Limited data confidence means ComicPilot found enough reliable information to return a basic, explainable recommendation, but some identity, edition, run, or relationship evidence is incomplete, ambiguous, or conflicted. It is a warning about the boundaries of the evidence. It is not a percentage chance that the recommendation is correct, a low collection-fit score, or an automatic Don't Buy.
The practical response is to read which facts were trusted, which were excluded, and which were unavailable. Then ask whether the uncertain fact could change the decision. An unresolved cover matters greatly to a variant collector. A missing story arc does not undo a collector-confirmed exact duplicate.
Keep three different answers separate
The collection-first buying guide separates three ideas that are easy to confuse:
- Collection Fit asks how strongly the comic connects to your collection, goals, wishlist, ratings, and trusted relationships.
- Buy Decision asks whether to buy this comic at the entered asking price, after duplicate and Not Interested vetoes and personal price guardrails are applied.
- Data Confidence describes how much of the identification and supported metadata was trustworthy enough to use.
A comic can therefore have strong collection fit and a Maybe buy decision because the price is above your limit. It can have a positive buy decision with Limited data confidence because the reliable evidence is narrow. It can also receive Don't Buy with High confidence when a confirmed local record proves an unwanted exact duplicate.
Data confidence is shown as High, Medium, or Limited, not as a percentage. The collection-fit percentage is not a hidden confidence score. Neither label estimates grade, condition, authenticity, physical completeness such as missing pages, market value, rarity, or future price.
What do High, Medium, and Limited mean?
High means the candidate identity and evidence needed for the important claim are strong enough for the app's highest confidence band. It does not mean every possible creator, story, event, cover, or relationship field is present. A collector-confirmed exact duplicate can be High even when richer provider metadata is unavailable, because storyline data is not required to prove that ownership match.
Medium means useful identity and metadata are present, but the result does not meet the stricter High boundary. A non-identity provider conflict can also leave the usable evidence in this band when the disputed field is excluded. Read the detail panel before deciding which claims remain supported.
Limited means the available evidence supports only a narrower recommendation. That can result from ambiguous identification, an unresolved run or edition, a trust-critical collection warning, conflicting identity data, or sparse metadata. ComicPilot still returns reasons; it does not hide uncertainty behind an unexplained score.
These bands are deterministic outputs from the same versioned rules for the same input. They are not calibrated odds such as “80% likely to be right.”
Why would a recommendation be Limited?
Common causes include:
- the issue candidate or variant remains ambiguous;
- publisher, series year, or volume is missing where two runs may share a title and issue number;
- the issue is known, but the exact cover or printing needed by the collector's variant policy is unresolved;
- a saved collection record lacks the run or edition detail needed for an exact comparison;
- a manual or unmatched imported record contains only title and issue;
- enough supported relationship or preference metadata is unavailable that the remaining evidence is sparse; or
- two metadata sources conflict on a field that could not be safely reconciled.
One absent optional field does not automatically produce Limited. The label reflects the remaining trusted evidence together with any ambiguity or trust-critical warning.
The comic-identification guide explains why title search, a barcode, and cover-text recognition produce candidates at different levels of certainty. A barcode can identify an issue family without proving the exact edition. The PWA cover-photo beta proposes editable text hints; it does not match artwork or establish a cover, printing, grade, or value.
A new or obscure comic missing from catalog search can still become a manual title-and-issue candidate. The absence of a catalog result is not evidence that the comic does not exist. It means the available lookup did not return a candidate safe enough to claim.
What can remain reliable when other data is missing?
Confidence should follow the specific claim. ComicPilot does not require story metadata to warn about a collector-confirmed exact duplicate, and it does not require a cover image to compare a confirmed title, run, issue, and edition already stored locally.
That creates an important distinction:
- “You already own this confirmed exact edition” can remain decisive without an event or story arc.
- “You appear to own this issue, but the edition is unresolved” supports a possible-duplicate warning, not an exact-edition claim, when variants are unique for your collection.
- With variants not treated as unique, a confirmed same-issue match can count as already owned without pretending the exact cover was established.
