Research Interests & Future Direction
Three papers are the starting point. This is where the research is going.
The three papers currently under peer review are not the destination | they are the first formal outputs of a much larger research program that has been forming since the decision to pursue MPhil/MS Data Science (AI) in 2025.
This page documents the complete research direction | every active and planned area of investigation, how each connects to the Cognitive Marketing Engine, and | critically | which research directions are being built into actual SaaS products, not just published as papers.
This is not a wish list. This is the roadmap.
The Unifying Thread
Every research interest below answers some version of the same question:
“Where is marketing still relying on guesswork, and what mathematical or computational approach can replace that guesswork with evidence?”
Five domains have emerged as the focus areas where this question is most urgent | and most underexplored:
1. Predictive & Causal Marketing Intelligence
2. NLP & Content Integrity for Marketing
3. AEO / GEO / LLMO | Marketing in the AI Search Era
4. Computational Marketing Infrastructure (CME Extensions)
5. Marketing-Focused SaaS Product ResearchEach is explored below.
Domain 1 | Predictive & Causal Marketing Intelligence
Extending the foundation laid by the first three papers
The first three papers established a foundation in lead scoring, sequential behavior modeling, and multi-touch attribution. The next research questions extend each of these in directions that current literature has not fully addressed:
Cross-Platform Propensity Transfer Learning
Can a propensity model trained on one client’s historical data | in one industry | be partially transferred to a new client in a related industry, reducing the “cold start” problem that makes early-stage ML modeling difficult for smaller businesses?
This question directly addresses a practical limitation of Loop 1 and Loop 4 of the Cognitive Marketing Engine | smaller clients often don’t have enough historical data for robust model training. Transfer learning research could make computational marketing accessible to businesses currently too small for it.
Causal Attribution Under Privacy Constraints
As cookie-based and device-level tracking continues to degrade (iOS privacy changes, browser restrictions, regulatory pressure), how can causal attribution models | like those explored in Paper 3 | be adapted to function reliably using aggregated, privacy-safe data only?
This is directly relevant to Loop 2’s Bayesian Media Mix Modeling approach | MMM is inherently more privacy-resilient than device-level attribution, but the research into optimal MMM configuration for small-to-mid-size businesses (not just enterprise) is underdeveloped.
Multi-Industry Customer Lifetime Value Benchmarking
Paper 2 applied deep learning to ecommerce CLV prediction. A natural extension: how do CLV prediction models need to be adapted across different industries | subscription SaaS versus DTC ecommerce versus B2B services | where purchase frequency, contract structures, and churn dynamics differ fundamentally?
Domain 2 | NLP & Content Integrity for Marketing
Where marketing content meets responsible AI
As AI-generated marketing content becomes ubiquitous, a new class of problems is emerging | problems that traditional marketing tools were never designed to catch.
Pre-Publication Content Integrity Detection
Marketing copy increasingly uses psychological persuasion techniques | some legitimate, some crossing into manipulative “dark patterns” (false urgency, hidden costs, manipulated defaults). As AI tools make it trivially easy to generate large volumes of persuasive copy, how can NLP models | using transformer architectures like BERT | automatically flag content that may violate platform policies (Google Ads, Meta Ads policies) or constitute manipulative dark patterns before publication?
This research direction has direct commercial application: a pre-publication integrity checker that scans ad copy, landing pages, and email content for policy-violation risk and dark-pattern language before it goes live | preventing account suspensions, ad disapprovals, and reputational risk.
Brand Voice Consistency at Scale
For businesses producing large volumes of content across multiple platforms (the exact situation of the Marketing Intelligence Lab itself, and many ecommerce/DTC clients), can NLP models be trained to detect brand voice drift | content that technically follows guidelines but has gradually drifted in tone, vocabulary, or framing from the established brand voice?
Semantic Content Decay Detection at Scale
Building on the LDA-based content decay detection already used in Loop 1 | how can this be automated and scaled to continuously monitor an entire content library (hundreds or thousands of pages) and flag content whose semantic alignment with current search intent has decayed, before rankings actually drop?
Domain 3 | AEO, GEO, and LLMO | Marketing in the AI Search Era
The next SERP doesn’t look like Google’s SERP
This is one of the most urgent and underexplored research areas in digital marketing today | and one with almost no rigorous academic literature yet.
The Shift:
Users increasingly get answers directly from AI systems | ChatGPT, Perplexity, Google’s AI Overviews, Claude | rather than clicking through to websites. This creates three interconnected new disciplines:
- AEO (Answer Engine Optimization) | optimizing content to be selected as the direct answer in AI-generated responses
- GEO (Generative Engine Optimization) | optimizing for how generative AI systems synthesize and cite information across multiple sources
- LLMO (Large Language Model Optimization) | understanding how LLMs represent, weight, and retrieve information about brands and topics in their training and retrieval processes
Research Questions Being Explored:
Citation Pattern Analysis
When AI systems answer questions and cite sources, what structural, semantic, and authority signals correlate with being cited | versus simply being indexed? This is fundamentally different from traditional ranking factor research, because the “ranking” is happening inside a model’s reasoning process, not a traditional algorithm.
Semantic Authority Mapping for LLMO
Traditional SEO authority is measured through backlinks and domain metrics. What does “authority” mean to an LLM | and can content be structured to increase the likelihood that an LLM retrieves and accurately represents it when answering related queries?
