Strategy · Marketing · AI
India → Boston
TANVI
Marketing Strategy · Data & AI · Brand Growth
Hover to reveal. Scroll to begin.
Business Development·AI Engine Optimization·Marketing Analytics·Python & R·Neuromarketing·HubSpot & NetSuite·B2B Strategy·Predictive Modeling·International Athlete·
Business Development·AI Engine Optimization·Marketing Analytics·Python & R·Neuromarketing·HubSpot & NetSuite·B2B Strategy·Predictive Modeling·International Athlete·
Strategy & Growth
A tech-marketing strategist who reads a dataset and a room with the same instinct.
I'm Tanvi. I grew up in Pune, the Oxford of the East, and now work out of Boston, the Athens of America. Apparently I have a type.
I work at the intersection of data, marketing, and AI. Right now that's a Business Development Manager role: growth and partnerships across the entire US, fifteen West Coast states of my own, and everything that comes with that, from marketing and content to analysis, sales, and AI-visibility strategy. Before this, my path ran through international sports, life sciences, and a startup I helped build from zero to profit, plus a STEM Master's in Marketing.
What I have gained has been unusually broad. Before I came to Boston I ran an entire commercial function, with a team of eight alongside me, across three residential and commercial projects, running the full marketing and sales engine.
None of it was a detour. What connects it all is how I think: understand the system first, find where the real leverage is, then execute. That thread runs through every page here.
37%
YoY Sales Growth
46%
Conversion Lift
1,200+
Enrollments Driven
2×
International Gold
Where to start
Pick a thread.
01 / Story
The Career →
From Pune to Boston, told in eight chapters.
02 / Projects
The Research →
Twenty-two studies with a finding attached to each one.
03 / AI
AI & Strategy →
My own study on how AI assistants pick brands.
04 / Record
Track Record →
The numbers, the range, and where the discipline started.
Who I Am
The long version
of a short answer.
Most marketing people haven't won international medals in Taekwondo, run clinical research in a hospital, produced a documentary, and closed $3M in real estate sales before specializing in AI. I have, and the order matters. Each one handed me something the next one turned out to need.
Sports came first, at five, and never really left, and it gave me the operating system I still run on: stay composed when things go wrong, recover fast, and show up long before the results do.
Science came next. A Microbiology focus taught me to think in systems and trust evidence over noise. Then came years of deliberate range: UN-recognized NGO leadership, event management, theatre, interpretation. Different rooms, different languages, one constant, which was paying attention to how people actually make decisions.
It points somewhere specific now. Marketing that starts with evidence and ends with an idea worth acting on. The eight chapters below are how it got there.
Strategy & Growth
Business Development (B2B & B2C)Advanced
Marketing & Brand StrategyAdvanced
Partnership DevelopmentAdvanced
KPI Design & Exec ReportingAdvanced
Data & AI
AI Engine Optimization (AEO)Advanced
Marketing Analytics (Python, R, SAS)Advanced
Neuromarketing & BehaviorAdvanced
iMotions Biosensor ResearchCertified
Predictive Modeling & SegmentationAdvanced
Tools & Platforms
HubSpot CRMAdvanced
NetSuite ERPAdvanced
Google Analytics (GA4)Certified
IBM SPSS · Atlas.ti · IllustratorProficient
Human Skills & Languages
Cross-functional LeadershipAdvanced
Executive CommunicationAdvanced
Stakeholder & Client RelationsAdvanced
EnglishNative
MarathiNative
HindiNative
The Timeline
Every chapter
was a choice.
A non-linear path, built on purpose. Each step added a different lens, and they all point in the same direction.
2013-2020
01
"It's not who I am underneath, but what I do that defines me."
The Undergraduate Years: Building in Every Direction
Pune University · UN / APD · Alliance for Global Education · MNC Events
My path didn't start in a classroom. It started in 2013, when at sixteen I became one of the youngest board members of Action for Pune Development, an NGO later honored in the Limca Book of Records. Over the next several years I led 50+ social missions, mobilized 2,500 students, and served as a UN Youth Ambassador for global campaigns. Alongside my BSc in Zoology, with a Microbiology focus, I worked as an interpreter, translator, and research intern with the Alliance for Global Education, running clinical research at KEM Hospital and producing a documentary on microfinance. I was also a lead organizer at MNC Events, executing State Bank of India events, heritage walks, and an ultra-cycling RAAM qualifier. I never treated college as a single track, and working across cultures under real stakes while still a student taught me more about working with people than any class did.
