Relatively Objective by Jonathan HartDuring the course of my career I've had the privilege to lead some amazing teams which have produced exceptional results for our clients. These are a few of the solutions that I am most proud of because of their impact, scale or novelty.
Texas Water FoundationCreated at Harbinger AI
Psychographic segmentation connected water to the things Texans already cared about, helping Texas Runs on Water become a consistent language for a statewide priority.
The StandardCreated at Meiji Yasuda
Smart Claims Assistant uses AI to connect contract language, employer data feeds, customer portal inputs, phone notes, emails and handwritten documents so complex claims reach an approval decision faster.
HumanaCreated at CINQCARE
Connected medical, social, community and environmental signals so outreach teams could understand the needs shaping each member’s life.
HSBCCreated at TBWA
Enriched digital leads with behavioral and third-party signals, then used AI to give relationship managers a prioritized, actionable briefing for every prospect.
CLEAR ChinaCreated at TBWA
Behavioral data revealed that CLEAR’s largest consumer segment was not a gender audience, but families already sharing the brand at home, leading to new products in market.
EurosportCreated at TBWA
Clustered streaming behavior into dynamic audience segments, giving Eurosport a way to sell media around demonstrated interests rather than generic fan categories.
United States Air ForceCreated at GSD&M
Combined marketing-mix and multi-touch attribution with a personalized recruiting website that adapted its content to each prospective Airman.
AT&TCreated at GSD&M
Used mixed-effects regression to separate creative performance from media cost, rotation and daypart effects, guiding a new direct-response television challenger.
Southwest AirlinesCreated at GSD&M
Machine learning classified customer-submitted Wanna Get Away stories into recurring human moments that informed the next generation of campaign creative.
TargetCreated at Accenture
Unified more than 200 data sources into a marketing measurement and optimization capability spanning departments, stores, campaigns, channels and geographic levels.
VisaCreated at Accenture
Combined time-series regression, diffusion modeling and non-linear optimization to balance growth and harvest investments across products, channels and markets.
Media AuditsCreated at Accenture
GENESIS pooled verified advertiser spend and matched it to Nielsen viewership data, creating benchmarks for media cost and quality by channel, market and global pool.
TargetCreated at Accenture
Target has long been recognized as one of the most innovative and forward thinking marketers in the world. Their sophisticated operation invests over $1B a year to execute 60 campaigns in 40 different channels across 5 geographic levels in service of 2,000 stores with 24 departments each.
With so much complexity, it was difficult to form a unified view of their investments and impacts. The Target CMO selected us as a partner to build out a best-in-class analytics capability to harmonize measurement, then create a data-driven process for channel planning and optimization that would maximize financial returns on this sizable investment.
Our first task was to identify, catalog and collect data from over 200 sources in the organization. With this accomplished, we built a Marketing Mix Model to evaluate the relationship between investments and short-term sales for every dollar on every department, in every store in every week of a three year period.
Deploying this solution took the efforts of more than 50 analytics professionals across 3 continents. Over a period of three years we generated $400M in incremental revenue, $100M in incremental profit and an analytical capability unrivalled in the retail sector.
VisaCreated at Accenture
Visa is the world’s leading electronic funds transfer network, processing more than 100 billion transactions worth over $7 trillion annually. Its marketing operations are funded globally, allocating money from a common pool to individual markets. Economic and regulatory factors cause these markets to differ widely in cash dependency, product availability and transactional-data granularity.
To create a strategic framework capable of balancing these factors, we developed a new measurement model blending time-series regression with Bass diffusion. We also developed innovative approaches for fluctuating exchange rates and reconciling data from different periodicities without losing degrees of freedom.
The approach measured the impact of each channel on each product, in every market, relative to economic maturity and used a non-linear optimization model to balance growth and harvest investments between markets.
The solution was deployed at Visa for more than 10 years and conducted annually across 17 international markets until the acquisition of Visa Europe. It drove annual allocations and gave leadership an unprecedented view of global marketing effectiveness.
