You check a Twitter account credibility score by running the account's own numbers through a calculator modelled on the reputation step X described when it open-sourced its recommendation code. Circleboom scores any X account on one screen with its Tweepcred Calculator, working either from a connected account or from ten values you type in yourself.
By hand: you read follower counts, engagement averages, and join dates across three tabs and still finish with a hunch. With Circleboom: the Tweepcred Calculator scores your X account against ten account, safety, and engagement inputs, then returns a number, a rank, and a distribution estimate on one screen.
→ credibility score for your X account
The number estimates X's model rather than reading X's own server-side value, and that distinction changes how you use it.
Why X scores the account before it ranks the post
Ranking on X starts one level above the post. The recommendation code carries a reputation value computed for the account itself rather than for any single thing it publishes. You can rewrite hooks for a whole quarter and still be held by an account-level constraint you never looked at.
Reading how tweets are ranked by the Twitter algorithm explains the pipeline. A Twitter account credibility score tells you where your account sits inside that pipeline before the pipeline ever sees your writing.
So the order most people use is backwards. They diagnose content first because content is the part they control, then get frustrated when better content produces the same flat impressions. Pulling a Twitter account credibility score first tells you whether content is even the variable in play.
The distinction shows up clearly in one common case. An account that has been running follow-for-follow campaigns has a large audience and a following count to match, plenty of posting history, and reach that keeps thinning anyway.
Nothing about the posts explains it. The account-level reading does, because the model looks at the shape of the graph around you rather than the wording of any single post.
What X's open-sourced code actually documents about the score
The score has a public specification, and it is narrower than most write-ups suggest. X's published tweepcred README describes a weighted PageRank run over a graph whose nodes are X users and whose edges are their interactions. That graph is seeded by a user mass file produced by a separate job upstream of the ranking itself.
Three things in that document are worth holding onto.
- The reputation step converts the logarithm of the PageRank value into a number between 0 and 100, which is why the scale is compressed at the top.
- A post-calculation adjustment reduces the rank of accounts that have a low number of followers and a high number of followings.
- The inputs are graph-level, so who interacts with you counts, not only how much interaction there is.
Now the part that matters for anyone shopping for a checker: the README documents the mechanism but does not publish a "good score" cutoff. The 65 figure that circulates in algorithm explainers is a community reading of the code. It is not a line in the specification, so treat any tool that draws a hard pass or fail mark as offering an opinion.
That is also why the follow ratio deserves separate attention. It is the one adjustment the published document names explicitly.
Every other input reaches the number through the graph, where your own behavior is one contribution among thousands. The background in what Tweepcred is and how to increase your score on X covers the wider signal set.
The ten inputs behind a credibility score on X
Whichever route you take, Circleboom reads the same ten values from your X account and scores them together. As an X Enterprise Developer with official API access, it can pull the account fields live on the connected route, so the numbers you would otherwise be recalling from memory arrive measured instead. Seeing them grouped explains why two accounts with identical follower counts can land in different bands.
Five inputs that describe the account
- Follower number, the audience size the graph starts from.
- Following number, which pairs with the follower count to form the ratio the specification adjusts for.
- Tweet number, standing in for posting history and activity depth.
- Join date, which sets account age.
- Verification status, entered as a simple verified or not verified.
Five inputs that describe behavior and safety
- Shadowban status, a toggle, with an in-form link to the check.
- Temporary label status, a second toggle with its own check.
- Average likes per post.
- Average reposts per post.
- Average replies per post, the input most people estimate rather than count.
The split is the useful part. The first group is fixed on the day you run the check and cannot be argued with. The second group is where a wrong entry quietly ruins the result, because three of those five are values you supply from judgment rather than read off a screen.
How to check your Twitter account credibility score, step by step
Screen-by-screen: where the account-quality signals behind a credibility score sit inside Circleboom.
https://www.youtube.com/watch?v=weBd6Gn_Yp0
To check the score on a connected account, sign in to Circleboom, open the Tweepcred Calculator from the Essential Toolbox menu, and pick the connected route rather than manual entry so the account metrics are read instead of estimated. The result screen returns a numeric score on a gauge, a rank, and a distribution estimate. The six steps below run in two phases: connect and read, then verify and save.
Connect the account so the inputs are read, not typed
- Log in to Circleboom Twitter and authorize your X account.
- Open the Essential Toolbox menu, where the free utilities live for signed-in accounts.
- Select the Tweepcred Calculator and choose Connect your X, which the tool marks as the more accurate of its two routes because it works from connected account data instead of numbers you estimate.
