A useful chart should explain what it rewards, what evidence it uses and where editorial judgment enters the picture. Readers should be able to distinguish a measured result from a selection made by the editors. Creators should know what makes a character eligible and how to question a result.
That is the purpose of the Chroma Chart. We begin with an editorial watchlist of five AI and virtual personalities. Future editions may rank eligible characters using comparable performance data and documented editorial assessment. Those rankings will require a published scoring specification, verified sources and editorial review. We have not announced a measurement window or a date for the first numerical ranking.
The Index, the watchlist and the rankings
The Index is our directory of AI and virtual influencers. It brings together profile information, public handles, categories and editorial context. The original homepage listings also show dated Instagram follower counts, ordered within categories. Those historical listings are separate from a Chroma Chart score.
Being listed does not mean that every performance metric has been verified or that the character automatically qualifies for a ranked edition. A profile can be useful even when comparable measurement data is unavailable.
The inaugural watchlist is a different editorial product. Granny Spills, Lil Miquela, Milla Sofia, Aitana López and Imma represent five approaches to virtual identity. Their display order is not a ranking. Granny Spills receives the cover treatment as an editorial feature, not as a declaration of first place.
A future measured edition will ask how eligible accounts performed during a documented period. Audience size alone will not answer that question. Growth, engagement, publishing activity and editorial assessment all need definitions that readers can inspect.
Who can be considered
We cover both generative AI personalities and CGI virtual influencers, and label the distinction. A character does not need to be photorealistic, female or built with one particular tool to fall within our editorial scope.
For a future Instagram-based ranking, the character must have an Index profile, an identifiable public primary account, activity during the measurement period and enough verifiable data for a fair comparison. The final scoring specification will state the minimum history and qualifying activity required. Watchlist inclusion does not waive those requirements.
- One character, one primary account. We will identify the account being measured. Secondary, regional and network pages will not be added together to inflate that character’s audience. Any account changes will be documented.
- Repost, fan and collab pages are outside the competitive ranking. A distribution page is not the same thing as the character whose work it shares. This includes Cybeauty One’s own collab pages.
- Cybeauty-owned or operated characters are excluded from competitive rankings. That includes Naomi Marie. We cannot independently grade our own work.
- Insufficient data means no measured placement. An unknown value will not become zero, and missing history will not be replaced with an estimate simply to complete the chart.
Creators can submit their characters for consideration. Submission does not guarantee a profile, editorial coverage, watchlist inclusion or a ranking position.
The four areas we will assess
Our published criteria cover growth, engagement, consistency and originality. They are a framework for developing the ranking, not a complete scoring formula. Some inputs are measurable; others require an editor’s judgment. Both kinds of assessment must be explained.
| Area | What it examines | Evidence and judgment |
|---|---|---|
| Growth | Audience change relative to the starting audience | Comparable dated follower snapshots |
| Engagement | Public likes and comments on qualifying posts | Documented post sample, observation time and comparison group |
| Consistency | Publishing activity and continuity of character identity | Posting records plus a documented editorial assessment |
| Originality | Concept, storytelling and creative execution | An editorial rubric, reviewer notes and examples |
Growth: change relative to the starting audience
The basic growth calculation is the change in followers divided by the starting follower count, multiplied by 100. Gaining 20,000 followers from a starting audience of 100,000 is 20% growth. The same increase from 2 million is 1%. Both are real increases; they describe different rates of growth.
That calculation alone does not produce a fair score. The final specification must explain the measurement boundaries, minimum starting audience, treatment of zero or negative growth, outliers and conversion from growth rate to points. Small accounts can show large percentage changes, so the comparison rules matter.
Unusual growth patterns will trigger review, not an automatic accusation. A follower spike without a matching change in visible likes or comments does not by itself establish manipulation. Viral distribution, publicity, audience behavior and reporting delays are possible explanations. Any adjustment or exclusion will need documented evidence and a stated reason.
Engagement: the same evidence standard for everyone
Our starting point is public likes and comments on qualifying posts. Before assigning scores, we will publish the post-selection rules, calculation, audience denominator and comparison groups. We will also explain how different formats, paid distribution, collaborative posts, pinned posts, hidden counts and unavailable or deleted content are handled.
Collecting every account on the same day does not ensure comparable engagement data. One tracker might average a different set of posts or use a different update time. A post published yesterday has also had less time to collect reactions than one published weeks ago. The scoring specification must address those differences.
Saves, reach and other private insights will not affect a common score when equivalent evidence is unavailable for every account being compared. A creator may voluntarily supply insights for context or a correction request. We will verify what we can and label any supplementary information separately. Supplying private analytics will not buy a scoring advantage.