2018-May 2020
02
"Anyone can do my job, but no one can be me."
Fitness Science & My First Business
K11 School of Fitness Sciences · Y1 Fitness
I earned a Personal Training diploma from K11, India's leading fitness-science academy, ranking 2nd in class, with a REPs International License valid in 10 countries. From there I helped build Y1 Fitness, a sports-psychology and nutrition startup, as Head of Business Development, driving 58% profit in its first year through 50+ trainer partnerships and 70+ clients counseled. It was my first real taste of building a business engine from zero.
Jun 2020-Jun 2021
03
"Logic is the beginning of wisdom, not the end."
Virtual Technical Support Specialist
Amazon · Pune, India
I joined one of the world's most operationally rigorous companies during the pandemic. I analyzed customer usage and product-performance data for Amazon Devices, applied structured frameworks like PESTLE and SWOT for internal reporting, and finished top three on my team in my first month's performance review. It's where I learned process discipline and data precision at a global scale. This was also the era I became an early adopter of generative AI tools, building them into how I researched and worked right as they first emerged.
Jul 2021-Mar 2022
04
"Just because something works doesn't mean it cannot be improved."
Y1 Fitness: Return & Relaunch
Y1 Fitness
After Amazon, I came back to Y1 Fitness to stabilize operations and relaunch growth. When something you helped build needs you, you show up. I picked up where I had left off, re-engaged our trainer network, and continued driving business development through the full lifecycle of the startup.
Apr 2022-Aug 2023
05
"I'm gonna make him an offer he can't refuse."
Marketing & Sales Head, Business Development
Kemse Constructions (formerly Sai Associates) · Pune, India
This is where strategy met serious stakes: ₹24 crore in sales, about $3M, and 37% year-over-year growth. Indian property prices run far below US ones, so the rupee figure is the truer measure of scale. I led the full commercial engine, marketing, sales strategy, and buyer analytics, for three high-value developments, managing an 8-member cross-functional team. I drove a 40% lift in brand awareness and 25% in revenue, maintained 500+ client relationships, and was named Employee of the Month for three consecutive months. I was also an early adopter of ChatGPT here, using it to sharpen marketing and research workflows when the tools were brand new.
Sep 2023-Dec 2024
06
"I don't have dreams. I have goals."
MS Marketing (STEM): Intuition Meets Precision
Suffolk University, Sawyer Business School · Boston, MA
GPA 3.8/4, a $20,000 scholarship, Graduate Fellow at X-Lab, and RA for the Sports Management Program. I came here to model markets, not just move them: neuromarketing, machine learning in Python, advanced analytics in SAS and R, behavioral segmentation. But the coursework was only half of it. Suffolk put me in rooms I would not otherwise have been in, working on live briefs with real companies, learning directly from professors who had run the functions they were teaching, and building a network across Boston's business and sports scene. My professors pushed me to use AI efficiently without becoming reliant on it. I left fluent in the science, confident with the tools, and comfortable in the room.
Sep 2024-Feb 2025
07
"Sometimes you have to run before you can walk."
Marketing Partnerships & Analytics Lead
Dream Venture Labs (Vylero) · Boston, MA
I designed and launched 3 influencer and university partnership campaigns that drove 1,200+ enrollments in 5 months, using A/B testing and campaign analysis to lift campaign performance by 18%. It was a live startup proving ground for applying graduate-level analytics, and AI tooling, to real growth.
Feb 2025-Present
08
"Make it simple, but significant."
Business Development Manager
Catholic Purchasing Services · Newton, MA
This is where every thread converges. I run business development across the entire US, with fifteen West Coast states as my own territory, working both sides of the market: developing vendor partnerships and winning the client relationships those vendors serve. Growth across acquisition, retention, and cross-selling has lifted conversion rates by 46%. Alongside that I handle marketing, including content creation, campaign analysis, executive reporting, RFPs, vendor workflows across NetSuite ERP and HubSpot CRM, plus a share of direct sales. I also lead our AI-visibility strategy, researching how the organization surfaces in AI-driven search, then driving hands-on adoption across the team. Strategy, data, marketing, sales, AI, and operations, in one role.
How I Think
Different fields,
one operating system.
The skills get you in the room. How you think decides what happens in it.
01 /
I map the system first.
That means inputs, dependencies, and what breaks if it changes, all before I touch the task. That's why my work compounds instead of just getting done.
02 /
Analytics with instinct.
