#7-case-studies-stories
Thread
👋 Curious how you’re all using AI for customer case studies
I’m doing some research into how content teams are actually using AI in their case‑study workflow, and I’d love to hear what’s working (or not) for you.
Are you using AI just to draft the core narrative, or are you also generating things like headlines, callouts, sidebars, pull quotes, and stat blocks?
Are you asking AI to help with brand voice consistency, narrative cohesion, or spotting content gaps across your story library?
And for anyone working closely with design teams — are you generating AI output in a way that can be automated into Adobe templates or other design tools?
Really curious how far folks are taking it and where the boundaries are right now.
Ai writes are case studies completely using Peerbound and content proofreads.
We use ChatGPT to completely draft the case study. (We use chat a lot because our company has an enterprise subscription.) Chat is still prone to hallucinations, so I check things like use cases and impact metrics.
Same with quotes: I will ask Chat to identify good quotes, but I verify that those are in the transcripts.
We actually use automation for design using a tool called Banner Bear. We have templates set up there and can push images over. In the past, I had our designer make templates in Canva so we could easily self-serve things like quote images, speaker images, etc.
We’re implementing Peerbound now, so will report back once we know how that impacts customer story and quote image development.
I usually start in NotebookLM to assess transcripts and build a mind map to see an example of how the topics the customer covers look laid out. I find NotebookLM is better at identifying (and not making up) quotes that I can weave into the narrative and use as pull quotes.
I'll often use Claude to draft the case study itself but then heavily refine it on my own afterward.
I've been working on a custom GPT that sort of helps address gaps in the content library. I load our customer stories into the knowledge and then you can query it, e.g. "Show me all the customer evidence related to X use case." Then, if we don't have anything that satisfies the ask, we know it's a gap. But this tool is primarily to quickly serve up customer proof when the GTM team needs an example to point to
I have two gems in Gemini, one a customer story draft creator and one a customer story analyzer. The draft creator uses several of our current stories as a template. I can then add the transcript of an interview, and it creates an early draft, which I then verify and edit to my liking.
The analyzer takes a URL of a case study (mine or a competitor's) and analyzes its strategy, pulling out things like the story structure, value framing, proof and persuasion tactics, and more. It ends with feedback: How it could be improved, where the opportunities are, etc. I think I got the base prompt from LinkedIn and tailored it to my needs. I'm happy to share it, if anyone wants.
I'm also investigating using Riverside for asynchronous customer interviews. With clients in Australia and other places where it's hard to match up time zones, I think it could be an interesting option!
We use AI to draft stories, but I always have to include in the prompt "Do not change the customer quotes AT ALL" because it's so prone to changing them. Drives me crazy, but once you have a good prompt it will usually stop doing it.