Media AuditsCreated at Accenture
After Accenture acquired Media Audits, the business inherited a global operating model built around legacy UNIVERSE systems. Those systems combined client purchasing data with consumer viewing data to evaluate the cost and quality of media buys.
The analytical logic was valuable, but the platform was fragmented. Each country had its own implementation, architecture, support model and business process, reflecting local media markets with different buying rules, from upfront commitments to auction-based bidding. That made a global media-audit practice difficult to operate consistently across markets.
Project GENESIS replatformed the core workflow from BASIC to SAS while preserving existing functionality through parallel validation. The system pooled verified advertiser spend, matched agency data against Nielsen Ad*Views and generated standardized outputs for analysis, workbooks and presentations.
The result gave Accenture Media Management a more stable global platform for assessing media effectiveness across channels, markets and client pools. It turned a country-by-country audit process into common measurement infrastructure for a practice that ultimately represented more than $40B in annual global advertising spend.
Southwest AirlinesCreated at GSD&M
When GSD&M decided to revive Wanna Get Away for Southwest Airlines, more than 50% of their most loyal customers reported having heard the phrase in the last week, despite the campaign having been off the air for 10 years. Few pieces of marketing have entered pop culture in such a sustainable way.
To create the next generation of Wanna Get Away material, we started with the Southwest approach: we listened.
Via Twitter contest, customers were encouraged to submit their own Wanna Get Away stories. With a machine learning algorithm, we analyzed and classified the more than 10,000 responses. It discovered that, while the people and places were different, the same 12 story arcs manifested over and over again.
This insight formed the basis of the new campaign, with each spot based on one of the 12 stories that represents an enduring human moment. Backed by actual customer experiences, the messages became immediately relatable.
United States Air ForceCreated at GSD&M
The all-volunteer United States Air Force was facing its most difficult recruiting period since the Vietnam War ended over 40 years earlier. Above and beyond the already rigorous physical, mental and moral standards, qualifying to become an Airman requires a high degree of technical competency.
To meet the most aggressive goal to date we needed to produce more leads, maintain lead quality and stay within an established appropriation. We reassessed investment across media, events and platforms by developing an analytical engine that built Marketing Mix and Multi-Touch Attribution models in tandem.
Next, we redesigned airforce.com to listen and respond to each unique user using a collaborative filtering algorithm that changed images, headlines, and copy to provide a customized experience. Lastly, we conducted extensive conversion-rate optimization through creative and content testing.
The data-driven allocation, machine-learning powered website, and incredible effort by 1,500 USAF recruiters culminated in a successful mission, welcoming the largest Airman class in recent history.
HSBCCreated at TBWA
Delivering on HSBC’s strong growth agenda for SMB in Asia Pacific required digital channels to deliver increasing acquisition volumes across markets in varying stages of digital transformation. Funnel analysis revealed that although digital lead generation volumes were high, conversions were low, and in some markets not a single lead had converted to a new account.
Following our philosophy of Designing for Data Creation, we re-imagined the SMB website, content strategy and PPC campaigns so interactions explicitly created signals about a prospect’s value, product interests and banking needs. Enriched with public data, this was synthesized using predictive models to guide Relationship Managers in prioritizing leads.
Artificial Intelligence summarized the collected insight into a single-page visual briefing, suggesting conversation pathways and content most likely to result in a new customer.
In Singapore, the business moved from delivering no new clients from digital channels in 2018 to achieving 100% of its annual target within the first six months of 2019.
CLEAR ChinaCreated at TBWA
CLEAR China had organized the shampoo category around the structure of its portfolio: male products and female products. In a haircare market worth roughly RMB 54B in 2018, that view was commercially intuitive, but it left the business blind to a large consumer segment already using the brand in a different way.
Using survey data, we clustered 500 respondents by the brand attributes that appeared to influence purchase decisions. The largest group was neither male nor female as a strategic audience, but married consumers with kids. Once this family segment was separated, the male and female groups also came into clearer focus because they were no longer being conflated with family behavior.
What had looked like a single hierarchy of brand needs became more specific. “Makes you feel confident” appeared to lead overall, but was strongest for families with kids. Once that group was accounted for, male consumers were more clearly oriented around what the product does, nourishing the scalp and moisturizing the hair, while female consumers were more clearly oriented around how it makes them feel, refined and fresh.