Read the gauge, then confirm what the gauge cannot see
- Read the score, the rank, and the distribution line together. The product page's own worked example shows a score of 40 with the rank Rising Star and the note that only around 4 to 6 of that account's posts get considered for distribution.
- Run the Twitter Shadowban Test on the same handle, because a visibility restriction and weak engagement produce the same flat numbers from the outside.
- Download the result or post it to X so the score becomes a dated baseline you can compare against later.
The sequence matters in that direction. The login supplies read access before anything is calculated, and the Essential Toolbox route puts you on the connected version rather than the manual one. The shadowban check comes after the gauge so you can tell a low score caused by weak engagement apart from one shaped by a visibility restriction.
Save the baseline last, because a baseline you cannot date is not a baseline.
Short version of the sequence: connect, read the gauge, confirm visibility, save the result.
Is a credibility score worth acting on?
It is worth acting on as a direction, not as a target. Because no external calculator can read X's server-side value, the honest use of any number here is comparative: your account this month against your account last month, on the same tool, with the same method. That is the same logic behind checking your Twitter score at intervals rather than once.
The five rank bands the calculator assigns give that comparison a shape:
- Emerging User
- Active Contributor
- Rising Star
- Expert Voice
- Elite Influencer
A band change over a quarter carries more information than a two-point move in a week. If you want a second reading built on different inputs, Circleboom's Twitter Quality Score scores your account on its own model. Two tools disagreeing is itself useful signal about how soft any single estimate is.
One habit I would keep: read the engagement inputs against your own history rather than against a benchmark. Knowing what a good engagement rate on X looks like across the platform is less useful than knowing what yours was before the drop you are investigating.
There is a second reason to prefer your own history. Rank bands group very different accounts together, so a niche account with 900 engaged followers and a broad account with 90,000 passive ones can land in the same band for opposite reasons.
The band tells you where the model places you. Only your own trend line tells you which direction you are travelling.
What changes once you have the number
The number turns a vague worry into a short list of inputs. Ten values go in, and each one is either fine, unknown, or a problem you can name.
That is a different working state from "my reach feels bad," and it is the reason the Tweepcred Calculator is more useful as a diagnostic starting point than as a scoreboard.
It also sets the order of your next moves. If the follow ratio is lopsided, that is the input with a documented adjustment behind it. If average replies sit near zero while likes stay healthy, the problem is conversation rather than reach. In that case how influence distribution is calculated on Twitter is the better next read.
For a brand account, the same reading has a reporting use. A dated score and rank, downloaded before and after a cleanup quarter, gives an account audit something concrete to open with. It is a weaker claim than a reach number, but it is honest about being an estimate, which is more than most account-health slides manage.
What the score will not do is promise you a movement. Recalculating after a cleanup shows whether the inputs you changed moved this estimate. It does not tell you what X did on its side, and any tool claiming otherwise is overselling.
Summary
A Twitter account credibility score is an estimate of the reputation value X's own code computes, assembled from ten account, safety, and engagement inputs and returned as a number, a rank, and a distribution note. Circleboom reads those inputs from your connected X account and scores them in one pass, which is what turns a hunch about reach into a list of named variables.
Read it as a baseline, recheck it against itself, and let the follow ratio be the input you act on first, since that is the one the published specification actually adjusts for.
→ score your X account's credibility
Questions the gauge does not answer on its own
Is this the same number X uses internally?
No. X calculates its reputation value server-side across the full interaction graph, and that value is not exposed publicly. Any calculator, Circleboom's included, produces an estimate of that model from the inputs it can see.
Do I have to connect my X account to get a score?
No. The Tweepcred Calculator also runs from manual entry, where you type the ten values yourself. Connecting is the more accurate route because the account metrics are read rather than estimated, but the manual route is there when you would rather not connect during that session.
Which of the ten inputs do people get wrong most often?
Average replies, in my experience. Likes and reposts are easy to eyeball from a profile, but reply counts are the number people round up from memory, and replies are the input most tied to conversation rather than passive reach.
Does a low score mean my account is shadowbanned?
Not on its own. A low score can come from a lopsided follow ratio, a young account, or thin engagement, none of which involve a restriction. The shadowban and label toggles are separate inputs precisely because they answer a different question. Run the visibility check before you conclude anything from a low number, since the two causes call for completely different fixes.
How often should I recalculate?
Often enough to see a trend and rarely enough that the trend means something, which for most accounts is monthly. Weekly recalculation mostly measures noise, since the underlying inputs move slowly and the 0 to 100 scale is compressed by design.
