Methodology explainer
Unmeasured is not zero: how UpgradeOrSave coverage and rankings work
Why a product can be ranked in one workload and absent from another, and what indexed, under review and not approved mean.
Editorial dates describe the text. They do not date the measurements. Measurement acquisition dates are unavailable in this view.
Rankings invite a simple reading: higher is better, and anything not on the list must be worse. On UpgradeOrSave that second half is wrong. A product that does not appear in a ranking has no published result for that workload, which says nothing about how fast it is. This explainer covers how coverage is scoped and what each state on our pages means.
Coverage is per workload and per reference domain
Each ranking orders one workload against one frozen reference domain. A processor can be indexed for rendering and absent from compression; a graphics card can have a ready rendering result on one Blender backend and none on another. When we report coverage, the count of ranked products and the population they are drawn from are scoped to the same workload and domain, so a ranking never silently changes what it is counting.
What each state means
- Indexed: the product has a verified, publicly approved result at the current reference for this domain, and it is ranked.
- Not enough evidence: measured evidence exists but does not yet meet the coverage the method requires. No score is published.
- Evidence under review: a verified result exists, but part of the evidence behind it is still in editorial or research review. No score is published until that review finishes.
- Not yet approved for publication: a verified result exists, but publication rights for one or more of its sources are not confirmed. No score is shown.
- Unindexed or not measured: there is no result for this workload at all.
None of these states is a zero. We never place a missing product at the bottom of a list, interpolate a value from neighbours, or borrow a result from a sibling model.
Filters never renumber ranks
When you filter a ranking by name or manufacturer, the rows you see keep their original rank within the full domain. Filtering is a view on the same published rows, not a new ranking, so a product's position never improves because others were hidden.
Freshness is a separate fact
A ranking page loading today does not mean its measurements were taken today. Measurement, publication and retrieval dates belong to the evidence, and where an exact source date is not exposed on a page, freshness is unknown rather than current by default. Our pages show the reference version a result belongs to, so you can tell which frozen scale it was measured against.
What this means for an upgrade
- If the product you are considering is not ranked for your workload, the comparison is unknown, not unfavourable.
- Check the reference domain before comparing two products; scores from different domains are not comparable.
Limitations and uncertainty
- Per-measurement dates are not yet shown on ranking tables.
- Coverage numbers change as evidence is admitted and reviewed; the states above stay the same.
Evidence and methodology
Source-rights checks and active research holds apply before evidence reaches this page. An unavailable result is omitted rather than estimated.
No commission-tracked links appear in this article. Any future affiliate links require separate activation and a clear disclosure near the link. Read our disclosure.