Built on the official X API
A report carries 54 metrics across 9 blocks. Every one of them comes from counting, summing, dividing or sorting the columns of the export you download, so you can rebuild any figure yourself. The data is gathered Twitter compliant from end to end.
What it tells you
How many times the posts were seen in total
What it tells you
How far the audience on paper is from the audience reached
What it tells you
The accounts that got most reach per post
The formula next to each metric uses the real column names of the export. It is notation, not marketing.
The numbers that open any report. How much was said, by how many, and how much of it was seen.
9
| Volume | What it tells you |
|---|---|
| Total tweets | How many posts the query returned |
| Unique accounts | How many different accounts took part |
| Total impressions | How many times the posts were seen in total |
| Total likes | Likes added up across every post |
| Total retweets | Retweets added up across every post |
| Total replies | Replies added up across every post |
| Total engagement | Likes, retweets and replies together |
| Potential reach | Followers of everyone who posted, added up |
| Period covered | From the first post to the last one |
Where a report stops being a list and starts saying something. Each one puts two raw numbers against each other.
8
| Ratios | What it tells you |
|---|---|
| Impressions per tweet | How much reach an average post got |
| Impressions per account | How much reach an average account got |
| Tweets per account | How much an average account posted |
| Engagement per tweet | How much reaction an average post got |
| Engagement rate | How many of the people who saw it reacted |
| Amplification | How often a post got passed along |
| Conversation | How much the posts were answered |
| Real reach against potential | How far the audience on paper is from the audience reached |
How much of the conversation a handful of tweets and accounts carry. Worked out by sorting and accumulating.
3
| Concentration | What it tells you |
|---|---|
| Reach held by the top tweets | How much of the reach the biggest posts took |
| Reach held by the top accounts | How much of the reach the biggest accounts took |
| Accounts making up half the reach | How few accounts it takes to reach half the total |
Sortable tables of the tweets that did the work, each with its author and its figure.
6
| Tweet rankings | What it tells you |
|---|---|
| Most impressions | The posts seen by most people |
| Most likes | The posts with most likes |
| Most retweets | The posts passed along the most |
| Most replies | The posts that got most answers |
| Most engagement | The posts with most reaction overall |
| Best engagement rate | The posts that converted views into reaction best |
The same by author. The efficiency ranking is the one that surprises people.
5
| Account rankings | What it tells you |
|---|---|
| Most impressions earned | The accounts that brought most reach |
| Most engagement earned | The accounts that brought most reaction |
| Most active | The accounts that posted the most |
| Most followers | The accounts with the biggest audience |
| Most efficient | The accounts that got most reach per post |
Percentage splits, each with its own impression and engagement breakdown, which is where the value sits.
6
| Composition | What it tells you |
|---|---|
| Originals, replies and quotes | How much came from originals, replies and quotes |
| With and without an image | Whether posts with an image did better |
| With and without a link | Whether posts with a link did better |
| Verified and unverified authors | How much came from verified accounts |
| By language | Which languages the conversation happened in |
| By declared location | Where the people posting say they are |
Granularity follows the period of the report. For a live event this is the main section.
5
| Movement over time | What it tells you |
|---|---|
| Tweets over time | How the volume moved day by day |
| Impressions over time | How the reach moved day by day |
| Engagement over time | How the reaction moved day by day |
| Peak hour and peak day | The busiest hour and the busiest day |
| Hour of day against day of week | Which hours of which days concentrate the activity |
Pure counting, no interpretation. What was named, linked and repeated.
7
| Content | What it tells you |
|---|---|
| Most used hashtags | The hashtags people used most |
| Most frequent mentions | The accounts people named most |
| Most linked domains | The sites people linked to most |
| Most shared links | The exact links shared most |
| Best performing images | The images that got most reaction |
| Most used emojis | The emojis people used most |
| Most frequent words | The words that came up most, minus the filler |
Who took part, aggregated from the author columns of every row.
5
| Author profiles | What it tells you |
|---|---|
| Account age | Whether the accounts are new or long established |
| Follower bands | How the audience sizes are spread out |
| Followers against following | Whether an account is followed more than it follows |
| Lifetime posts | How much the accounts post in general |
| Most repeated locations | The places that repeat most among the authors |
Leaving these out is a decision, not an oversight. Each one needs a model to interpret text, and a wrong label is worse than no label.
Topics and clusters
Automatic topics and clusters. Same reason.
Bot detection
Bot detection. It can be approximated with hard signals like very new accounts posting a lot, but calling an account a bot is a strong claim and it needs judgement.
Estimated reach
Estimated reach. There is nothing to estimate when the API returns the real impressions of every tweet.
Every plan carries the same product. The only decision is how many tweets you need to analyse each month.