Hello, I’m Sneha
I’m a linguist who works on a simple question: when an AI system such as Google’s AI Overviews answers someone’s question, why does it quote some texts and ignore others, and how can being quoted turn into customers and revenue for a business?
I created the SARC Content Methodology for AI visibility and conversion to answer that question in practice, and I founded Written Intelligence to test it in public. This page tells the story of how I got here: through three degrees in language, media and linguistics, through a website I started because I needed proof of my own work, and through a method that grew out of that website until it became the subject of a doctoral proposal.
Education: three degrees, one subject
Looking back, all three of my degrees were about the same thing: how language works on the people who read it.
BA in English, Journalism and Psychology, Bangalore University, India (2018). My undergraduate degree combined three ways of looking at a text: as literature, as news and as something that acts on a reader’s mind. I studied on a merit scholarship. Psychology in particular stayed with me. It was my first introduction to the idea that how something is said changes what people believe and do, which is the idea behind the Conversion letter in SARC.
MA in Mass Communication and Journalism, Jain University, India (2020). My master’s degree moved from single texts to whole systems of communication: how messages are produced, distributed and received at scale. I graduated as Bronze Medallist and top subject scorer. Journalism training also gave me the habits that now sit inside the Authority letter: name your source, check your claim and put the most important fact first.
MSc in English Language and Linguistics, University of Stirling, United Kingdom (2024). I moved to the UK for my second master’s degree, supported by a country-specific scholarship. Linguistics gave me the tools I had been missing: ways to describe precisely how a sentence is built, how it signals what is known and what is new, how it marks where its evidence comes from, and how readers decide what is relevant. These are the concepts SARC now rests on: information structure, evidentiality and stance, and relevance theory.
The order matters to me. Journalism taught me how to write for readers, psychology taught me why readers respond, and linguistics taught me how to describe both precisely enough to test them.
How SARC started
SARC did not start as a research idea. It started because I needed to prove something about myself.
A portfolio built out of necessity
After my MSc, I found it hard to get hired. I knew I could write well, but employers wanted evidence, and I didn’t have the industry track record they were looking for. So I built the evidence myself. I started snehamukherjee.info as a portfolio of my own writing, before I had even settled on a niche.
That decision changed the direction of my work. A portfolio only works if people find it, so I had to learn how my own pages got discovered. And at exactly that moment, the way people found things online was changing. More and more questions were being answered by AI systems such as ChatGPT, Perplexity and Google’s AI Overviews, which read the web and quote a handful of sources rather than listing ten links. My pages were no longer only competing to rank. They were competing to be quoted.
Noticing a pattern
As I wrote and published, I started to notice that the pieces AI systems picked up had something in common, and it was linguistic. They answered the question in their first sentence. They named things in full instead of saying “it” or “this”. They said where their facts came from. They stayed on one question at a time. These were not tricks. They were features of clear writing that I had been trained to describe in my MSc, now showing up as the difference between a text that gets cited and one that gets ignored.
The first test
I turned those observations into a set of rules and applied them deliberately to my own site. The result was the first evidence that the rules mattered: 15 articles on snehamukherjee.info earned 3,000 AI impressions in four months, without paid promotion or backlink campaigns.
Giving it a name
Once the rules worked more than once, I organised them into four parts, each answering a different question about a text:
- Search. Can it be found for the questions people actually ask?
- Authority. Can every claim be traced to a source?
- Relevance. Does each passage answer its question, and only that question?
- Conversion. Does the reader have a clear next step, even if they only ever see a quoted fragment?
The first letters gave the method its name: SARC. I wrote about it on my site as part of a series of articles on AI visibility and content structure published between April and August 2026.
Why Conversion is the point
The four letters are not equal. Search, Authority and Relevance exist to serve the last one, because visibility on its own does not pay a business’s bills.
AI Overviews made that problem urgent. When Google shows an AI-generated summary, people click through far less. In Pew Research Center’s 2025 study of real browsing data, users clicked a traditional search result in 8% of visits when an AI summary appeared, compared with 15% when it did not, and they clicked a source link inside the summary in only about 1% of visits. A business can be cited in an AI Overview, seen by thousands of people, and still gain nothing, unless the words that get quoted do the selling.
