physical and digital consumer analytics.

AOL has been enjoying its second life since completing its evolution from an internet service provider into an ad network and then into programmatic platform company even before being purchased for $4.4 billion by Verizon.

Whether or not AOL platform rival Yahoo is acquired (or not) by Verizon, the company is already moving forward on a number of fronts to deliver a clearer set of online-to-offline attribution solution.

While “omnichannel attribution” is being promised by practically every major ad tech company, Amy Mitchell, AOL, GM and head of Convertro (the attribution provider AOL acquired over two years ago), offered her take on the state of digital-to-physical marketing data, the integration of Millennial Media and revamping of MapQuest has had on its capabilities.

Lastly, Mitchell also outlined AOL’s approach to achieving higher levels of geo-data accuracy as it becomes the crucial differentiator among marketing platforms, a point that was also outlined in this week’s Forrester Wave report which ranked Convertro as a “strong performer.”

GeoMarketing: How would you describe the state of online/offline attribution? What are the main challenges right now?

Amy Mitchell: Online/offline attribution is much more advanced today than it was seven years ago when we started. However, challenges are still there and need to be acknowledged. There are four different types of challenges: data, attribution methodology, activation (acting on the insights) and service.

On Data

The quality of attribution is only as high as the quality of inputs of data and the types of data ingested. Back in the days, when MTA was purely digital-focused, it wasn’t a real issue- you just ingest log files or place pixels and there you have it: great, granular data.

Today, when you talk about offline attribution, location, people moving across devices, algorithms that account for macro-economic factors etc. getting access to data and ingesting it becomes a critical thing.

Getting access and ingesting the data at that scale is extremely difficult- marrying together different data types, different taxonomy, different cadence and always maintaining consistency is key.

But even if you have a really good automated data-ingestion and quality control mechanism, you still need to obtain data that can help you explain people’s behavior. For example, to understand that this user was buying at GAP because he/she has a habit of going there every month. This can only be obtained by getting highly precise, highly granular location data which Verizon and AOL obviously have and attribution company like Convertro can utilize.

Attribution Methodology:

Whether we’re assessing online and offline attribution separately, or looking at how they work together, one of the most fundamental challenges for enabling smarter attribution is the lack of a unified methodology within the industry.

Traditionally, sophisticated marketers employ multiple solutions to understand marketing ROI and forecast optimal spend. Most use two separate models – Marketing Mix Modeling (MMM) and Multi-Touch Attribution (MTA) – to understand the true return on advertising investments and inform decision-making. MMM explains the 40,000-foot story to executives and leadership while MTA explains the 4-foot-high individual consumer story to media planners, buyers and technologists. Each works well in its own right. However, because the models are fundamentally different, developed using different data sets and methodologies, the results are most often inaccurate or biased when combined and cobbled together. This impacts all forms of attribution – online, offline as well as understanding the connectivity across the two domains.

Recently, the industry has moved to more closely marry these two models. Having a unified model – MMM plus MTA – that exists within a single framework enables marketers to more effectively execute and optimize their efforts. In addition, it provides a consistent and more accurate set of metrics, drives more informed decisions for the allocation of marketing budgets across channels, and helps brands and marketers understand and act on the entire customer journey to achieve key objectives. This is the key to enhancing attribution, at large.

Last year, Convertro launched its own unified service – our Unified Marketing Activation Platform (UMAP). UMAP adds all of the benefits of MMM on top of impressive MTA to deliver hyper-accurate evaluations of omni-channel marketing measurement and campaigns.


No one cares about attribution for the sake of attribution. Everyone cares about the action that they can take and the industry’s job is to make it easier for them. Stand alone solutions are great but when you plug attribution into a buying system and automatically make changes based on attribution intelligence you’re able to move faster and close the loop to increase your ROI or achieve your other objectives faster.

Convertro is part of AOL ONE platform that empowers marketers by giving them access to premium inventory at scale across display, mobile, video and TV. Combining these buying tools with the intelligence of unified holistic attribution when there’s zero “cookie loss” is key to increasing performance.

In addition to the above, more programmatic world, good attribution vendors have to help customers adopt the recommendations and get buy-in and acceptance within the organization. Organizations that have long been working on a last click basis and are crediting search for everything good that TV is doing will need to move away from that practice when introducing advanced attribution.

This aspect of change management and service is a key that is sometimes being overlooked by vendors in our industry.

Tech-enabled service

As mentioned above, service is key. But it’s not any service- it’s the right service. Marketers don’t want to pay a vendor for doing work that machines can do, so the key in tech-enabled service is to leverage technology for automation and creation of tools that enable customer success representatives to focus on change management, strategic recommendations, deep vertical expertise and helping organizations adopt.

What has Convertro, and by extension, AOL done recently to advance its capabilities in online and offline attribution?

AOL and Convertro have been investing across all pillars that were outlined before: data, methodology, activation and tech-enabled service.

Since the Verizon acquisition Convertro has been getting access to deterministic cross device data for attribution and analytics purposes.

The launch of our new UMAP platform has been core to our attribution strategy. With a unified engine that seamlessly blends MMM and MTA, our customers can more accurately realize and understand historical performance in their advertising campaigns while predicting future outcomes across offline – print, OOH, TV, direct mail, catalog – and digital – search, display, video, email, social, affiliates, and more – channels.

Through UMAP, we believe that we have meaningfully strengthened our leadership position in the market – which is only confirmed by today’s Forrester Wave. According to the report, we are the highest-ranked across several key categories, including “technology platform.” This speaks to the investments we’ve put forth in attribution, knowing how increasingly critical it is for all advertisers and agencies seeking to understand ROI and optimize spend.

