Blog / Affiliate marketing
AI for Generating Case Studies: Build Trust Without the Writing
You have data. The MyLead panel shows you specific numbers: CTR, EPC, conversions, month-over-month growth. The problem is that raw statistics don't convince anyone on their own - a table of numbers doesn't build trust the way a well-told story does. A case study is the bridge between data and the emotions that drive a decision.
AI won't invent results you don't have - and it shouldn't. But it can take your real data and turn it into a narrative that actually persuades: structure, language, the turning point in the story. In this article, I'll show you how to do it step by step - and where the red line is that you must never cross if you want to build trust instead of destroying it.
What you'll learn from this article
Why case studies build trust more effectively than marketing claims
How to turn raw MyLead panel data into a persuasive narrative without adding anything of your own
A ready-to-use workflow with 4 prompts: from input data to a formatted case study and real user opinions
Where the red line lies between building trust and destroying it
What legal requirements (GDPR, affiliate content disclosure) a case study must meet
Why case studies build trust better than marketing claims
"Our method increases conversions" is a claim. "After implementing smart scheduling, conversions in my finance campaign rose from 2.1% to 3.5% over 6 weeks" is evidence. The difference isn't cosmetic - it's the difference between an ad and a case study.
A case study works because:
It shows the process, not just the result - the reader sees which steps led to the outcome, so they can judge whether the same logic would work in their own situation.
It has concrete numbers - "a 67% increase" is more convincing than "a significant increase," because it sounds like an actual measurement, not a rhetorical flourish.
It shows the starting point - a good case study doesn't begin with success. It begins with a problem the reader can identify with.
This is why case studies convert better than generic product descriptions - but it only works when the data is real. A fabricated case study that gets exposed destroys trust faster than any case study could ever build it.
How to turn MyLead panel data into a narrative
Most publishers have data but don't know how to tell it as a story. AI excels at this specific task: it takes the numbers and facts you provide and arranges them into a narrative structure - without adding anything that wasn't in the input data.
The classic case study structure that works:
Starting point (the problem) - what was the situation before the change? What results, what frustrations, what specific problem needed solving.
Action (what you did) - specific steps, tools, decisions. Without this part, a case study is just an announcement of a result, not a story.
Result (the data) - specific before-and-after numbers. The more precise, the more credible.
Takeaway (what it means for the reader) - how the reader can apply the same logic to their own situation.
AI works best on steps 1, 2, and 4 - turning raw facts into a readable narrative. Step 3 must come exclusively from your real data.
Workflow step by step with prompts
Step 1: Gather input data from the MyLead panel
Before opening AI, collect specific numbers: campaign statistics before and after the change, the time period, and exactly what change you made (e.g. a creative change, landing page optimization, targeting change). The more precise the input data, the better and more credible the resulting case study.
Step 2: Generate the narrative structure
PROMPT 1 - Building a case study from real data:
You are an affiliate content marketing expert. Help me build a case study based on MY REAL campaign data. Do not add any numbers, quotes, or facts that I haven't provided below.Campaign data:- Niche: [e.g. personal finance]- Starting problem: [e.g. low CTR, high CPL, low conversion]- Values before the change: [specific numbers]- What was changed: [specific description of the action]- Values after the change: [specific numbers]- Time period: [e.g. 6 weeks]Build the case study using this structure:1. Starting point - describe the situation and problem (2-3 sentences)2. Action - describe step by step what was done (3-5 sentences)3. Result - present the before/after data in a readable format4. Takeaway - what this means for other publishers in a similar situationIMPORTANT: use only the data I provided. If I haven't provided something (e.g. the exact cause of the increase), flag it as a "likely factor" rather than presenting it as fact.
Step 3: Add context and benchmark comparisons
PROMPT 2 - Industry context:
Based on the case study below, propose (as a separate, clearly labeled section) a reference to publicly available industry benchmarks that could help the reader judge whether this result is typical or exceptional.Case study:[paste the case study from step 2]Requirements:- Don't invent benchmarks - if you're not certain about industry data, state that it should be verified against current industry reports- Clearly label the benchmark comparison as approximate, not scientific
Step 4: Format for publication
PROMPT 3 - Formatting for the blog:
Format the case study below for publication on the MyLead blog.Requirements:- Title with a specific number from the result (e.g. "How I Increased Conversion by X% in Y Weeks")- Introduction building context (1-2 sentences)- Clear headings for each section of the structure- Highlight key numbers (bold)- Closing with a practical takeaway for the readerCase study:[paste the final version]
How to collect real user opinions and format them with AI
Genuine opinions and testimonials are powerful social proof - but they must be real and collected with the consent of the person who said them. AI has exactly one role here: helping you collect, organize, and format real opinions. It never creates their content from scratch.