Trained in predictive modeling and neuromarketing, and seasoned by years of reading people as a counselor, interpreter, and coach, my instinct usually flags something first, and the data either backs it up or corrects it.
03 /
AI as leverage, not a crutch.
I use AI to move faster and think bigger, but the judgment stays mine. The failure I watch for is outsourcing the thinking and calling it efficiency.
04 /
I'd rather be wrong early.
I run the cheap version of the test first. It costs less to find out early, and by the time the decision actually matters, I'm working from evidence instead of instinct.
05 /
Early ownership.
From running UN youth campaigns as a student to leading AI strategy for an organization, I don't wait for permission to take responsibility. Sports taught me that hesitation is the most expensive move there is.
06 /
Impact over motion.
I keep asking which part of this actually moves the number. When my own work isn't earning its place, I'm comfortable cutting it and starting again.
Academic & Research Projects
Real data.
Real names.
Not classroom hypotheticals.
Real organisations, production tools, and a finding at the end of each one.
These are the ones I can show. The client and company work stays where it belongs.
Filter by method to see how I work.
22
Projects Shown
7
Analytical Tools
13
Sectors
18k+
Records Analyzed
X-Lab · Suffolk University · Paid Research Role
I've measured what people won't tell you.
Most marketers researching consumer behavior have run a survey. I spent my Master's as a paid Graduate Fellow at X-Lab, the first neuromarketing lab in Massachusetts. I set up the instrumentation, ran participant sessions on the iMotions biosensor stack, and worked alongside the lab's director on the analysis, across the lab's own studies and research for external clients.
Surveys capture what people are willing to say about a product. Eye tracking and skin conductance capture what happened before they decided what to say.
Surveys capture what people are willing to say about a product. Eye tracking and skin conductance capture what happened before they decided what to say.
Eye Tracking
Facial Expression Analysis
Skin Conductance
Brain Imaging
iMotions Certified
X-Lab is directed by Dr. Mujde Yuksel, who teaches the graduate neuromarketing course at Sawyer Business School.
Start here
Original Research
AI Visibility Audit: How Assistants Pick Brands
Built a 20-question prompt library across four buying stages and ran it, logged out. 23 answers, 55 brands, 69 mentions. Brands are named in 7 of 7 comparison questions and 1 of 5 problem-aware ones. No brand holds double-digit share. And a third of the time, the source cited for a recommendation is the brand's own website.
Study Design · AEO · AnalysisAug-Sep 2026
Start here
Machine Learning · Causal Inference
Uplift Model: Who Buys Because You Contacted Them
Built a two-model uplift estimator from scratch in NumPy, no libraries. It answers the question marketing budget actually turns on: not who will buy, but who buys only because you reached them. Validated against known ground truth at r = 0.99. It put zero value-destroying customers in its target list. A conventional response model put in 7.2%.
Python · NumPy · Causal InferenceSep 2026
Start here
Neuromarketing · X-Lab
Amusement Parks: What the Eye Does Before the Mouth Speaks
A between-subjects biosensor experiment in Suffolk's X-Lab: 34 participants viewed amusement park images with or without people while eye tracking, facial coding and skin response ran. Participants locked onto the crowded park four times faster, the only significant result. Nothing they self-reported reached significance.
iMotions · Eye Tracking · GSRMar-May 2024
Framework · AEO
The AEO Diagnostic
A six-step diagnostic anyone can run their own brand through: build a prompt library, run it clean, score what comes back against published benchmarks, then find out why the brands that surface do. Ends with a fix order, and with what the method cannot tell you.
AEO · Framework DesignSep 2026
Research Methods
AI-Driven Interview Research, ARTES
Used AI to run structured interviews against defined consumer profiles, then coded every response by topic, code and theme the way you would a human transcript. Produced two product recommendations, and named the method's own limits.
AI Interviewing · Thematic CodingOct-Nov 2024
Startup Strategy · Professional
Market Research, Vylero
Market analysis for a live startup on where AI was reshaping project management: predictive risk forecasting, automation, and the shift to flexible team assembly in a growing gig economy.
Market Research · Competitive Analysis2024
Product Strategy · Market Entry
Market Entry Analysis, Clip Mouse
Assessed a gesture-controlled wearable mouse for entry into the IT peripherals market: the technology, the competitive field, adoption barriers, and whether the position was defensible.
Market Entry · Competitive Analysis2023
Brand Strategy · Case Analysis
Brand & Growth Case Studies
A set of strategy case analyses across categories: Entomo Farms (cricket protein facing a new entrant), Tata Steel, Steinway, Lexmark, Supreme and Gillette.
