The Art And Craft Of Feature Writing

Getting the Structure Right

The hardest part about feature writing isn't finding the angle. It's keeping the reader engaged when the subject matter could easily become a dry list of specifications or a puff piece for a product manual. I learned this the hard way about five years ago when I was assigned to write a 1,200-word feature on a mid-tier project management software company. The CEO gave me the press release, the feature sheet, and three client testimonials. I sat down to write a standard inverted pyramid piece and ended up with something that read like a press release with more words. My editor cut it in half and told me to talk to the people who actually used the software, not the people who sold it. That's the core of The Art And Craft Of Feature Writing: you're not documenting facts. You're reconstructing a narrative from raw material that rarely comes organized in any useful way. The difference between a feature and a news report is that a news report tells you what happened. A feature makes you understand why it mattered, or at least gives you enough texture to care.

The Art And Craft Of Feature Writing

Beginners treat features like extended news stories. They open with the lead, stack the facts, and hope the ending lands with some emotional weight. That approach works for quick turnaround pieces on predictable events. It falls apart when you're writing about something nuanced, counterintuitive, or culturally specific. The feature format demands a different architecture. Start with a nut graph. This is the paragraph that tells the reader explicitly what the piece is about and why they should invest their time. News writers often skip this because the headline and lead do the work. Feature writers can't afford to be lazy about it. Without a nut graph, you've got atmospheric prose that floats above empty content. I usually draft the nut graph last. Writing the rest of the piece first forces me to clarify what I'm actually arguing before I summarize it. If I can't nail the nut graph after I know the full shape of the story, I don't know the story well enough yet.

How I Actually Approach a Feature

My process isn't linear. I start with background research, usually two to three days of reading, browsing forums, watching interviews, and skimming annual reports or academic papers depending on the subject. This phase alone determines whether a feature is going to be shallow or substantive. Most writers rush this because deadlines pressure them into interviewing and drafting faster. I've found that spending an extra day on research typically saves three days of rewriting because you stop hitting walls where you realize you don't understand a key concept your source is describing. After research comes the interview phase. For features, I conduct at least three to five interviews, mixing primary sources with secondary experts who can provide context. I record everything. Transcription services run about $1.50 per minute of audio, which means a two-hour interview costs roughly $18 to transcribe. Doing it yourself takes about four hours of work and is almost always less accurate. I use Otter.ai for rough drafts and then manually correct the transcript against the recording. This combined approach usually takes me about three hours per two-hour interview. The drafting stage is where most features go wrong. Writers produce 3,000 to 5,000 words and then cut aggressively, often removing the strongest material because it doesn't fit the structure they've imposed. I write the first draft backward. I start with the ending, then the middle, and only then write the opening. This feels counterproductive but it prevents the common problem of a strong lead that the body can't support. When you write the ending first, you know exactly where the piece is going and every paragraph earns its place by moving toward that destination.

A Specific Problem I Encountered

Last year I was writing a feature on a regional water authority that had been trying to modernize its infrastructure for eight years. The story had genuine tension: budget constraints, political interference, aging pipes, and a community that barely noticed the problem until three neighborhoods lost water pressure for a week straight. I had all the data. I had quotes from engineers and residents and council members. The piece was technically complete at 2,800 words and completely unreadable. The problem was structural. I had organized the piece chronologically, which meant the conflict built too slowly and resolved too abruptly. The nut graph appeared on page two instead of page one. The human element was buried under municipal meeting transcripts and budget line items. I spent six hours rebuilding it from a timeline structure to a thematic one, grouping sections by the actual stakes involved rather than the order in which events occurred. The revised version came in at 2,100 words and ran in the Sunday magazine section. The key insight: chronological organization is the default structure. It's also usually the wrong one for features.

