How to measure reaction to a launch on X

A launch post gets a burst of attention and then fades. If you want to know how it landed, you need to capture that burst in an organized way, or you'll end up judging the launch by the five replies you happened to see first.

This method works for a new product, a feature, a redesign, a price change or a public announcement. It assumes you collect posts by hand, as described in the collection guide, and score them with the checker.

Before the launch: take a baseline

You can't say reaction was "more positive than usual" without knowing what usual is. In the week before the launch, run your normal brand searches and score them. Note the mention count and the positive, neutral and negative split. If you post regularly, also score the replies under two or three ordinary posts from your account.

This baseline is rough, and that's fine. It keeps you from reading a typical Tuesday as a triumph or a crisis.

Write down what you expect too. "We expect questions about pricing and some complaints about the old design going away" is a prediction you can check later.

Decide what counts as reaction

A launch on X produces several separate streams:

• replies to your announcement post
• quote posts of your announcement
• posts that mention you or your product without quoting
• replies and quotes on posts by other people who covered the launch, such as press, partners or creators

Collect them separately. Replies tend to be questions and direct feedback. Quote posts carry the more public opinions, including jokes and criticism people want their own followers to see. Independent mentions show whether the news traveled beyond your followers.

Collect in time windows

Reaction changes over the first few days. Early replies come from your most engaged followers, who are usually positive. Later replies include people who saw it shared elsewhere, and sometimes a critical wave if the launch hit a nerve.

Split collection into windows, for example:

Use since: and until: in search for the mention stream, and note the time you copied the replies under the announcement. Score each window on its own, then together.

Clean before you score

Remove posts by your team, duplicate posts, and obvious spam. Cap how many replies each account contributes, so one enthusiastic fan or one angry customer can't fill a whole window. The guide on reading sentiment covers this in more depth.

Score, then read

Paste each window into the checker and record the split. Then filter to negative lines and read them all. Do the same for lines flagged as possibly sarcastic, since a launch is exactly when people write "oh great, another redesign".

Tag each negative line with a reason in your spreadsheet. Common launch reasons are price, missing features, bugs, confusing changes, the old version going away, and general dislike of the brand. Do a lighter pass on positive lines to find what people liked most. Skim the neutral lines and pull out the questions.

Build a one-page summary

People who weren't watching the launch want a short answer. A useful one-page summary has:

1. a sentence on overall reaction compared with the baseline, such as "more positive than our usual mentions, with a negative spike on day two"
2. the counts for each stream and window, in a small table
3. the top three reasons for negative reactions, each with a count and one representative post written in your own words
4. the top three things people liked
5. the questions people asked most, which often point to gaps in the announcement or the help pages
6. what you collected and how: queries, windows, accounts removed and the date you collected

Paraphrase rather than paste full posts into internal reports if they might be shared widely, and leave out usernames unless there's a reason to include them.

Watch for things the split hides

A launch can score positive overall while one serious problem is building. Ten replies about a payment bug matter more than a hundred "looks great" replies. Scan for repeated complaints that point to something broken, and pass those to the right person on the same day.

The opposite also happens. A loud negative reaction from people who were never customers can drag down the split while your actual users are happy. Check whether the critical accounts mention using the product. If they don't, the reaction is still worth knowing about, but it is a reputation question, not a product one.

Compare launches carefully

If you launch often, keep the same windows, the same searches and the same cleanup rules each time. Then you can compare one launch with the last. Remember that the topic shapes the mood. A price increase will rarely score like a new feature, and that doesn't mean it went badly. Compare price changes with price changes, features with features.

Keep the raw files

Save the exported CSV for every window, along with a note of the queries used. When someone asks six months later how the redesign was received, you can answer from evidence instead of memory.

Questions

How soon after a launch should I measure?

Take a first look within a few hours to catch problems early, then do the main measurement after 48 to 72 hours, when most of the reaction to a typical post has come in. Repeat once a week or two later if the launch is large.

What if the reaction is mostly neutral?

That often means people are asking questions or tagging friends. Read the neutral lines. A neutral-heavy reaction full of "how much is it?" and "does it work with X?" tells you your announcement left out information people wanted.

Twitterment is independent and not affiliated with X Corp. Features and search operators on X change from time to time; if something here doesn't match what you see, check X's own help pages.

Open the sentiment checker