HASHTRACKER

Social listening 2 min read

Brand mention monitoring without drowning in noise

Brand mention monitoring takes ten minutes to set up and a few iterations to make usable. What to track besides the name, and the monthly figure nobody checks.

Cover drawn as mentions landing over time

Setting up brand mention monitoring takes ten minutes. Setting it up so it does not drown you takes a few iterations, and the difference is all in the query.

The difference is almost entirely in the query.

What to include besides the brand name

The brand name alone misses most of what is said about you. Four additions usually double the useful volume. Misspellings, because people type fast. The @handle, which behaves differently from the plain name. Product names, which often travel without the brand attached. And the shortened forms your customers use, which are rarely the ones in the style guide.

"your brand" OR @yourbrand OR yourbrnad OR "your product"

Tell us what to track and we will build the query before you commit to anything.

Tell us what to track

How to cut the noise

Brand names collide with ordinary words constantly, and the fix is exclusions applied one at a time.

"your brand" -is:retweet -job -hiring

Check what each exclusion removed before keeping it. A term that looks like noise sometimes carries real complaints, and you will only notice by reading what disappeared.

Reposts deserve their own decision. Exclude them when counting how much was said. Include them when measuring how far it travelled.

Separating the three kinds of mention

Most mention streams are three different things mixed together, and they need different responses. Customer service, which is people with a problem. Conversation, which is people discussing you without needing anything. And noise, which is recruiters, bots and coincidence.

Splitting by whether a post is a reply, by whether it carries a link and by author type gets you most of the way there without any interpretation.

Which numbers to report monthly

Volume, unique accounts, impressions and the share coming from your own account. That last one is the check nobody runs. If a large part of your mention volume is your own posts and their reposts, the brand is talking to itself and the number looks healthier than it is.

Spotting something before it becomes a problem

A spike matters when it is unusual for you, which means you need a baseline first. Two or three months of monthly figures is enough. Watch the ratio of replies to original posts. A rise in replies with flat original volume usually means people are answering something rather than starting conversations, and that something is worth finding.

Also watch concentration. A spike carried by one large account is a different situation from a spike carried by four hundred, and they need opposite responses.

Making it defensible

If the monthly number is going to be shown to anyone who decides budgets, it needs a stated query, a stated period and rows someone can check. We gather everything within the Twitter compliance rules and hand over the dataset alongside the report. Tell us the brand and the competitors and we will scope the monitoring around them.

Tell us the brand and the competitors

Choosing a subscription

Every subscription carries the same product. The only decision is how many tweets you need to analyse.

See the subscriptions