HASHTRACKER

Brand reputation 3 min read

Online reputation management that is not just panic

Online reputation management works when it starts before anything goes wrong. Measure a normal month, set thresholds in advance and learn to tell a person from a wave.

Cover drawn as a line that climbs and falls

Online reputation management is sold as a service and practised as a panic. Something appears, someone escalates, everyone improvises, and nobody writes down what happened.

The version that works is duller and starts long before anything goes wrong.

Tell us what you are exposed to and we will set up the measurement that tells you when to act.

Tell us what you are exposed to

Reputation X Twitter rank
Reputation X Twitter rank

What online reputation management actually involves

Three things, in this order: knowing what is normally said about you, noticing when that changes, and deciding whether a change deserves a response. Most of the effort goes into the first, which is unglamorous and is what makes the other two possible. Without a baseline, every spike looks like a crisis and every quiet week looks like success.

Start by measuring a normal month

Run your brand terms over a period where nothing happened and write down four figures: daily volume, the share of mentions that are replies, how many accounts make up half the volume, and total impressions.

That is your baseline. It takes an afternoon and it is the difference between an alert that means something and an alert everyone mutes by the second week. Redo it every quarter, because normal drifts.

Decide in advance what deserves a response

Write the thresholds down while nothing is happening, because judgement is worse under pressure. A useful rule has two parts: a volume trigger relative to your baseline, and a reach floor.

Volume alone is a bad trigger. A thousand angry posts nobody saw is a smaller problem than fifty that reached a million people, and a policy built on mention counts will send you chasing the wrong one.

Separate a person from a wave

Work out how many accounts produced half the volume. If the answer is one or two, you are dealing with a post and the response is a conversation with a person.

If it took four hundred accounts, replying to individuals will not touch it and the response has to be public and structural. The two situations look identical on a volume chart, which is why concentration belongs in every reputation report.

Know which mentions are yours

Split mention volume into your own posts, the accounts you have a relationship with, and everyone else. A brand where most mentions come from itself and its partners has a distribution list rather than an audience.

That distinction rarely surfaces until someone splits it, and it changes what reputation work even means. You cannot manage a conversation that is mostly your own voice, you can only notice that it is.

Responding, and when not to

The instinct is always to respond. Often the correct move is to let something decay, because a reply hands a small post the audience it did not have. Check the reach before deciding. If the posts driving a spike have low impressions, responding publicly is what will give them reach. If they are already being seen widely, silence reads as confirmation and the calculation flips.

What to keep after it is over

Reconstruct the timeline hour by hour: volume, impressions, the first posts that moved, the account that pushed it past the point of no return, and how long the decay took.

The archive makes this possible after the fact, so even a crisis nobody was tracking can be documented properly. That document is what makes the next response faster, and it is the part almost everyone skips because by then the pressure is off.

Everything we gather stays within the Twitter compliance rules and the dataset comes with the report. Tell us the brand and the competitors and we will build the baseline first.

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