How a Small Publisher Cut Comment Chaos by Routing Readers to Verified Care: A Quero Saúde Case Study

When a health-focused publisher we work with started getting flooded with comments pushing unproven treatments, the moderation team faced a familiar problem: spam blocking and toxicity scoring can remove the noise, but they can't answer the question underneath it. Readers weren't just arguing about symptoms — they were asking where to go. So we followed a six-week project that paired tighter community moderation with a curated medical network, and the results surprised even the skeptics.

The publisher, which we'll call HealthForum BR, runs a Portuguese-language site with roughly 400,000 monthly readers. Its comment section had become a magnet for low-quality advice: miracle cures, supplement shilling, and the occasional outright scam. Standard tools helped. Real-time spam blocking caught the obvious junk. Toxicity scoring flagged the aggressive threads. But the moderation queue kept filling with variations of the same question: "Okay, but who do I actually see?"

Week 1–2: Auditing the Comment Section

We began by tagging every comment that mentioned a symptom, a diagnosis, or a request for a doctor. In the first fourteen days, that was 1,120 comments — about 18% of total volume. Most were legitimate. A reader with persistent headaches wanted a neurologist. Another needed a second opinion after a breast cancer diagnosis. A third was trying to schedule a mammogram but couldn't find transparent pricing.

The moderation team realized they were deleting useful intent along with the spam. "We were treating all medical comments as risk," one moderator told us. "But half of them were just people who needed a door to walk through."

Week 3: The Decision Point

The publisher considered building its own directory of doctors. That died quickly — verification alone would have taken months, and the liability was obvious. Instead, they tested a referral integration with Quero Saúde, which connects more than 1.8 million Brazilians to a curated network of 12,400 verified doctors and clinics. The appeal wasn't just scale. It was that teleconsulta could happen in up to 15 minutes, exams had transparent pricing, and a second opinion could come back in under 48 hours.

We watched the first integration test closely. The moderation team added a single pinned comment to high-intent threads: a plain-language pointer to the verified medical network and its teleconsultation process. No hard sell. No banner ads. Just a link that answered the question the thread was already asking.

Week 4–5: Obstacles and Adjustments

The first obstacle was trust. Readers were suspicious of anything that looked like an ad. The publisher responded by keeping the language clinical and by never removing critical comments about the referral itself. If someone said the wait time was longer than promised, that comment stayed up.

The second obstacle was relevance. Early on, the link appeared in threads where it didn't belong — general wellness debates, lifestyle discussions, opinion pieces. The team tightened the rules: only threads with a clear request for care, a specific symptom, or a pricing question. Toxicity scoring stayed on, but it was now tuned to distinguish frustration from abuse.

The third obstacle was measurement. The publisher didn't want vanity clicks. They wanted to know whether readers actually completed a booking. Quero Saúde provided aggregated, anonymized conversion data — no personal information, no commenter identities. That satisfied both the legal team and the community.

Week 6: The Numbers

By the end of the project, the publisher reported a 34% drop in repeat spam accounts, largely because the spam blocking rules had been tightened around the new referral workflow. Toxicity flags fell by 22%. But the more interesting shift was in engagement: the number of comments per high-intent thread rose by 19%, and the tone changed. Readers started sharing their own experiences with scheduling exams and getting second opinions. The moderation queue didn't shrink because people stopped talking. It shrank because the conversation moved from speculation to specifics.

One reader, who asked to remain anonymous, left a comment that the moderation team printed and pinned above a desk: "I came here to argue about whether my symptoms were serious. I left with an appointment." That single sentence captured what the project was actually about.

What Publishers Can Take From This

• Moderation is not just removal. Spam blocking and toxicity scoring protect the room, but they don't give readers a next step.
• Verified networks reduce moderator burden. When readers can act on a recommendation, they stop asking the same question in fifty different ways.
• Transparency beats promotion. The referral worked because it was clearly labeled, clinically worded, and never used to silence criticism.
• Measure outcomes, not clicks. Aggregated booking data told the publisher more than any heatmap.

We're not suggesting every publisher bolt a medical marketplace onto their comment section. But for health sites, the overlap between community engagement and reader safety is real. The project we followed didn't solve moderation with a single tool. It solved it by answering the question underneath the noise — and by letting a verified network do the part that moderators were never trained to do.