# True Stories > True Stories is an AI-powered qualitative research analysis platform that applies 200+ behavioural science, consumer psychology, and marketing frameworks to interview and focus group transcripts. It delivers framework-driven analytical conclusions — not summarisation — from qualitative market research data. Built by qualitative researchers, for qualitative researchers. - Primary audience: insight agencies and client-side insight teams - Website: https://www.truestories.earth/ - Platform: https://ai.truestories.earth/ - Book a demo: https://calendly.com/truestoriesai/30min - Contact: alex@truestories.earth --- ## Why True Stories ### Built for one purpose Built from the ground up for one purpose only: deep analysis of qualitative research data. True Stories is not a general-purpose AI tool adapted for research. Nor is it a swiss army knife with analysis bolted-on as a second or third order feature. Every architectural decision, every AI model, every analysis framework, every prompt has been engineered specifically for extracting the deepest possible insight from human conversation. ### Multiple best-in-class AI models working in unison True Stories doesn't rely on a single AI model. It orchestrates multiple specialist models across the analysis pipeline: embedding models that understand semantic meaning across languages, retrieval models that find the most relevant evidence for each query, reranking models that prioritise the strongest evidence, and large language models with extended thinking capabilities that produce the actual analysis. Each model is selected for its specific strength at its specific task. ### Theory-grounded analysis Every analysis applies established marketing, consumer psychology, and behavioural science frameworks to organise evidence into analytical conclusions. Jobs-to-be-Done, Self-Determination Theory, Aaker's Brand Identity Model, Innovation Resistance Theory, the COM-B Model, Cognitive Dissonance Theory, and many others. The AI doesn't just summarise what people said. It explains why they behave the way they do, using the same theoretical frameworks researchers already work with. ### Comprehensive prompts that minimise hallucination Each specialist analysis enhancer uses a purpose-built prompt of 3,000-5,000 words that specifies the analytical framework, theories to be applied, evidence standards required, output structure, and explicit instructions against fabrication. Every finding must originate from the evidence in the transcripts. The system is designed to say "the evidence is limited" rather than invent confidence it doesn't have. ### "Show me the thinking" not "show me the quotes" Most platforms treat qualitative analysis as a sorting exercise: group similar quotes, label the piles, show the receipts. This is evidence of data summarisation, not of data analysis. True Stories takes a fundamentally different approach. An insight about why customers resist switching isn't hiding in one verbatim quote. It emerges from the pattern across dozens of data points, interpreted through the lens of theory that explains human behaviour. The system documents where AI analysis and human researcher observations converge or diverge, and provides evidence strength ratings against the most critical insights. Every analysis response supports follow-up questions. Ask "can you explain how you arrived at this insight?" and the AI responds with a contextual, reasoned answer — explaining which transcripts contributed, which segments showed the pattern most strongly, where the evidence is robust and where it's tentative, and why the theoretical framework matters. This is more powerful than a static list of matching quotes. It's a conversation with an analyst who has read every transcript, applied every relevant theory, and can explain and discuss their reasoning on demand. --- ## Project Setup ### Creating a project Create a new project from the homepage. Give it a meaningful name that ideally includes the client and research topic (e.g. "Company X - Brand Perception Study"). This name appears throughout the platform and on any exported project documents. ### Choosing a project type **General** — for all standard qualitative research projects. Usage and attitudes studies, brand research, customer journey mapping, segmentation studies, and any project where the research covers broad topics. **Creative** — for projects where the primary focus is evaluating the audience response to advertising creative. The Analysis Genie, analysis enhancers, and report generation are all optimised for creative evaluation, including per-idea diagnostic scorecards. **Concept** — for projects focused on testing product concepts and/or features. Includes specialist concept response analysis with Kano-model feature diagnostics and Go/Develop/Rethink/Drop verdicts. You can still analyse the audience response to different ideas in a General project, but the Analysis Genie and final report documents will not treat this as the central focus of the research. ### Assigning a client Assign your project to a client at any time — during creation or later via the Account Management menu. Assigning a client helps organise projects and restrict team access: team members can be given access at the client level, so they only see projects belonging to their assigned clients. Client assignment will also enable cross-project analysis in the future. ### English vs Multilingual Choose English if all transcripts are in English. Choose Multilingual if transcripts include any non-English content. The multilingual option triggers backend optimisations for processing and analysis of non-English transcript data. Once set, this cannot be changed. ### Research Brief Aside from transcripts, the Research Brief is the single most important input data for a project. The suite of AI models uses it to prioritise which themes to explore in depth, frame the analysis around research objectives, and produce