Consistency: publishing activity and editorial continuity
Posting frequency and gaps in activity can be measured. Whether a persona maintains a recognizable identity is an editorial assessment. These are separate parts of consistency, and we will not describe both as purely numerical.
Visual continuity does not mean that a character must look identical forever. An intentional redesign, a change of medium or an evolving artistic style should be assessed in context. We will distinguish deliberate creative choices from unexplained inconsistencies, and document the reasoning with examples. Different production methods should not be treated as interchangeable technical tests.
Originality: an explained editorial judgment
Editors will consider the character’s concept, storytelling and creative execution. What gives the persona a point of view? How does the work develop that idea? Is the craft serving an identifiable creative purpose?
These questions need a published rubric and reviewer notes, not an unexplained preference. Originality is not the only area involving judgment: the character-continuity component of consistency does too. Any future numerical contribution from either assessment must be disclosed in the approved scoring specification.
What must be settled before we publish ranks
We have considered a provisional 100-point framework: 35 points for audience growth, 30 for engagement, 25 for consistency and 10 for originality. Those weights are not approved, are not in force and have not been applied to Edition 01. This article does not activate them.
Before a measured edition can be published, its versioned specification must define:
- The eligible account pool, minimum activity and minimum data history.
- The measurement start and end, timezone, capture schedule and observation cutoff.
- The qualifying posts, content formats and treatment of paid, collaborative or unavailable content.
- The approved weights, normalization rules and any audience-size bands or other comparison groups.
- The treatment of missing data, outliers, zero starting audiences and negative growth.
- The editorial rubric, review process, tie-breaking rules and correction procedure.
Each ranked edition will identify the methodology version it uses. A change to the method must be explained and dated. We will not silently rewrite an earlier edition’s rules or manufacture historical data to make a comparison possible.
Where the evidence comes from
Public trackers such as Social Blade and CreatorDB, together with accessible public posts, can provide useful evidence. Availability, calculation methods and historical coverage vary. Social Blade notes that its history depends on when an account entered its system; a tracker should not be assumed to hold a complete record from the account’s creation.
For every measurement used in a ranking, we will record the source, relevant account, capture timestamp and timezone, underlying measurement window or post sample, and verification status. We will distinguish when we retrieved a value from when that value was measured or updated. Where a source does not disclose enough information, we will flag the limitation or omit the measurement.
Cross-checking means investigating differences, not choosing whichever provider gives the more flattering number. If sources disagree materially, we will retain the references and explain the resolution. We will not average incompatible figures simply to fill a gap.
Some historical snapshots may not be reproducible through a later live lookup. We will preserve supporting records and publish the source references and dates needed to understand a result. We cannot promise that every reader will see the same current value on a changing profile.
Edition 01’s profile identity checks are not performance measurements. It has no comparative metrics dataset and no Chroma Scores. Its editorial selections should be read on that basis.
What cannot influence a competitive result
Network-wide audience totals, unverified marketing claims, earnings assertions and a chatbot’s unsupported answer are not substitutes for evidence about the account being measured. Historical importance belongs in our reporting and profiles; it is not evidence of this month’s performance.
Commercial work and editorial selection must remain separate. A paid collaboration does not automatically qualify a character for The Index, a watchlist or a competitive chart. Sponsored features will be identified, and sponsorship will not improve a score. Sponsored placements must remain separate from competitive ranking entries.
Offers of promotion, personal relationships, pressure or threats cannot change a score or create an entitlement to coverage. Cybeauty-owned characters and our collab pages remain excluded from competitive rankings. Relevant editorial conflicts will be disclosed.
Questions, corrections and a record of changes
If you believe a profile or future chart measurement is wrong, email [email protected] with the account, edition or page, disputed value, relevant date and supporting source. A dated tracker reference or clearly identified account insights will help us investigate. Please omit passwords, access tokens and information unrelated to the issue.
We will review the evidence and explain the outcome when we can establish it. We are not promising a fixed response deadline. A request for a correction is assessed on its evidence and is not a request for paid placement.
Material corrections will show the previous value, corrected value, reason and correction date. If a correction affects a score or position, we will identify that effect. Original and replacement measurement records will be retained so the change is traceable.
The same principle applies to the method itself. Our public methodology page will identify the current version and explain material changes. This version clarifies that private insights are supplementary, that consistency includes editorial judgment, and that numerical weights and normalization remain pending.
For now, explore the inaugural watchlist as an editorial selection. A measured ranking will follow only when its evidence and rules are ready to stand alongside it.
References and current documents
- Chroma Chart — Edition 01: The Inaugural Watchlist
- Chroma Chart public methodology and independence safeguards
- Social Blade: historical data coverage
- CreatorDB: profile and performance information
Published November 2, 2026. These are editorial and ranking principles, not an approved numerical scoring specification.