Skip to main content

Social listening

Most social listening tools count mentions. A number goes up, you get an alert, and you're left doing the actual work yourself: reading the thread, figuring out who started it, deciding whether it matters.

Counting is the easy half. This playbook is about the other half — what was said, whether it's spreading, and whether you need to do anything.

Before you start​

Social sources in Kinn are public content you point at rather than accounts you connect: YouTube, TikTok, Instagram, X, and Reddit.

That's a real difference from Discord or Steam. You're not connecting a firehose — you're choosing what to watch. Which means step one is making sure you're watching the right things.

1. Find what you're missing​

You can't analyze what isn't connected, and the conversations that matter most are usually the ones nobody on your team happened to notice.

Find new YouTube videos about our game that we haven't connected yet. Sort by newest.

Search Reddit for discussion of our app from the past month that we're not already tracking.

Kinn returns results you haven't indexed, with enough detail — title, creator, date, engagement — to judge what's worth bringing in. When something looks relevant, ask Kinn to connect it and it starts pulling in the comments.

See Discovering new content. Run this monthly at minimum. A creator with 200k views who you've never indexed is a blind spot, not a small one.

tip

Do this before concluding there's no data. A thin answer often means you haven't connected the conversation, not that it isn't happening.

2. Follow whole channels, not one video at a time​

The pattern teams get the most out of: rather than adding videos one by one as someone notices them, point Kinn at an entire body of video content and let it keep up.

On YouTube you can connect at three levels and mix them freely:

  • A channel — your own, or a creator's, so every new video's comments are picked up as they publish
  • A playlist — a launch series, a tutorial set, a campaign collection
  • Individual videos — when you only care about a few

Connecting your own channel means every tutorial, trailer, and devlog you've ever posted becomes one searchable body of feedback:

Across all the comments on our channel, what do people most often get stuck on? Which videos generate the most confusion relative to their view count?

That second question is a genuinely hard one to answer by hand and a genuinely useful one to have answered. A video with a high confusion-to-views ratio is either explaining something badly or documenting something that's badly designed.

Volume tells you something too. Comment volume across a whole channel or playlist is a decent proxy for which topics people care about — and where it's rising or falling over time:

Chart comment volume across our channel by month for the last year, and tell me which topics are growing.

tip

Tracking a whole channel pulls in older, low-traffic videos as well. That's usually what you want for a long-running body of content — but if you only care about a launch or a handful of key videos, index those directly to keep the signal tight.

3. Read the comments, not the post​

The video is marketing. The comments are the feedback. That's where the actual signal lives, and it's the thing mention-counting tools consistently miss:

For the three most-viewed videos about us from the last month, what are the comments actually saying? Group by theme and tell me which are positive, negative, or asking questions.

The "asking questions" bucket is quietly the most valuable of the three. A recurring question in comments is usually a documentation or onboarding gap — the cheapest kind of problem to fix.

4. Understand a spike before you react​

When volume jumps, the instinct is to respond immediately. Ask first:

Mentions spiked on TikTok in the last three days. What's driving it — is it one creator, a specific clip, or lots of independent posts? Is the sentiment positive or negative, and is it spreading to other platforms?

Those three questions separate the cases that need completely different responses:

  • One big creator — a single conversation to have, not a crisis.
  • Many independent posts — something real is happening; find out what.
  • Spreading across platforms — it's got legs. Act today.
  • Contained to one platform — watch it, don't panic.

5. Track sentiment, not just volume​

Volume tells you people are talking. Sentiment tells you whether that's good:

Chart sentiment across our social sources over the last 90 days, and mark where the 2.4 release landed. Where the line moves, tell me what changed.

That last clause matters. A sentiment chart without explanation is a pretty picture; a sentiment chart where Kinn tells you why each move happened is a decision.

6. Compare social against your owned channels​

Social skews casual and creator-driven. Your Discord skews invested. Your store reviews skew angry. Reading one in isolation misleads:

Is the complaint about the new monetization showing up in Discord and Steam too, or mainly on social?

A social-only complaint often means people reacting to a clip rather than to the product — a real perception problem, but a different one from a real product problem, and it needs a different fix.

7. Make it recurring​

Every Monday at 9am: summarize what's being said about us across our social sources over the last 7 days. Group by theme, note any spike and what drove it, and flag anything that also appeared in Discord or Steam. If nothing notable came up, say so in one line.

See Scheduled tasks.

If you're running creator or influencer campaigns, the questions are different enough to deserve their own routine — Kinn reads performance metrics and comment sentiment for the same videos, which is what lets you rank creators by what their audience actually said rather than by reach alone.

See Measure creator campaigns.

Common mistakes​

Only watching what you already connected. The whole point of discovery is that your blind spots are, by definition, invisible. Make it a monthly habit.

Treating a mention count as a finding. "Mentions are up 40%" is not information until you know what was said and who said it.

Assuming a spike is about the product. Sometimes it's a meme, a creator's personal drama, or a completely unrelated story your name got attached to. Ask what's driving it before you respond.

Ignoring non-English conversation. Kinn matches across languages, so describe the problem rather than the exact English phrasing you expect and you'll pick up discussion your team can't read.

Next​