What makes a
little space valuable?

Which X analytics do we import?

Creators can request a snapshot of up to 200 original posts from the previous 28 days. We exclude replies and reposts, deduplicate posts, and retain aggregate counts rather than post text. Available measurements include follower count, median post impressions, engagement rate and profile clicks driven by sampled posts. Private click metrics appear only when X returns them for the connected owner.

Metrics are cumulative totals on the sampled posts at import time. A snapshot is not a count of all activity during that calendar period, and overlapping snapshots are never added together. Incomplete samples and unavailable metrics are labelled.

Do post impressions measure banner views?

No. A post impression is a view of a post. Profile clicks on posts are a signal of profile interest, but do not count all profile visits or prove a banner was seen. API post impressions may include promoted impressions. We do not turn these measurements into a promised number of banner views.

How are activity rankings calculated?

Creators who opt in can appear in the post-impression ranking. It orders profiles by median impressions per sampled original post, using a snapshot from the last seven days with at least five posts that include impression data. Missing analytics do not receive invented scores. Follower and sponsorship-spend rankings remain separately available.

How does the banner price estimate learn?

The first model uses comparable marketplace transactions. It considers fully paid, unrefunded placements after the entire booked period, the creator’s delivery claim, a clear delivery review and a confirmed transfer to the creator. The approved publication must belong to the booked X account and fall within its booked dates. A delivery claim is an attestation; it does not independently prove continuous banner display. Open delivery issues, uncertain transfers, bank disputes, examples and older payments without this completion record are excluded. A later refund or bank dispute removes a previously qualifying sale.

Historical audience snapshots must be from the 30 days before the auction closed, so later growth cannot rewrite the audience evidence. The model considers at most the 1,000 most recently paid qualifying sales from the previous 180 days. Comparables must match the audience category and sponsorship duration, have a pixel area within 10%, and have a follower count between one quarter and four times the current creator’s count. The model ranks them by similarity in logged follower count, median post impressions and engagement rate, and uses up to 30 nearest matches from the previous 180 days.

A repeat creator–buyer pair contributes only its latest qualifying sale. An estimate requires at least five sales across three creators and three buyers. We display the median and the 20th–80th percentile range. These are comparable sale prices, not a statistical guarantee about a future auction. Each new qualifying sale can update the estimate.

Why might I see “Learning the market”?

The estimate tells you whether it needs a fresh analytics import, more measurable posts, more completed sales, or more distinct creators or buyers. A brand-new market needs actual outcomes before a dollar estimate is defensible. Creators can still choose their own starting price. Early and developing data labels describe sample maturity, not predictive accuracy.

Who can see my account analytics?

Your analytics workspace is private by default. You can choose to publish summary metrics on your creator page and activity rankings, and contribute qualified outcomes to aggregate price estimates. Daily refresh is optional and requires an active marketplace scheduler. Disconnecting X deletes stored analytics and removes them from public rankings and future comparisons.

Can I verify Premium analytics without sharing exact stats?

TLSNotary is the required route for verified-analytics opportunities and the future valuation model using authenticated private claims. Standard accounts can continue to list and bid using baseline audience signals. Verification establishes eligibility; actual proven signals and buyer demand determine price. That is the intended purpose of Private proofs mode. Selecting it stops API analytics imports, deletes saved analytics and disables public metric sharing. The cryptographic verification path is a research prototype and is not active.

A zero-knowledge proof can establish a statement such as “at least 10,000 profile visits” while hiding the exact value. It must also prove that the private value was authenticated by X for the right account and dates. An editable, unsigned CSV does not establish that provenance.

The candidate design uses a local authenticated TLS session, a hidden commitment to the relevant response data, and a separate proof of the threshold. The marketplace would receive only the public claim and proof. Public bands reveal a minimum or range by design, so repeated arbitrary threshold queries must be prevented. Exact rankings of hidden numbers require an additional private-computation design.

A ready-made X analytics proof provider and audited verifier have not been integrated. No private-proof badge is issued by this version.

For local source research, download the offline analytics inspector. Extract the download and open its HTML file on your own computer, then select a sanitized HAR capture there. It can show response values locally and save a restricted structural report. Keep the original capture private. An editable capture and its report do not prove X provenance or unlock verification.

What could improve valuation over time?

More completed campaigns, repeat buyer behaviour, category demand, price stability and measured campaign outcomes can support future model revisions. The current method is identified as comparable-settled-placements-v2. We do not claim a trained prediction model or banner attribution system before those capabilities have been validated.