Collecting opinions
A short survey after a success - if you run a publisher community or are in touch with people using your materials, send a short form (Google Forms, Typeform) asking about a specific result, along with consent for publication.
Transcribing conversations - if you talk with someone (e.g. a webinar recording, a voice call), AI (e.g. Descript) can transcribe the recording, making it easier to pull out quotes for further formatting - without changing their content.
Written consent - always ask explicitly whether the person agrees to have their opinion published with their name, initials, or anonymously. Keep that consent on file.
Formatting opinions with AI
PROMPT 4 - Formatting a collected opinion:
I have a real user opinion that I want to format for publication. Do NOT change the substantive content, meaning, or numbers provided by the person - you may only fix grammar, punctuation, and readability.Original opinion (from [name/initials, with consent]):[paste the verbatim opinion]Task:1. Fix only grammatical and punctuation errors2. Shorten if the opinion is very long, but preserve the meaning and all specific facts/numbers3. Do not add any statements the person didn't make4. Suggest a visual format (quote, short case, "user voice" section)
This prompt differs fundamentally from "generating testimonials" - AI never creates the content of an opinion, it only helps organize and format it while preserving the literal meaning of what the person actually said.
Red lines: what you must never do
This section is the most important part of the entire article. The line between building trust and destroying it is very clear here.
Never fabricate quotes - a made-up opinion attributed to a "satisfied user" is false social proof, regardless of whether the person is named or anonymous. It misleads the reader and, in many jurisdictions, violates consumer protection law.
Never invent numbers - "customers see an average 40% increase" without any real source data is fabrication, even if the number "sounds realistic."
Never publish an opinion without consent - even a genuine opinion, published without the author's consent, violates their rights and can carry legal consequences.
Never present a hypothetical scenario as an actual result - if you want to illustrate a strategy's potential without real data, always and clearly label it as an illustrative scenario, not as a specific client's case.
Practical rule: if you're asking yourself whether a piece of content "stretches the truth a little, but for a good cause" - that's exactly the signal that you're crossing a line you shouldn't cross.
Legal requirements - transparency and consent
Consent for use of likeness and opinions - in the EU and Poland, consent is required from a person before publishing their opinion, photo, or personal data (GDPR). Keep documentation of that consent.
Disclosure of sponsored content - if a case study promotes a specific affiliate offer, it must be disclosed as content containing affiliate links, in line with consumer protection authority requirements and general advertising transparency rules.
Prohibition on misleading practices - unfair commercial practices regulations (in Poland: the Act on Combating Unfair Market Practices) prohibit presenting fictional opinions or results as real. Consequences include administrative penalties and civil liability.
Disclosure of AI-assisted content - if AI helped draft the case study (even without fabricating data), it's worth considering a general disclosure about AI use in the editorial process, in line with growing transparency standards.
Summary - case study checklist
I have real data from the MyLead panel (before/after, specific numbers, time period)
The case study is based exclusively on facts that actually happened
AI was used to build the narrative from established facts - not to invent new ones
If the case study includes a user opinion - I have written consent for publication
If I'm presenting a hypothetical scenario - it's clearly labeled as an illustrative example
Any comparisons to industry benchmarks are labeled as approximate
The case study includes affiliate content disclosure if it promotes a specific offer
I keep documentation of consent for use of opinions and personal data
FAQ
Can AI invent data for a case study if I don't have enough results yet?
No, and this is an absolute boundary. AI can only build a narrative from facts you actually provided. If you don't yet have sufficient data, the better approach is to wait for real results or clearly label a hypothetical scenario as an illustrative example rather than presenting it as an actual case study.
Can I publish a case study without giving exact customer numbers?
Yes, you can use your own data from the MyLead panel or anonymized statistics, as long as they're real. The key is that the numbers come from actual results and aren't invented "for a more convincing effect" - even rounding or averaging should stay consistent with the actual data.
How does AI help with collecting user opinions if it can't create them?
AI helps transcribe recordings (e.g. from webinars or voice calls) and format and organize opinions that have already been genuinely collected - it improves grammar and readability, but never changes the meaning or adds content the person didn't actually say.
Do I need to disclose a case study as affiliate content if it only mentions a MyLead offer?
Yes, if the case study promotes a specific affiliate offer, it must be disclosed as content containing affiliate links, in line with consumer protection authority requirements and general advertising transparency rules in effect in Poland and the EU.
What should I do if I'm not sure whether a part of my case study "stretches the truth a little"?
That uncertainty is itself a warning sign. If you're asking whether something is fine "because it's for a good cause," it likely means you're crossing a line you shouldn't cross. When in doubt, remove the section or clearly label it as unconfirmed or illustrative.
Have data from your own MyLead campaigns worth turning into a case study? Log in to your account and check your panel statistics - that's a great starting point for writing your first, fully credible story about your results.
Have any questions? Feel free to reach us through our channels.