We rolled out what we call “act button” which is enabling customers to activate recommendations for their search campaigns with one-click from within the Convertro platform.

On the tech-enabled service front, we started making more visible the quality of data that is going into the algorithm and recommendations. Transparency is key and automated transparency is enabling us to “open the kimono” to all of our customers. We added a data quality section that shows you the current state of integrations and data that is flowing in. This is based on a data monitoring and alerts system that we put in place to validate monitor the data and early-detect issues.

How does geo-data factor into online and offline attribution?

For attribution, geo-location data is central to contextualizing the customer journey. We must determine accurate location and overlay this against the customer journey to help our advertiser and agency partners truly understand what’s happening both online and offline. We also ingest location data, using it as a tentpole guide across the non-media, context-based factors we account for in our model. Non-media factors are econometric (unemployment, housing, etc.), environmental (weather, etc.), and brand-related (affinity, customer loyalty, purchase history, etc.). Geography is often the backbone for everything as we localize these factors down to the individual for the most accurate intelligence and predictive insight.

Specifically, user-level geo data can be two things:

Explanatory factor as a driver to outcome (someone who’s going into the shop every month), this improves the accuracy of marketing effectiveness measurement

Outcome factor (understand what drives people to visit in store)

You can then join this information with additional geo-related conditions such as weather and help CMOs spend more wisely.

For example, we can then look at the city level and decide to spend more heavily in specific DMAs where I can be more efficient given conditions in that market.

How does AOL make use of its other units — from Convertro to Millennial Media to Mapquest — in order to generate attribution intelligence for clients? Or are those integrations still a work in progress (or case by case basis)?

Great attribution intelligence is all about: (proprietary) data, attribution methodology, activation and tech-enabled service.

A lot of AOL/Verizon synergy for attribution is around proprietary data like:

  • Device graph
  • Audience data
  • Location data to identify store visits and then which marketing is more effective in driving people to store (leveraging several units here: mapquest, millennial, verizon)
  • TV data
  • Connected cars, smart homes

Work here is ongoing and we continue to add data to make the attribution smarter.

A big part is also about activation- single platform to have closed-loop ROI optimization machine with zero loss of data in between. Convertro is already integrated into the ONE platform and continues to partner with other AOL unites.


What are your goals around enhancing attribution capabilities looking ahead to 2017?

We believe that location data/analytic and understanding people’s behavior offline is key for 2017. The more we can understand the customer journey offline in the same way we understand it online today the better attribution we get. Leveraging our proprietary assets to understand how people move between different locations can help the OOH industry, retailers, CPG, financial institutes and more.

For marketers, this translates into being able to understand better where to place OOH messages and how to be more accurate, timely and targeted in those messages.

For consumers, this translates into getting more relevant and personal messaging delivered to them at the right place, at the right time and on the right device.

We also believe that the industry no longer cares about MTA, MMM or any other acronym. The industry is looking for attribution that is practical- that enables you to get answers to your marketing questions (who/where/what/when) and powerful- built on superior technology, fast, real time and user level.

What we continue to do in 2017 is to get access to more user level data and use of attribution for activation. We already rolled out the ability to act on automated recommendations from within the platform with some vendors and we will continue to add more partners there

Is there anything in particular that brands should ask attribution providers when it comes to evaluating a solution? Is it about the source of the data?

Evaluating effective attribution providers really comes down to three components: 1) the data; 2) the science making it actionable; 3) and the platform everything is encased in.

On the data side, the best vendors relentlessly pursue consumer-level data for more granular understanding. Convertro, for instance, has access to unique consumer-level data points provided by AOL and Verizon. This add layers of accuracy and consumer insight available through no other measurement provider. How fast can you get this data in and how automated the ingestion process is are additional key questions. Legacy MMM providers take pride in ingesting multiple types of aggregate data but as we all know getting MMM report back takes months partially due to slow and manual ingestion.

But it’s not enough to simply have data — you have to have the science to enable it and drive actionality. Our UMAP unified platform is the first such approach to holistically model both aggregate data and consumer-level data parameters simultaneously. Our model can also be customized quickly to account for unique media and non-media factors that influence the performance of marketing campaigns. These science capabilities allow providers to account for context, which is a necessity. You need to truly understand HOW they are actually combining the data and the science. You will hear many claim to have a “unified” solution, but often that just means they are doing a “top-down”/”bottom-up duct-taped together solution. Two inherently biased models don’t equal one unbiased model. The HOW can lead to real quantitative quicksand (analysis paralysis) because of the discrepancies that emerge. You really need a single framework that uses the same science.

Additionally, the platform itself – the technology – is critical. Advertisers and agencies want data and science in an advanced tool that empowers the users. For instance, we pride ourselves on being an open platform that has system-to-system integrations with many SEMs and DSPs for faster activation of spend recommendations. That connectivity is key within our industry – and we are the leader in providing a flexible and integrated solution to partners. However, Convertro also provides an intuitive interface that lets users focus on the insights that can move their business forward. This combination of capabilities and streamlining is hugely powerful.

One last key question is “what’s the service like”? Brands should strive for quality service that is helping them with implementation, adoption of technology, change management and strategic recommendations but they shouldn’t pay the same price that they would have paid a consulting firm. Vendors should be lean enough to provide quality service at scale without needing to “throw people” at the problem.

About The Author
David Kaplan David Kaplan @davidakaplan

A New York City-based journalist for over 20 years, David Kaplan is managing editor of A former editor and reporter at AdExchanger, paidContent, Adweek and MediaPost.