Case Analysis · Strategy2023-2024
Machine Learning · Python
Supervised Modeling of Wellbeing & Listening Behavior
A complete scikit-learn pipeline on survey data: engineered a binary target, trained a logistic regression, then read the model through odds ratios. Against a 72.15% baseline it reached 100% precision at 20% recall on a 20-row test set. The odds-ratio reading is the point, not the score.
Python · scikit-learn · pandasNov-Dec 2024
Sports · Fan Strategy
Gen Z & Gen Alpha Engagement, Boston Celtics
Surveyed 380 respondents and ran in-depth interviews to find why younger fans follow the Celtics online but don't attend. The gap wasn't interest, it was format: they consume the team through short-form video rather than broadcast. Recommended exclusive TikTok and YouTube content, and flagged India as untapped.
IBM SPSS · Survey · IDIOct-Dec 2023
Sports · Market Analysis
Sports Betting in India: Market Trend Analysis
Mapped a market where legal and illegal turnover runs to an estimated $130-150B, with roughly $200M moving during a single ODI match, while legal online betting exists in just three states. The commercial opening is fantasy sports, legal nationwide and lowest on regulatory risk.
Qualitative · Secondary ResearchOct-Nov 2023
Public Transit · Qualitative
In-Depth Rider Research, MBTA Boston
Fourteen in-depth interviews, a focus group of six and on-field observation across three lines, all coded in Atlas.ti. Built to test how rider satisfaction shifts across the winter season and what actually drives the decision to ride.
Atlas.ti · Focus Groups · IDISep-Dec 2023
Streaming · Segmentation
Customer Segmentation, Netflix User Behavior
Clustered roughly 15,000 subscribers using PROC FASTCLUS into five segments. The useful result was a negative one: sci-fi affinity was near-identical across all five clusters (0.70 to 0.71) while viewing frequency ranged 1.75 to 4.33. Device and habit segment this audience. Genre does not.
SAS · Cluster AnalysisMar 2024
Travel · Paid Search
Agoda vs Marriott: OTA vs Hotel Performance
Regression on paid search found Agoda converting at 2.71% CTR against Marriott's 0.36%, roughly seven times better. In both models higher bids and ad-quality scores carried significant negative coefficients, so the intermediary was winning on intent capture rather than spend.
SAS · RegressionApr-May 2024
Sports · Sponsorship Research
Fan Profiling, New England Free Jacks
Built the fan-profiling study for a Major League Rugby club: professional background, purchasing influence and media habits, to arm sponsorship conversations. The question the client most wanted answered was why young professional women are converting to rugby faster than any other segment.
Survey Design · AnalysisSep-Dec 2024
Art Commerce · Consulting
Market Strategy & Competitor Analysis, ARTES
Two research streams for one client: social listening across online art-collecting communities, coded thread by thread, plus a competitive benchmark against Artsy. Delivered competitor analysis, qualitative content analysis and a pitch deck.
Social Listening · Competitive AnalysisSep-Dec 2024
Hospitality · Modeling
Predictive Analytics for Customer Loyalty
Modeled 1,500 guest records across 25 variables, 18 of them satisfaction measures to isolate which ones actually predict overall satisfaction, testing variable transformations rather than assuming the raw scales were usable as collected.
R · Predictive ModelingApr 2024
Real Estate · Go-to-Market
Marketing Research & SWOT, Surajnagar Site
Full go-to-market plan for a mixed residential and commercial development in Pune: segment and competitor analysis, positioning built around luxury and community living, branding and promotion tactics, the sales and distribution plan, and budget allocation.
SWOT · Market ResearchApr-Jul 2023
Retail · Behavioral
Price, Promotion & Market Penetration Across Seven Markets
Analyzed 2,080 weekly observations across two competing products in 20 stores and 7 countries, comparing price and promotion effects on sales by store and by country to find where penetration strategy should differ by market.
R · Behavioral AnalysisMar 2024
Hospitality · Retention
Customer Segmentation & Retention Tactics
Segmented the same hotel guest base by satisfaction profile rather than modeling it, turning the prediction work into retention tactics aimed at the segments most likely to churn.
R · SegmentationMay 2024
Health Tech · UX
VR Fitness: Consumer Journey & Empathy Mapping
Empathy and journey mapping for a VR fitness platform, built around a defined persona and tracked by emotional state across awareness, consideration, decision and retention. Found the drop-off points sit at information overload, setup cost and technical friction.
