Rebuilding Blog Production for a 70-Client Book

Blog production was the bottleneck the whole content operation felt. There was no consistent system for it, the work was split across too many hands, a single post could take hours, and deadlines slipped further behind as the client book grew.

I built the system that replaced that process. I now run blog production for the entire book of more than 70 clients myself, ahead of schedule, work that used to take a team.

Results at a glance

The Challenge

Before I built the system, blog production had no reliable shape. It moved through several roles, a single post could take hours, and the schedule fell further behind as the client roster grew. It was the slowest stage in the pipeline and the one most likely to run late.

The clients are interventional-psychiatry practices, so the writing carries real regulatory weight. Off-label indications, device-clearance wording, and controlled trademark usage each have a correct form, and getting one wrong is a compliance problem, not a typo. Any faster way of producing this content had to hold that line at speed, across the whole book, every time.

The Approach

Build the quality in, so scale never lowers it

I decided early that at this volume the checks had to live inside the system, because the final human read is the first thing that gets squeezed when the schedule is tight. Three failure modes worried me most, and I designed the system to make each one impossible.

1. Live-crawl links

   Every internal link comes from a live crawl of the client's own site.

2. Source verification

   Every source is checked against the published record.

3. Compliance enforcement

   Off-label language, clearance wording, and trademark usage are enforced on every draft.

4. My approval

   Nothing reaches a client until I review and approve it.

I had seen what a dead or wrong link does in a healthcare blog, so I set the system to draw every internal link from a live crawl of the client's own site. It can only use pages that actually exist. It cannot invent one.

A citation can look authoritative and still point at a study that does not support the claim, so I made verification non-negotiable. Every source is checked against the published record before anything ships, and nothing rests on a reference that only looks right.

The regulatory rules that carry the most risk, off-label language, clearance wording, and trademark usage, are the ones I made the system enforce on every draft. I did not want compliance to hang on whether a reviewer was fresh at the end of a long day.

Find the failure that design did not anticipate

Some problems only surface at volume. A compliance rule I had written, the one that forces a brand-name treatment into its required all-caps form, was also rewriting that same word inside real client URLs, turning links that were already correct into links that pointed nowhere. The system flagged them as broken, and no revision could clear the flag, because the copy had been right before my own rule touched it. I traced it to the text-repair step reaching inside link addresses, and I fixed it so a correction can never modify a URL. Nothing published with a broken link.

Keep the final call mine

I set the system to do the production work and to leave the judgment to me. It flags problems; it does not overrule them. Nothing reaches a client until I have reviewed and approved it. That gives me the speed of an automated pipeline and the accountability of a human editor in one workflow.

The Results

The bottleneck is gone. What used to move through several roles and run late now runs through me alone, and it comes in early. I delivered a full quarter of blogs across the entire book about two months ahead of schedule.

The system I built supports more than 630 client blogs a quarter, at roughly a dollar per campaign. A post that once took hours is now one of hundreds I carry in a quarter, for a roster that used to take a team. Consistency and compliance hold across all of it, because I made the system enforce them on every draft.