- “No story relationship is available” means the app cannot use that relationship; it does not prove that no relationship exists.
The duplicate-or-variant guide shows which physical details to compare before relying on an edition-level warning. If an exact local duplicate is confirmed, ordinary positive signals such as wishlist, price, character, or story interest do not override it.
Offline status alone does not force Limited. A previously confirmed identity and reliable cached ownership record can still support High confidence for an exact duplicate. Offline use does limit fresh provider retrieval, and the result should say that provider relationships may be stale or unavailable.
How do I read the data detail panel?
The recommendation screen shows the supported metadata groups considered for that decision in three categories:
- Trusted fields were present and eligible for use where a rule needed them.
- Conflicted and excluded fields had contradictory evidence and were not used for scoring.
- Not available fields were absent from the candidate evidence.
Conflicted and missing do not mean the same thing. A conflict says evidence was present but disagreed. Missing says the app had no supported value for that group. Neither category is silently filled with a guess.
The panel can also show attached provider provenance and metadata or cover-rights status. ComicPilot's data-source and safeguards page explains which sources are active, local, permission-gated, supplemental, rights-reviewed, or excluded. A source appearing on that page is not by itself proof that it is enabled, reachable, complete, or allowed to provide every field.
How can I improve Limited confidence?
Use the smallest correction that resolves the decision-changing uncertainty:
- Confirm the series title and issue number against the physical comic.
- Add a reliable publisher, series year, volume, or publication date when the title has multiple runs.
- Check the indicia, printing statement, cover designation, and full barcode when edition identity matters.
- Choose a catalog candidate only when its displayed run and edition details agree with the comic.
- Use Wrong Comic? immediately when the candidate is not the comic in your hand.
- Correct an incomplete saved collection record rather than forcing a new candidate to match it.
- Reconnect when a fresh catalog search or provider relationship is necessary, while remembering that another lookup may still have incomplete coverage.
For unmatched collection rows, preserve the valid local record rather than inventing metadata. The collection-import guide explains how separate Series, Issue, Series Year, Volume, Variant, Cover, Printing, and Barcode fields make later comparisons safer.
Do not add a guessed year, cover, or printing just to make the confidence badge change. More fields help only when the values are reliable.
When should Limited confidence change what I do?
Ask one question: could the uncertain field reverse the ownership or price-aware decision?
Pause when the unresolved detail determines whether the comic is a duplicate, whether it belongs to the correct series run, or whether the relationship you care about actually applies. Confirm the copy before buying when the purchase is expensive or difficult to reverse.
You may still have enough evidence for a modest decision when the missing fields are not central. A manually entered issue can match a reliable local issue record under the collector's variant policy. A wishlist entry can remain explicit intent. A low asking price can remain inside a personal budget target. The explanation should show exactly which of those signals fired and omit unsupported ones.
Does Limited mean the recommendation is wrong?
No. It means the recommendation is deliberately narrower than one backed by stronger evidence. Read the reasons and warnings instead of treating the label as a verdict by itself.
Can Should Buy appear with Limited confidence?
Yes. The decision and confidence answer different questions. If trusted collection and price signals support a purchase while other metadata is missing, the app can show a positive decision with Limited confidence. Confirm any missing fact that would matter to you before acting.
Can an offline duplicate warning still be High confidence?
Yes, when the comic and the locally stored exact edition were reliably confirmed and there is no trust-critical identity or variant conflict. Missing network relationships do not weaken the exact ownership fact they were never needed to prove.
Does a low asking price improve Data Confidence?
No. Price belongs to the buy-decision layer. It does not establish the issue, series run, edition, relationships, grade, condition, or market value, so lowering the asking price cannot strengthen the evidence label.
Does missing story data mean the comic has no story connection?
No. It means ComicPilot cannot support that connection with the evidence available for this recommendation. The app should omit the relationship from scoring rather than infer it from a shared character, nearby date, or similar cover.
Limited confidence is useful when it names a boundary you can check. The goal is not to make every badge High. The goal is to know which parts of the recommendation are reliable before you buy the comic.