That is what SARC is built to do. Each letter moves a reader one step closer to becoming a customer:
- Search gets the business’s page retrieved for the questions buyers actually ask, including the long, specific questions that trigger AI Overviews.
- Authority makes the quoted passage trustworthy enough for both the AI system and the reader to rely on.
- Relevance makes sure the passage that gets quoted is the one that answers the buyer’s question, not a generic paragraph.
- Conversion makes sure that quoted passage carries the business’s name, what it offers and a reason to act, so that even a reader who never clicks remembers who answered their question.
The path SARC is designed for runs from an AI Overview appearance to brand recall, then to a branded search or a click, then to an enquiry, sign-up or purchase, and finally to revenue. In practice, that means writing the sentences most likely to be quoted so they name the business and its offer, answering the pre-purchase questions (price, comparisons, how it works, what happens next) that AI Overviews increasingly answer, putting proof next to every claim, and giving every page one clear next step.
It also changes how success is measured. An AI impression is a means, not an end. The real measure of SARC is whether AI visibility shows up in branded searches, sign-ups, enquiries and revenue, which is why the method tracks all of them together.
Testing it more widely
One site is not enough to show that a method works, so I kept testing. Across 80 independently published articles, 37.5% were cited by AI systems, and the same articles earned 33,700 organic Google impressions. I also launched Written Intelligence, a publication where I could apply SARC to a new subject and watch the results build in public.
From method to research
The more I used SARC, the clearer its limits became. I could show that pages written with it did well, but I could not show why, or which of the four parts mattered most, or whether the results would hold on someone else’s site. Those are research questions, not practice questions. In October 2026 I set up sarcmethodology.com, and in November 2026 I published the working paper that sets out SARC’s linguistic basis and twelve propositions that a study could test. The method that started as a way to prove I could write is now the subject of my doctoral proposal.
Written Intelligence
Written Intelligence is the publication I founded and edit. It covers two subjects that are changing faster than most people can keep up with:
- AI Search: how AI systems find, choose, rank and cite their sources.
- AI Safety: how AI systems are kept safe, from guardrails to risk and trust.
The two belong together. Both are about whether we can rely on what an AI system tells us: AI Search asks where its answers come from, and AI Safety asks what stops those answers from causing harm. My approach is the same for both. I read the research, test the claims and explain what matters in plain language, in writing and on video.
Why I started it
Written Intelligence does two jobs. For readers, it is a clear guide to two fast-moving fields. For my research, it is a live test of SARC. Every article is written with the method, on a subject where I had no existing audience, so its results are a cleaner test than my portfolio site, which already had a history.
What it has achieved so far
- Together with my portfolio site, Written Intelligence has earned about 10,087 AI impressions (as of October 2026).
- I have been featured on Qwoted as an expert source for Written Intelligence.
- Each episode of my podcast links to a full written guide on Written Intelligence, so the writing and the video work as one body of work.
Where it is going
Written Intelligence is where the research behind SARC will be most visible while the PhD is under way. As the study produces findings, I plan to publish plain-language summaries there, so that writers and businesses can use what the research shows without waiting for the journal articles.
Other public work
- AI Safety & Guardrails (podcast). A YouTube podcast I host on AI safety, AI agents and AI search. Every episode links to a full guide on Written Intelligence.
- LinkedIn newsletters. I write four newsletters, which together have 1,875 subscribers and more than 50,000 impressions.
- Content experiments. Alongside SARC, I run small tests on how AI systems choose sources and share what I find, including the 80-article citation test described above.
- Certifications. HubSpot Academy Digital Marketing (2026, scored 59/60) and Semrush Academy Content Toolkit (2026).
Why a PhD
SARC works in practice, but practice can only take it so far. A PhD would let me find out which parts of the method really drive results, why they work in linguistic terms, whether they hold across different AI systems, subjects and websites, and, most importantly, how much of that AI visibility turns into enquiries, sign-ups and revenue. The goal is a tested method that businesses can rely on to turn AI Overview visibility into revenue, not just a method that worked for me.
I am looking for a supervisor whose work touches discourse analysis, computational linguistics, information retrieval or the language of digital media.
Get in touch: snehamukherjee@snevo.info · LinkedIn · Written Intelligence