AEO Content Structuring Using Transformer Analysis
Using the same SBERT and transformer-based techniques already applied in Loop 1 for search intent drift | can content be analyzed and restructured specifically to increase its retrievability and citation probability in AI-generated answers, while maintaining quality for human readers?
Why This Matters Now:
This is not a future concern | it is happening now, and almost no marketing practitioners have rigorous frameworks for it. Early research in this area, backed by the same computational rigor applied to traditional SEO, represents a significant first-mover research opportunity | and a direct extension of the Cognitive Marketing Engine’s diagnostic philosophy applied to an entirely new “search” environment.
Domain 4 | Computational Marketing Infrastructure | CME Extensions
Making the framework itself a subject of research
The Cognitive Marketing Engine is not a finished product | it is a living framework that improves as research progresses. Several open questions about the framework’s own architecture are active research interests:
Optimal Retraining Frequency Under Concept Drift
Loop 4 currently uses monthly retraining as a standard cycle. But is monthly optimal for every model type and every industry? Research into drift detection metrics that could trigger adaptive retraining schedules | more frequent for fast-moving industries (fashion, ecommerce flash sales), less frequent for slower-moving ones (B2B, real estate) | would make Loop 4 more efficient and effective.
Multi-Objective Portfolio Optimization Beyond Markowitz
Loop 2 currently applies Markowitz Efficient Frontier for budget allocation | a model originally designed for financial portfolios with a single risk-return tradeoff. Marketing budget allocation often has multiple competing objectives simultaneously | revenue, brand awareness, customer acquisition cost, and lifetime value. Research into multi-objective optimization techniques (beyond simple Markowitz) that can balance these competing objectives mathematically is an active interest.
Explainability in Causal Marketing Models
As CausalML and Bayesian models become more central to client strategy (Loop 2), how can the outputs of these models be explained to non-technical stakeholders in a way that builds genuine understanding | not just trust based on “the AI said so”? This connects directly to the practice’s core value of “data over opinion” | clients need to understand why, not just what.
Domain 5 | Marketing-Focused SaaS Product Research
Where research becomes a product
This domain is different from the others | it is where academic research is deliberately being designed for commercialization.
The mission has always included building Micro SaaS products that make advanced marketing intelligence accessible to businesses that cannot afford enterprise data science teams. Every SaaS product in the pipeline is directly derived from a research question above | not built independently of the research program.
Product 1 | Google Ads Waste Spend Detector (Target: 2027)
Research foundation: Domain 1 (Causal Attribution) + Domain 4 (CME extensions)
A tool that applies Isolation Forest anomaly detection and causal incremental lift analysis | the same techniques used in Loop 1 and Loop 2 of the CME | to a business’s own Google Ads account, identifying spend that is not generating true incremental results, even when platform-reported metrics (ROAS, CTR) look healthy.
This is Loop 1 and Loop 2 of the Cognitive Marketing Engine, packaged as a self-service tool for businesses too small to engage the full CME process.
Product 2 | Pre-Publication Content Integrity Checker (Research-Stage)
Research foundation: Domain 2 (NLP & Content Integrity)
Building directly on the NLP-based content integrity research | a tool that scans ad copy, landing pages, and marketing emails for platform-policy-violation risk and dark-pattern language before publication, using transformer-based classification models.
Product 3 | AEO/GEO Content Audit Tool (Concept-Stage)
Research foundation: Domain 3 (AEO/GEO/LLMO)
As research into AI-search citation patterns matures, a tool that audits a website’s content library and scores each page’s likelihood of being cited or retrieved by AI answer engines | with specific restructuring recommendations based on the semantic authority research findings.
The Pattern:
Research Question (Academic Rigor)
↓
Validated on Real Client Data (CME Application)
↓
Packaged as Self-Service SaaS Tool
↓
Accessible to Businesses Beyond Direct Client BaseThis is the mechanism by which the mission | making advanced marketing intelligence affordable globally | actually gets executed. Not through scaling a consulting practice indefinitely (which has natural limits), but through productizing the research itself.
How This Connects to the Long-Term Vision
Every domain above feeds into the same long-term picture:
PhD Track (2027 onwards): The strongest research threads from these five domains | particularly AEO/GEO/LLMO (currently almost unexplored academically) and causal attribution under privacy constraints | are strong candidates for PhD-level research, given how underexplored they currently are in academic literature.
Global Practice: As CME applications across these domains are validated with international clients (UK, USA, UAE), the research gains real-world validation across multiple markets | strengthening both the academic contributions and the commercial products.
Micro SaaS → Technology Company: Each SaaS product above is designed to operate independently, but all three share underlying infrastructure (data pipelines, ML model architecture, client onboarding). Over time, these products form the foundation of a technology company | built entirely on original research, validated on real client data, before ever being offered as standalone software.
The Honest Summary
Most “research interests” pages are aspirational | a list of topics someone finds interesting, disconnected from anything they’re actually doing.
This page describes work that is already in progress | three papers under review, active coursework directly informing client strategy, and SaaS product concepts already mapped to specific research domains.
The research interests aren’t separate from the practice. They are the next 5-10 years of the practice, already planned.
Want to see where this research started?
→ The Three Papers Under Review (link to /research/papers)
→ The Academic Journey (link to /research/ms-ds-journey)
→ How Research Becomes Client Strategy (link to /approach/my-framework)