What Beginners Miss

Most aspiring feature writers focus on voice and style. They practice writing beautiful sentences and develop a distinctive tone. This matters, but it's secondary to structure and sourcing. A beautifully written feature with weak research will disappoint readers within the first two paragraphs. A solidly structured feature with plain prose will hold attention because the ideas are clear and the evidence is convincing. Another common mistake is over-quoting. Sources say things. Your job is to summarize, paraphrase, and selectively quote. If a source's exact words are sharper than anything you could construct, use the quote. If you're using a quote to fill space or add authority, replace it with your own synthesis. I aim for a quote-to-prose ratio of roughly one to three. Every quoted sentence should earn its slot. Roughly 60 to 70 percent of my draft word count should be my own writing, not my sources'. When that ratio flips, the piece loses its authorial voice and becomes a collage of other people's thoughts.

The Reality About Feature Writing Tools

There's no single tool that handles feature writing end-to-end. What I use day to day is a combination of transcription software, a reference manager, and a drafting environment that lets me rearrange sections without rewriting. Notion and Scrivener both handle the organization piece. Notion is faster for research aggregation because you can paste articles, highlight passages, and tag sources in seconds. Scrivener is better for maintaining a clear view of the complete manuscript structure, especially on longer features over 3,000 words. For transcription, I've tested Descript, Rev, and Otter. Descript's editor interface is clean and lets you edit text to edit audio, which speeds up quote extraction. But the accuracy drops noticeably with heavy accents or overlapping speech, which happens constantly in real interviews. Rev's human transcription is accurate to about 98 percent but costs $25 per hour of audio and takes 24 hours. Otter sits in the middle at about $8 per hour with roughly 93 percent accuracy on clear speech and 85 percent on muddled recordings. For a typical feature project with four interviews averaging 90 minutes each, I use Otter for the base transcript and spend about two hours correcting it manually. I also keep a spreadsheet tracking every source, their contact information, what they said, and which section of the piece it supports. This sounds administrative but it prevents the specific failure mode where you discover three days before deadline that you have no direct quote supporting a key claim in your third section. The spreadsheet usually takes me about 45 minutes to set up and another 30 minutes per interview to populate. The time investment pays off during the revision phase when you're searching for a specific quote or verifying a detail under pressure.

Where This Approach Breaks Down

Writing features this way assumes access. You need sources willing to talk, subject matter that hasn't been exhaustively covered, and editorial patience for a process that typically takes two to three weeks for a 1,500-word piece. If you're working on tight turnarounds, daily news cycles, or topics where primary access is blocked, the method slows you down considerably. In those scenarios, a hybrid approach works better: use the nut graph and thematic structuring but compress the research and interview phases into a single day. You'll get a thinner piece, but it will still be structurally sound. The backward drafting method also doesn't work well for event-based features where the narrative arc is inherent in the sequence of events. A concert review, a race report, or a political convention recap benefits from chronological progression because the experience itself unfolds in time. For these, stick to a forward structure but still write the nut graph last and the ending first. That small adjustment prevents the lead-body mismatch without forcing you into an unnatural organizational framework. There's also the problem of source reliability. Feature writing depends on people telling you what happened and why. Human memory is fallible. People lie. Sources omit details that don't serve their narrative. I verify any factual claim that could be disputed by cross-referencing at least two independent sources. If a source describes a meeting, a decision, or a financial figure, I find documentation or an independent witness. This verification step adds roughly 15 to 20 percent to the total production time but prevents the embarrassment and credibility damage that comes from publishing an unverified claim in a feature.

Practical Takeaways

The single most impactful change you can make to your feature writing is restructuring your process. Stop leading with the opening paragraph. Draft the ending first, identify the nut graph after you know where you're going, and organize thematically rather than chronologically unless the event itself demands linear progression. Track your sources in a simple spreadsheet. Aim for one quote per three paragraphs of your own prose. Verify disputed claims with a second source before including them. These habits won't make your writing more interesting on their own, but they'll make it more reliable and more coherent, which is the actual foundation of a feature that holds attention from the first line to the last.