Improved Relevance Filtering Across Every Source

We shipped a platform-wide improvement to how SocialCX scores mention relevance, reducing false positives across every connected source.

What changed

Short brand names and acronyms are common — and common words coincidentally contain them more often than you'd expect. We tightened the relevance model so that short brand identifiers now require the full tracked keyword to actually appear in context, rather than triggering on a coincidental substring match alone.

In practice, this means mentions that only technically contained a fragment of your brand name — without being genuinely about your brand — are now correctly filtered out instead of showing up as "high relevance" in your inbox.

What you'll notice

If your brand name or a tracked competitor uses a short code or acronym, you may see your mention volume adjust as this rolls out — that's the filter doing its job, not a drop in real coverage. Every mention affected by this change is preserved in your data, just correctly reclassified as lower relevance rather than deleted.

Why we built it this way

We'd rather show you fewer, more accurate mentions than a higher count padded with noise. Relevance filtering is core to how SocialCX earns trust in the numbers it shows you, and this update makes that filter meaningfully more precise.

See how relevance scoring works across your mentions — explore Sentiment & AI Insights or start a free trial.