Somewhere in This Transcript Is a Story
How I turn SME interviews into narratives, and where AI fits in my process.
My first cover story for a publication was about Dr. Sam Foote, the Arizona Veteran Affairs whistleblower. I interviewed him at length on a recorded phone call, found what I thought was a great segue into his story, and built my lede around it. It involved the moon landing, and I promise it made sense to me at the time.
When my editor read the draft, though, he didn’t see the connection clearly enough. More importantly, he didn’t think our readers would either. They needed some context about Foote and why they should care about his story before I took them somewhere that, without it, seemed pretty far afield.
I was disappointed because I’d become attached to the lede, a habit I’ve since learned to quell (knowing content can change at any moment saves my sanity!). My editor didn’t make me throw the idea away, though — we made a compromise instead. I could work the moon landing and how it influenced the course of Foote’s career into the story later, but the lede needed to give readers an inkling of why they were reading his story in the first place. This was a story about a whistleblower, and we needed to bring that angle into that story before asking our audience to follow us somewhere else.
Interviews give writers all kinds of material that is interesting, surprising, or simply fun to write about. Sometimes what catches my attention while I’m talking to a subject really does become the lede. Sometimes I get so caught up in a detail that I give it more importance than it deserves. Learning to tell the difference is part of turning an interview into a story.
I record nearly every interview I conduct, but I still take notes. Those notes aren’t always detailed because I’m not trying to capture what someone is saying word for word. Because the recording has that covered, my notes tend to look more like this:
14:32 – possible lede
27:18 – quote?
34:02 – fact check
I’ve been interviewing subject matter experts (SMEs) for most of my career, and I’ve learned to pay attention when something in an answer makes me sit up a little straighter. It could be an unexpected way of explaining something, or an answer that gets more interesting after I ask a follow-up. I may not know yet how I’ll use it in the finished piece, so I mark the time and keep listening.
Once I have the transcript in front of me, I go back to those timestamps and find the passages I flagged. From there, I think about how the story might unfold and which moment gives me the most interesting way in. The comment I marked as a possible lede during the interview may still be my favorite, but reading the full conversation sometimes changes my mind. An idea that didn’t seem especially important in the moment might connect with something the SME said later and send me in a different direction.
The transcript gives me the raw material, but there’s still a lot of work between recording the conversation and the story someone will eventually read.
Not everything worth using belongs in quotation marks
One thing I’ve learned from writing interview-driven pieces is that a good answer doesn’t necessarily make a good quote. Most of us don’t speak the way we write, and a SME can give me exactly the information I need while taking a fairly circuitous route to get there.
When the idea is valuable but the spoken answer doesn’t work well on the page, I’ll often carry the substance into the narrative instead. I can explain the point clearly while staying faithful to what the SME told me, then use a direct quote when I want the reader to hear their actual language.
“Most of us don’t speak the way we write, and a SME can give me exactly the information I need while taking a fairly circuitous route to get there.”
This is also where writing from a transcript becomes much more than assembling an article around the best quotes. I’m deciding what information the reader needs to understand the subject and where the SME’s own words will make the piece stronger. I’m also rearranging a conversation that happened in one order into a narrative that may work better in another.
When I’m ghostwriting thought leadership, I approach the interview a little differently because I’m usually not looking for direct quotes. I listen for the SME’s thinking and try to understand the argument well enough to write it in their voice. The interview gives me the expertise and direction for the piece, and it often points me toward additional research I need to do. If the SME makes a compelling argument that would benefit from evidence, for example, I’ll look for credible statistics or data that can support it or give the reader more context.
By the time I begin drafting either kind of piece, I’ve already made many editorial decisions that won’t be obvious to someone reading the finished article. That’s part of why I’ve found AI most useful when it helps me move through the material more efficiently without asking it to make those decisions for me.
Where AI has made the process easier
I can’t timestamp every interesting moment while I’m interviewing someone because I’m also trying to stay present and actively listen during the conversation. There are plenty of times when I remember something the SME said and know I want to revisit it but have no idea where it appears in a long transcript. I used to search until I found it (a handy CTRL+F search if I remember some of the phrasing). Now I can ask AI to locate the passage, then go back to the transcript to check the wording and surrounding context myself.
Once I have a general sense of how I want the article to flow, I’ll also use AI to help me work through an outline. Having that direction first is important to me because I’m not asking the tool to decide what the story should be — I'm using it to help me organize a story I’ve already started to see.
The same is true when I get stuck on a sentence. Sometimes I’ll ask AI for another way to phrase what I’m trying to say and use a suggestion as a starting point. Other times I don’t like any of its suggestions, but seeing a few approaches that don’t work can help me figure out what does.
Headline writing is another place where I’ve happily welcomed the help. I’ve joked in presentations for years that it’s one of my weaker skills, and I can spend an unreasonable amount of time trying to come up with the handful of words that will sit above an article I’ve already written. Having a brainstorming partner who doesn’t mind generating more options has been genuinely useful, even though most of them never make it onto the page. Sometimes one gets me where I need to go, and sometimes rejecting several of them helps me figure out what I’ve been looking for.
That’s generally how AI fits into my editorial process. I’m not looking for it to hand me finished copy that I can paste into an article. I find it much more useful when I already know what I’m trying to accomplish and can use the tool to get there more efficiently.
The work starts before the transcript
For all the time AI can save me once an interview is over, it hasn’t changed much about how I conduct the interview itself. I’m still listening closely enough to notice when an answer goes somewhere I didn’t expect, and I’m still willing to follow that thread instead of moving immediately to my next prepared question. Somewhere along the way, I’m probably also scribbling “possible lede” next to a timestamp so I can find that moment again later.
By the time I put a transcript into an AI tool, I’ve already started making sense of the conversation. I was there as it unfolded, and the notes I took preserve some of my first reactions to what I heard. Those instincts aren’t always right, so I don’t treat my notes as instructions I have to follow. I’ve abandoned plenty of possible ledes once I started writing because something else turned out to be more interesting.
AI lets me move through the material faster and handle parts of the process that used to take more time. I’m happy to let it help me find the comment I vaguely remember from 35 minutes into an interview, and I’m especially happy to have another source of headline ideas.
The time I save there gives me more room for the work I actually care about: listening closely enough to recognize when someone has given me something worth pursuing, then figuring out how to turn what I heard into a story someone else will want to read.