UX · Journey MappingApr-May 2024
Real Estate · Growth
Digital Marketing Growth, Kemse / Sai Associates
Digital growth and research program for a live property developer, taking the academic analytics work back into a business I had run commercial strategy for.
Digital AnalyticsNov-Dec 2023
Three sections. Pick one to open it.
Before the boardroom,
there was the mat.
I started playing at five, and I haven't stopped since. Long before any boardroom or dataset, sports taught me how to perform under pressure, recover from a loss, and outwork the room. It's the foundation everything else is built on, and it's still how I operate.
19 Medals · 2× International Gold · Black Belt
Taekwondo
Two international gold medals. Seventeen more across state, district, and national competitions. Champion of Champions title as a junior, Black Belt, and a certified National Referee. Nineteen medals in total. The lessons were infinite.
State Level · Nationals
Athletics
Long jump, sprinting, and the 4×100 relay, competing at State level and Nationals. Pure speed, power, and the discipline of incremental gains.
District & State Level
Cricket & Baseball
District-level cricket as batsman and wicket-keeper, plus the Pune University baseball team's playing nine, competing at district and state level. Team sports, read under pressure.
Gold
International Open Taekwondo
Sub-Junior Lightweight
Gold
Champion of Champions
Sub-Junior Taekwondo
⬛ Black Belt
Highest Honor
National Referee
🏃 Athletics
Long Jump · Sprint · Relay
State · Nationals
⚾ Baseball
University Playing 9
District · State
🏏 Cricket
Batsman & Wicket Keeper
District Level
Sports isn't a line on my résumé. It's the source code.
The composure to present to a room full of executives, the discipline to learn a new domain from scratch, the instinct to read a situation before the first move: all of it traces back here.
The experiences that
built the perspective.
Most of these ran alongside my studies. They don't fit on a one-page résumé, but they're a real part of why I see problems the way I do, and why I'm comfortable across so many kinds of rooms.
Social Impact · United Nations
Nine Years on an NGO Board
I joined the board of Action for Pune Development at sixteen, in 2013, and stepped down at the end of 2022 when I moved to Boston. The role needed someone on the ground. The NGO holds a Limca Book of Records entry for outstanding contribution, and I was part of more than 100 of its activities across traffic safety, civic awareness and education. The UN work came through it: Youth Ambassador for MY WORLD 2015 and World We Want, and 1,000 signatures gathered for Save Kids Lives, which the UN certified.
100+ activities · 1,000 signatures · 9 years on the board
Research · Interpretation · Media
Research Intern, Interpreter & Documentary Producer
With the Alliance for Global Education, I bridged US exchange students and Indian institutions: interpreting, translating, running clinical research at KEM Hospital, and producing a documentary on microfinance as director, producer, and voiceover artist. Cross-cultural communication, lived daily.
Clinical research · Translation · Film production
Event Management
Lead Organizer, Corporate & Sports Events
Lead organizer at MNC Events: running events for State Bank of India, leading heritage walks for Symbiosis MBA students, and serving on the core team for Deccan Cliffhanger, an ultra-cycling RAAM qualifier. Logistics, stakeholders, and pressure, at scale.
SBI events · Heritage walks · National race ops
Theatre & Performance
Actor, Director & Screenplay Writer
Competed in Maharashtra's most prestigious theatre competitions (Purushottam & Firodiya Karandak) as actor, lighting designer, screenplay writer, and assistant director. It's where I learned storytelling, stage presence, and how to hold an audience, skills that quietly power every presentation I give now.
Acting · Direction · Screenwriting · Stagecraft
Four pieces of work where I owned the outcome. Each one is a system I built and left running.
Real Estate · Business Development
37% YoY Growth, $3M Closed
Owned commercial strategy end to end across three high-value developments at Kemse Constructions and personally closed ₹24 crore, about $3M, of the sales (home prices in India run well below US ones, so that covers far more property than the dollar figure suggests). Built the targeting from regional demand and buyer analytics, and ran the multi-channel campaigns with a team of eight. Named Employee of the Month three months running.
↑ 40% brand awareness · ↑ 25% revenue · 500+ client relationships
AI Strategy · B2B Growth
46% Conversion Lift & Hands-On AI Strategy
At Catholic Purchasing Services, I rebuilt the acquisition and cross-sell engine with data-driven segmentation and HubSpot workflows across 15 states, delivering $1M+ in annual sales, while researching and developing an AI Engine Optimization strategy to keep the organization visible as buyers shift to AI-driven search.