recommendations that address genuine business needs. A project with a strong brief will produce dramatically better analysis than one without. Include: the client and business context, the research objectives and hypotheses, an overview of the methodology and sample, and any background the AI should know — such as constraints or sensitivities. ### Ideas Being Tested For Creative and Concept projects (and any General project that tested specific ideas), use the Ideas Being Tested section in the Project Context to define each idea. Give each idea a clear name, a description of what participants were shown or told, and any aliases participants might use to refer to it. Visual stimulus material — storyboards, concept boards, packaging designs — can be uploaded directly for each idea. The platform analyses each image with AI to build a detailed understanding of what it depicts. When participants talk about "the one with the guy and the lemon" or "the blue packaging," the AI knows exactly which idea they mean. ### Tagging transcripts Tags are sub-sample descriptors attached to transcripts, e.g. age group, gender, country, audience type, or any other dimension in the sample frame. Tags enable two features: the AI uses them to identify and report on sub-sample differences in every analysis, and Targeted Analysis can focus specific queries on subsets of the data. **Manual tagging:** Create tag categories (e.g. "Geography") and values within each (e.g. "London", "Manchester", "Edinburgh"). Then assign the relevant tags to each transcript individually. **Automatic tagging:** For projects with many transcripts, download the tag map Excel template, fill in the spreadsheet with tag categories, values, and filename mappings, then upload the completed file. The platform automatically creates all tag categories and values, and when transcripts are subsequently uploaded, each file is automatically tagged based on the filename mappings. This is significantly faster than manual tagging for larger projects. ### Uploading transcripts Supported formats include Word (.docx), PDF, and plain text. For best results, ensure transcripts clearly label who is speaking — use "Moderator:" or "Interviewer:" for the researcher, and participant names or identifiers for respondents. The platform automatically detects speakers, extracts timecodes where available, and splits content into optimally-sized chunks for analysis. --- ## Analysis ### What is Analysis Context? Analysis Context is the accumulated understanding True Stories builds about the research data. Each time an analysis is run using one of the Analysis Enhancers, the platform stores the key findings and uses them to enrich all subsequent analysis. The more context built, the sharper and more interconnected the insights become. Think of it as layers of understanding. A Usage & Attitudes analysis establishes the behavioural landscape. A Segments analysis identifies who the audience groups are. A Drivers analysis — now informed by both U&A and Segments — can explain what motivates each segment specifically, not just the audience in general. Each layer deepens the next. ### The Analysis Enhancers Eight specialist analysers, each grounded in established theoretical frameworks: **Usage & Attitudes** — establishes the behavioural and attitudinal baseline. How people currently behave, what they think, what they do. This is almost always the right place to start. **Customer Segments** — identifies distinct audience groups based on their attitudes, behaviours, and needs. Segments are referenced throughout all subsequent analysis, making every finding more actionable. **Brand Perception** — how the brand lives in people's minds. Evaluates brand identity across multiple dimensions including product associations, organisational perception, personality, and symbolic meaning. **Customer Drivers** — what motivates each segment. Goes beyond stated preferences to uncover the psychological and functional drivers behind behaviour. **Barriers & Friction** — what blocks adoption, switching, or change. Identifies both rational barriers (cost, complexity) and psychological barriers (loss aversion, status quo bias). **Customer Journey** — maps the experience from first awareness through to loyalty or churn. Identifies moments of truth, pain points, and opportunities at each stage. **Creative Effectiveness** (Creative projects) — evaluates advertising creative across 10 diagnostic dimensions including brand linkage, believability, emotional engagement, and persuasion. Produces per-idea scorecards with Go/Develop/Rethink/Drop verdicts. **Concept Response** (Concept projects) — evaluates product concepts with Kano-model feature diagnostics that distinguish must-have features from delighters and identify reverse features that actively undermine appeal. ### The Analysis Genie The Analysis Genie reads the Research Brief and designs a complete analysis plan tailored to the project. It sequences the analysis enhancers in the optimal order, crafts specific questions for each, and includes a "Challenge" question that deliberately tests assumptions in the research brief — surfacing things the researcher might not have thought to ask. The plan can be reviewed and modified before running. Press "Run All" and the Genie automatically cycles through every recommended analysis, building context with each layer. Or run each question individually for more control. ### Asking your own analysis questions Beyond the Genie's recommendations, any question about the data can be asked. Start a new chat, type the question, and optionally select an Analysis Enhancer to frame the analysis. The AI draws on all the context built so far, plus the raw transcript data, to produce a thorough, evidence-grounded response. ### Targeted Analysis Use the filter to focus analysis on specific transcripts based on their tags. For example, in a multinational study, analyse only the UK respondents, or compare male vs female