The model also separated intended brand associations from the attributes actually associated with purchase. Some attributes the brand had treated as positives were detrimental for specific groups: men were turned off by “make your hair soft and smooth,” while women were turned off by “used by professionals.” Because families were sharing the product, the finding also created a packaging opportunity around larger unit sizes.
AT&TCreated at GSD&M
AT&T’s Home Solutions division operated a massive Direct Response Television program used to generate product inquiries for its call center and e-commerce storefront. We developed an extensive testing methodology culminating in an in-market “Champion/Challenger” test.
To become Champion, new creative went head-to-head against the current winner for five weeks. The approach minimized client risk and created substantial improvements for years, but diminishing marginal returns eventually took their toll and 14 consecutive Challengers failed to win.
Deeper analysis showed that fluctuating media costs, different rotation weights and daypart mixes profoundly affected outcomes. We used a mixed-effects regression model to control for those factors while identifying which characteristics top-performing creative had in common.
Under the model’s guidance, we constructed a new Challenger. It outperformed the Champion in every week of the test, lowering Cost per Call by as much as 24% in a single week and 12% overall.
EurosportCreated at TBWA
Eurosport entered the 2018 Winter Olympics facing a difficult subscription challenge: much of the audience could watch Olympic coverage for free through traditional broadcast options, while Eurosport needed to prove the value of a paid digital streaming experience.
The larger opportunity was not simply to find “ski fans.” Streaming behavior created a much richer picture of how people actually watched the Games: which events they followed, which combinations of sports suggested deeper intent, and which moments created the strongest reason to return.
By clustering this behavioral data, we identified more than 10,000 dynamic audience segments and used them to shape both the consumer experience and the commercial offer. Viewers could receive more relevant clips, highlights and recommendations, while media buyers could buy audiences aligned to observable interests rather than broad demographic or sport-level assumptions.
This created a more valuable media product around the Olympics. The same system that helped drive 500,000 new subscribers during the 18-day Winter Games also gave Eurosport a clearer basis for charging a premium to advertisers seeking more precise, behaviorally defined audiences.
Texas Water FoundationCreated at Harbinger AI
Water has always been essential to Texas, but for years it was politically quiet. In an early conversation with the Texas Water Development Board, the problem was stated plainly. Talking about water was not going to get anyone reelected.
Texas Water Foundation needed a way to make water feel immediate across a state with different regions, priorities and political instincts. Harbinger AI created psychographic segments from survey data to understand how Texans related to water, growth, conservation and responsibility.
The work identified five distinct groups and a small set of questions that could classify someone into those groups with 90% accuracy. That gave the campaign a repeatable way to connect water to jobs, health, cost, growth, recreation, pride and the future of local communities.
The segmentation helped shape the strategic foundation for Texas Runs on Water, giving the campaign consistency without forcing every Texan into the same message. The public policy story continued to grow, culminating in constitutionally dedicated water funding over 20 years through the Texas Water Fund, approved by 71% of voters in November 2025.
HumanaCreated at CINQCARE
CINQCARE was building a care model for members whose health needs could not be understood through medical data alone. For many people in this population, the conditions and social needs shaping their lives were undocumented, which made them effectively invisible to traditional population health methods.
As Founding AI/ML Engineer, I designed and built a culturally aware data and model store designed to give CINQCARE a fuller understanding of Black communities and the individuals it served. The platform combined medical history, white-labeled health and social risk intelligence, third party consumer data, psychographics and community signals, then used predictive models to infer likely ICD-10 and SDOH codes.
That intelligence was used to personalize outreach scripts for 40,000 Humana members. Instead of treating every person as the same clinical opportunity, the system gave outreach teams language shaped by each member’s likely needs, barriers and motivations.
The work increased program enrollment and created an estimated $22M in annual recurring revenue. It also established a practical use for culturally aware AI inside care delivery, not as an abstract model, but as a way to help the right conversation happen sooner.
Your message has been sent.