$1M+ annual sales · 46% conversion improvement · AI visibility strategy
Partnership Marketing · Analytics
1,200+ Enrollments in 5 Months
Three partnership channels, influencer and university, A/B tested against each other until the winning combination was clear. Performance rose 18% as enrollment volume scaled. Left behind a system the team could rerun.
1,200+ enrollments · +18% campaign performance · 3 campaigns
Startup Growth · Sports Science
58% Profit in Year One, From Zero
As founding business-development lead at Y1 Fitness, I built the partner network of 50+ trainers, established the revenue model, and drove 58% profit in year one, while counseling 70+ clients and finishing my undergrad.
58% profit · 50+ partnerships · 70+ clients counseled
Credentials
Where the rigor
came from.
A through-line from life sciences to marketing science, each stage adding a layer of how I analyze, test, and decide.
Certifications
Google AI Essentials, Google
AI for Marketing, HubSpot
iMotions Academy Certificate, iMotions
Google Analytics (GA4), Google
Google Ads Search, Google
Research Involving Human Subjects, A+, CITI Program
Green Revolution, Grade A, United Nations ICCE
Global Certification Program, Grade A+, United Nations ICCE
CPT-EFS, Ministry of Skill Development, India
Degrees & Diplomas
Sep 2023-Dec 2024 · Boston, MA
MS, Marketing (STEM)
Suffolk University · Sawyer Business School
GPA 3.8/4 · $20,000 scholarship · Graduate Fellow, X-Lab · RA for the Sports Management Program · Secretary, Graduate Finance & Entrepreneurship Club.
Coursework: Neuromarketing, Machine Learning for Marketing (Python), Advanced Digital Analytics (SAS), Advanced Marketing Analytics (R), Qualitative Research, Customer Centricity, Global Perspectives for Consumers.
Coursework: Neuromarketing, Machine Learning for Marketing (Python), Advanced Digital Analytics (SAS), Advanced Marketing Analytics (R), Qualitative Research, Customer Centricity, Global Perspectives for Consumers.
Jun 2016-May 2019 · Pune, India
BSc, Zoology (Major), Microbiology (Minor)
Savitribai Phule Pune University
First Class degree with first-year Distinction. My grounding in microbiology and the scientific method, hypothesis, testing, evidence, is where my analytical instincts started. Alongside academics: university baseball, competitive theatre, NGO leadership, and research internships.
2019 · Pune, India
PG Diploma, Personal Training
K11 School of Fitness Sciences
GPA 3.8/4, ranked 2nd in class. K11 was founded by Kaizzad Capadia, India's first personal trainer. REPs International License valid in 10 countries, with PD Approval and SPECL (Skill India) certification.
Dec 2019-Aug 2020 · Pune, India
Diploma, Modern Applied Psychology
Academy of Modern Applied Psychology
Deepened my understanding of human behavior and decision-making, the same curiosity that drew me to neuromarketing, applied to how people actually think and choose. Finished in eight months alongside everything else I had going on.
AI & Strategy
Fluent in AI.
Grounded in strategy.
Everyone claims to use AI now. Here is what I have actually produced with it, and where I recorded that it could not be trusted.
Original research · September 2026
I ran my own study on how AI assistants pick brands.
Everyone in this field cites the same second-hand statistics. I wanted first-party data, so I built a 20-question prompt library across four buying stages and ran it, logged out, scoring every brand and every source that came back. 23 answers, 55 brands, 69 brand mentions.
Three of the four engines I designed for could not be measured anonymously any more. That turned out to be the first finding.
Three of the four engines I designed for could not be measured anonymously any more. That turned out to be the first finding.
Figure 1. Brand presence by buying stage, 23 answers collected 13 to 14 September 2026.
Brands do not appear until the buyer is already shopping
Share of answers naming any brand, by buying stage: 1 of 5 problem aware, 4 of 6 solution aware, 7 of 7 brand comparison, 4 of 5 purchase intent.
"How do I start strength training if I have never lifted before" returned technique advice and named nobody. So did the questions about weekly exercise and about cardio against weights. The assistant answered like a coach rather than a shop.
That reverses the usual AEO assumption. Top-of-funnel is where most brands compete hardest, and it is exactly where the assistant names no one.
"How do I start strength training if I have never lifted before" returned technique advice and named nobody. So did the questions about weekly exercise and about cardio against weights. The assistant answered like a coach rather than a shop.
That reverses the usual AEO assumption. Top-of-funnel is where most brands compete hardest, and it is exactly where the assistant names no one.
