I read posts before I trust a brand mention count. The count alone does not tell me which questions need a reply.
For a small B2B SaaS company, I’d track a few numbers. I count posts about our brand, repeat problems, and posts that got a reply.
The social listening metrics I keep
I start with posts that name our brand. I remove wrong matches and count the rest for one month. I keep the links so I can check the count.
I note why each person posted. Repeat setup questions go in our help page plan. Fault reports go to product staff, with links to the posts.
The report counts posts that need a reply and posts that got one. I list open problems with an owner so I can see what still needs work.
My first monthly report uses:
- Relevant brand mentions by source.
- Repeated questions or problems, with linked examples.
- Posts that needed action, posts handled, and open items.
I count competitor posts on a separate page. Brandwatch’s metrics guide lists share of voice and sentiment too. Before I use either number, I check how the tool counts posts.
Check the counts against the posts
I save the keywords, sites, and dates with each report. If I add LinkedIn this month, I’ll find more posts. I note the new site before I compare this month with last month.
I open posts from each site and mark copies of the same story. Replies to questions can hold details, so I read those too.
I read the complaints behind any jump in the count. When the same fault appears again, I send the links to product staff. If the post count falls, I check whether we changed a keyword or stopped watching a site.
The social listening report guide shows how I put these counts beside the work still open.
When share of voice and sentiment mislead
Share of voice compares your brand post count with the count for other brands you track. Sprout Social’s help page shows how to work out this share. If I change the list of brands, the share can change too. I save that list with the result.
Tools can read a joke as a bad review. I check the posts behind the label and correct it before the report goes out. Short posts need special care because a few words can change the meaning.
Conclusion
I won’t put a sentiment chart in the report until I’ve read the posts behind it. I want the first page to show repeat problems and the replies we handled. Those are the counts I can check from the source links.
