HASHTRACKER

Social listening 2 min read

Social media monitoring starts with one good query

A social media monitoring setup fails in week one or works for years. What decides it is the hour spent on the query: start narrow, widen deliberately, exclude on purpose.

Cover drawn as mentions landing over time

A social media monitoring setup either fails in the first week or works for years. What decides it is the hour spent on the query at the start.

Here is the order that works.

Social media monitoring
Social media monitoring

If you would rather skip the setup, tell us what you need to track and we will build and test the query with you.

Tell us what you need to track

Start with one term and look at what comes back

Before building anything elaborate, run the single most obvious term and read fifty posts. You are not measuring yet. You are finding out what the term collides with, how people actually spell it and whether the volume is ten posts a day or ten thousand.

Almost every wrong assumption in a listening project is visible in those fifty posts.

Widen deliberately, not all at once

Add terms one at a time and check the volume after each.

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

If one addition triples the volume, stop and read what it brought in. A term that looks specific to you often belongs to something else entirely, and you want to know that now rather than in month three.

Tell us what you need to track and we will build and test the query with you.

Tell us what you need to track

Exclude the obvious collisions

Brand names share space with ordinary words, films, other companies and fandoms. Exclusions handle it.

"your brand" -recruiting -hiring -"other company"

Apply them one at a time and always check what disappeared. An exclusion that removes 40% of the volume has either saved the project or destroyed it, and there is no way to know without looking.

Decide on reposts explicitly

This single choice changes every number downstream, so make it on purpose and write it in the methodology. Excluding reposts measures how many people said something. Including them measures how far it travelled. Reporting one while calling it the other is the most common mistake in listening reports.

Most setups want both, as two separate figures.

Set the period and the granularity together

A query without a period returns whatever the tool feels like returning. Bound it.

"your brand"

Then match the granularity to the question. Monthly for trends, daily for campaigns, hourly for events and crises. An hourly chart of a quiet month is noise, and a daily chart of a live event hides everything that happened.

Write down what you did

The query, the period, the exclusions and the repost decision. Four lines. Without them, next quarter's report is not comparable to this one, because nobody remembers exactly what was measured. This is the least glamorous part of listening and the one that decides whether the data is worth anything a year later.

We hand over the query alongside the report and the dataset, gathered within the Twitter compliance rules. Tell us the question and we will put the setup together.

Tell us the question

Choosing a subscription

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

See the subscriptions