Figure 2. Share of voice against benchmark bands, and publisher concentration.
Nobody owns this category
55 distinct brands across 69 mentions. The most-cited, Nike Training Club, holds 7.2% share of voice. Published 2026 benchmarks put a leader at 40 to 70%, and nothing here reaches even the 10 to 20% top-ten band.
An empty leader slot is an opportunity finding rather than a neutral one. The cost of taking share is currently low. Windows like that close.
An empty leader slot is an opportunity finding rather than a neutral one. The cost of taking share is currently low. Windows like that close.
One publisher keeps reappearing
Garage Gym Reviews was cited in 4 of the 16 answers that named brands. No other publisher appeared more than twice, across 44 distinct publishers. It also showed up in an earlier run four weeks before, on a different engine.
One review site is doing a disproportionate share of the deciding, which means a strategy aimed at your own website is optimising the wrong asset.
One review site is doing a disproportionate share of the deciding, which means a strategy aimed at your own website is optimising the wrong asset.
A third of the time, the source is the seller
In 8 of 23 answers, a brand named in the answer was also cited as the evidence for it. One query returned three apps, all sourced to a domain belonging to one of the three.
The assistant gives these the same visual authority as an independent review. A buyer sees a citation and reads it as verification. Roughly a third of the time it is the seller's own marketing.
The assistant gives these the same visual authority as an independent review. A buyer sees a citation and reads it as verification. Roughly a third of the time it is the seller's own marketing.
The answer changed in ten minutes
In an earlier run I put the same query to the same engine twice, ten minutes apart. A brand that led the first answer had vanished from the second. That makes AEO a monitoring discipline rather than a one-off audit.
Engines agree on the winner and on almost nothing else
Across the queries run on both engines, the same brand was named first in 2 of 3. But overlap between the full brand sets ran 0.09 to 0.20 on Jaccard, so beneath the winner they name almost entirely different fields.
Future led on both engines here, and led on two different engines four weeks earlier. Getting into an answer is a per-engine fight. Owning the top slot looks like a position that transfers.
Future led on both engines here, and led on two different engines four weeks earlier. Getting into an answer is a per-engine fight. Owning the top slot looks like a position that transfers.
What this does not show. Twenty-three observations is a pilot, so treat every percentage as directional rather than settled: the 7.2% share of voice at the top of this study carries a 95% interval of 3% to 16%. The structural findings hold at this size, the percentages do not. One engine carries most of the weight, since 20 of the 23 observations are Bing Copilot and the cross-engine comparison rests on three shared queries. Counting brands is a judgment call: a name in a shopping module was counted the same as one recommended in prose. This is one category, one country, one week, so the gatekeeper effect is almost certainly category-specific. And appearing in an AI answer is not a click, a visit or a sale. Nobody has a clean attribution model for this yet. Claude was excluded deliberately: this study was assembled with Claude's help, so including it in a competitive ranking would be a conflict of interest.
The data behind this study
Every figure above comes from these four files. This is the September 2026 run: 23 observations across Bing Copilot and ChatGPT, collected 13 to 14 September. The August run referenced above is described in the write-up, not included here.
queries.csvThe 20-question prompt library, four buying stages1.5 KBobservations.csvEvery answer collected, brands in order, sources cited5.6 KBaudit_results.jsonEvery figure on this page, computed from the observations2.3 KBRESULTS.mdThe write-up, including what the study does not show6.3 KBFramework · September 2026
Then I wrote the method for fixing it.
The study says what is happening. This says what to do about it: a six-step diagnostic anyone can run their own brand through, built from the data above rather than from other people's blog posts.
The fix order the data produced, ranked by expected effect.
It argues against the standard advice
Almost every AEO guide starts with your own content and schema markup. The data says start somewhere else. If one publication gatekeeps your category, being absent from it is the largest single gap and no amount of owned content closes it. Third-party presence first, entity infrastructure second, content third, schema fourth.
If a brand cannot clear the third-party threshold, owned-content work has nothing to amplify.
If a brand cannot clear the third-party threshold, owned-content work has nothing to amplify.
You run your own brand through it
Build a 20-question prompt library across four buying stages. Run it clean, logged out, with personalization off, because your own account has been trained on your own behavior and will show you a flattering answer no prospect will ever see. Score what comes back on a five-point scale, measure against published benchmark bands, then score every surfacing brand on five entity signals to find out why they surface.
