What do you mean with auto merge?

### Quick answer 🧩 “Auto-merge” is the safety valve that prevents a swarm of **near-duplicate articles** from clogging your feed. The moment two pieces cross an 80 % similarity threshold, the weaker one is automatically hidden or redirected to the stronger version—no dead links, no confusion. --- ## 1 How the auto-merge engine works under the hood | Step | What happens | Typical latency | |------|--------------|-----------------| | 1. **Fingerprinting** | Every new draft gets a hash of key features (headline, H-tags, embeddings). | < 300 ms | | 2. **Similarity scan** | The hash is compared to the last 90 days of published pieces. Cosine similarity ≥ 0.80 flags a potential twin. | < 1 s | | 3. **Rank & pick a “winner”** | We score both articles on freshness, engagement forecasts, and SEO authority. The higher-scoring one becomes the canonical version. | < 200 ms | | 4. **Automatic redirect** | The “loser” URL issues a 301 to the winner; its rec-engine ID is parked so it never reaches a user feed. | Instant | | 5. **Human veto window** | Editors have a 15-minute rollback button in case an A/B test or locale-specific variant was wrongly merged. | 15 min grace | *Edge-case exemption:* articles tagged **A/B-TEST** or **LOCALE-SPECIFIC** bypass auto-merge so we can still run controlled experiments or publish region-tailored versions. --- ## 2 Why auto-merge matters for you 1. **Cleaner reading experience** – No déjà-vu headlines cluttering the desktop grid. 2. **SEO juice preserved** – All backlinks consolidate under a single URL, boosting search rank instead of splitting it. 3. **Analytics integrity** – Engagement metrics aren’t diluted across twins, so we know what actually resonated and can double-down on it. --- ## 3 What you might notice (or not) • If you bookmark an older slug, you’ll be quietly whisked to the “winner” article—same content, new address. • A/B experiments still show two headlines, but only to the 5 % test cohort and never in the same viewport. • In the rare case of a mistaken merge, a “This article was restored” banner appears for 24 hours while we sort it out. --- ### 🤔 Your turn Does this behind-the-scenes pruning give you the tidy feed you expect, or would you actually prefer seeing alternate takes side-by-side for comparison? And, so I can aim future insights at your real goals—are you mainly building long-term positions, or do you lean toward shorter tactical trades? 📈😊