responses, or focus on a specific age group. The AI analyses only the selected transcripts but still draws on the full analysis context for framing and comparison. ### Follow-up questions Every analysis response supports follow-up questions within the same chat. Ask for more detail on a specific finding, challenge an insight, request evidence for a particular claim, or explore a tangent. The AI maintains the full conversation context and builds on its previous analysis. To start a completely new line of enquiry, select "New Chat" from the sidebar. Each chat preserves a distinct analytical thread. ### Adding analysis to the Project Report All Genie analysis and all manual analysis that uses an analysis enhancer (e.g. U&A, Brand, Barriers) will automatically be added to the project report as analysis sections. For all other analysis responses, use the three-dot menu to add it to the Project Report as a New Section. Analysis added to the Project Report becomes the input for the automated report generation process. --- ## Reporting ### How report generation works When there are at least three analysis sections in the Project Report, automatic report generation can be triggered. The platform runs a multi-stage pipeline that reads all analysis, triangulates findings across different analysis types, identifies cross-cutting insights and tensions, and produces a complete set of debrief-ready documents. For Creative and Concept projects, the specialist analysis (per-idea scorecards, diagnostic ratings, verdicts) is preserved in full and flows through to every report document. ### How moderator notes are integrated Before generating the project report, add moderator notes to the project. These could include insights, observations, hypotheses, and ideas. The report generation process rigorously tests these observations against the systematic AI analysis, documenting where they converge (highest-confidence findings), where they add unique value (things only a human in the room could observe), and where they diverge (which itself is informative). ### The report documents **Mini Report** — a straightforward synthesis of the most important findings and recommendations. Contains the key insights, recommendations, segment profiles, contradictions in the data, and a direct response to each of the research objectives and hypotheses listed in the research brief. **Strategic Narrative** — a more readable, story-driven version of the key insights and recommendations. Likely to be the basis of any slide deck or client presentation. **Analysis Digest** — a synthesis of the project analysis. Every analysis section is re-examined from the ground up: the original summaries are rewritten with the benefit of the completed deep analysis, and moderator notes (if provided) are rigorously integrated. This produces a more readable, more insightful version of the analysis corpus. **Appendices** — the complete analysis corpus for reference. Every project report analysis section in full, organised by type. This document is more for records than something that necessarily needs to be pored over. ### Exporting documents Every report document can be exported as Word (.docx) or PDF via the download menu in the report viewer. Exports include proper headers with document type, project name, and date. Full multilingual support — Chinese, Japanese, Korean, Arabic, Hebrew, Thai, Hindi, Russian, and 50+ other languages render correctly. ### Adding verbatim quotes After report generation, verbatim participant quotes can be added to the Mini Report and Strategic Narrative documents. Use the Add Quotes feature to select any line from the report to substantiate with verbatim and the AI searches the transcripts, presenting a range of options based on relevance. Quotes can be filtered based on transcript tags and quotes taken from non-English transcripts can be translated at the touch of a button. File names and timecodes can be displayed for each added quote to facilitate video editing. ### Regenerating the report If more analysis is added after generating the report, or to improve the output, the report can be regenerated at any time. The platform rebuilds all documents from scratch using the latest analysis corpus. --- ## Key Differentiators - **Analysis, not summarisation**: applies theoretical frameworks to explain why people behave the way they do, rather than grouping and labelling quotes - **200+ established frameworks**: Jobs-to-be-Done, Self-Determination Theory, Aaker's Brand Identity Model, Innovation Resistance Theory, COM-B Model, Cognitive Dissonance Theory, and many more - **Evidence-grounded**: every finding must trace back to transcript data; the system says "the evidence is limited" rather than fabricating confidence - **Speed**: a full qualitative project can move from transcript upload to client-ready report documents in a single afternoon - **Multilingual by design**: analyse transcripts in 75+ languages without translation; produce all outputs in English - **Multiple specialist AI models**: embedding, retrieval, reranking, and large language models each selected for their specific task - **Moderator note integration**: tests human researcher observations against systematic AI analysis - **Three project types**: General, Creative (with per-idea scorecards), and Concept (with Kano-model diagnostics) --- ## Company True Stories Ltd (Company No. 16186619) is incorporated in England and Wales. Registered office: Unit 3 Chiltern Court, Asheridge Road, Chesham, Buckinghamshire, HP5 2PX, United Kingdom. VAT No. 511340937. Founded by Alex Charlton, a qualitative researcher with 1,000+ group discussions, depth interviews, and consumer ethnographies across clients including the Advertising Association, Amazon, BBC Studios, Carat, Fox, ITV, Microsoft, Liberty Global, Netflix, Reach, Samsung, Sony, Starbucks, TD Bank, Telefonica, Twitter, UKTV, and Warner Bros Discovery. Contact: alex@truestories.earth | +44 7780 868 785