It ends with a schedule, because readings expire
Given the drift I measured, any reading has a half-life of weeks. Re-run monthly at minimum, weekly in a fast-moving category, and treat every figure as a point estimate with a wide band around it. At 23 observations a 7% share of voice carries a 95% interval of 3% to 16%. To quote a share that size within five points you would need around 104 observations.
Most published AEO numbers do not disclose their sample size, which should tell you something.
Most published AEO numbers do not disclose their sample size, which should tell you something.
What the method cannot do. It does not measure revenue: appearing in an AI answer is not a click, a visit or a sale, and nobody has a clean attribution model for this yet. The link between entity signals and visibility is a pattern strong enough to prioritize against rather than a proven mechanism. Every reading expires. AEO is roughly where SEO was in 2003: real, commercially significant, and full of people asserting certainty they have not earned.
100%
Model precision, at 20% recall
Python, scikit-learn
Python, scikit-learn
I read models as well as run them.
Engineered a binary target from survey data, trained a logistic regression on a 70/30 split, then read it through odds ratios rather than stopping at the score. I did the modeling on a four-person team project. Odds ratios are what turn a model into something a non-technical stakeholder can act on.
25%
Of my graduate exam grade
was prompt quality
was prompt quality
My AI use was graded, prompts included.
My analytics exam required ChatGPT and scored the prompts at a quarter of the mark. Pasting the question in scored zero. Marks went to how far you pushed the model and where you pushed back on it.
2024
AI as a research
instrument
instrument
I ran AI-led research, then documented where it could not be trusted.
Built consumer profiles, wrote a structured interview protocol, ran the interviews through AI, and coded every response by topic, code and theme. It produced two product recommendations and a limitations section stating the profiles were hypothetical and the scope narrow.
I found the gap before anyone asked me to look
Buyers now ask AI assistants what to buy instead of typing into a search box. Organisations that do not appear in those answers lose ground quietly, because nothing in their analytics tells them it is happening.
I spotted that exposure inside a B2B organization I work with and built an AI Engine Optimization strategy to close it, carrying it from research through to an implementation plan. Nobody briefed me on it.
AI rollouts rarely fail on the product
Teams rarely fail at AI because they picked the wrong tool. They fail because nobody designed the workflow around it, so it gets used enthusiastically for a fortnight and then quietly abandoned. The work that decides it is unglamorous: finding where the tool genuinely removes friction, shaping the habit around it, and going back a month later to check whether it held.
Knowing the technology is the easy part. Getting a team to work differently because of it is the harder skill.
The Arc
An early adopter, deliberately.
I didn't arrive at AI when it became unavoidable. I started early, learned to use it responsibly, and grew into leading strategy on it.
2022-23 · Kemse
First Mover
Early ChatGPT adopter for marketing and research workflows, the moment the tools appeared.
2023-24 · Suffolk
Trained Properly
Learned to use AI efficiently without becoming reliant on it, paired with Python for coding, research, and neuromarketing.
2024-25 · Vylero
Applied in the Wild
Used AI tooling to accelerate campaign analysis and partnership work in a live startup.
2025 · CPS
Hands-On Lead
Driving AI-visibility strategy and day-to-day adoption of AI tools across the team.
Tools I actually use
What I have actually built with each one.
Python + scikit-learn
Trained a logistic regression classifier end to end and interpreted it through odds ratios. Also pandas for preparation, EDA and feature engineering.
AI Engine Optimization
Built an AEO strategy from scratch and carried it to an implementation plan. Custom frameworks for how a brand surfaces inside an LLM answer.
ChatGPT
Using it since 2022, and formally assessed on prompt quality at graduate level. Research, competitive analysis, and AI-run interview protocols.
Claude
Document workflows, operations summaries and structured research. Where I go when the thinking needs to hold across a long document.
NotebookLM
Rolled out for institutional knowledge: source-grounded synthesis across document sets, so answers cite the source rather than inventing one.
SAS & R
Cluster analysis on 15,000 subscribers, regression on paid search performance, and predictive modeling across 1,500 guest records.
iMotions
Certified. Eye tracking, facial expression analysis and skin conductance for consumer research at Suffolk's X-Lab.
Perplexity AI
Real-time research and source verification, mainly for competitive and market intelligence where recency matters.
Google Analytics 4
Certified. Campaign performance, funnel analysis and behavioral data behind growth decisions.
Let's Connect
Let's build
something real.
I'm always up for a good conversation about growth, AI, or where marketing is heading next. A role, a collaboration, or just comparing notes, all of it works.