How the comparison was made

The comparison measures planning capability: what each product's engine can be told to take into account, and what its plan reports. It does not measure ease of use, speed, plan quality, integrations, price or support.

The evidence is what a prospective buyer can read before talking to a vendor: websites, help centres, manuals, API references, release notes and blog posts describing the product. Nothing behind a login or a trial, no demos, no sales material and no video - for every product, ours included. Every verdict quotes the vendor's own words with a link. Third-party reviews and competitor-written comparisons were used only to find products, never as evidence.

1. Choosing the products. A market scan ranked candidates by visibility, overlap with general-purpose 3D load planning, and how much public documentation they have. Eight were audited in depth; the reasons for every inclusion and exclusion are in the scoping report.
2. An open audit of each competitor. One auditor per product recorded everything its documentation shows the engine can do, with no checklist, so nothing was framed around Cargo-Planner. The auditors never saw Cargo-Planner's website. (Brief)
3. Writing the requirements. A separate author, who saw only the competitor audits, merged about 660 recorded capabilities into requirement rows, each worded as something a shipper would ask for, with a pass condition two auditors would agree on. The academic constraint literature (Bortfeldt and Wäscher, 2013; Silva and colleagues) filled in what the market did not document. (Brief)
4. Scoring every competitor on every row, including rows its first auditor was never looking for. A row with no evidence was not a fail until someone had searched for it, and every "not documented" lists where. (Brief)
5. Independent verification. Fresh verifiers tried to prove the scores wrong on a sample drawn by a fixed-seed script. (Brief)
6. Fixing the rules. The uncertain verdicts fell into recurring types. Each type got one rule, applied to every product, decided before Cargo-Planner was scored. (Rules)
7. Scoring Cargo-Planner from its public website by an auditor who had not worked on the competitors, to the same briefs and rules, followed by the same verification and extra scrutiny of its uncertain passes. (Brief)

Most verdicts are clear-cut. For the rest, these rules decide whether a documented capability counts. They were fixed after the competitors were scored and before Cargo-Planner was.

In the v1.0 audit, for each competitor, verifiers re-checked ten randomly drawn documented verdicts, every combination of settings, every marketing-only verdict, and eight randomly drawn "not documented" verdicts, which they searched for again independently. Of 223 competitor verdicts checked, 212 stood unchanged. Of the 64 "not documented" verdicts searched again, 63 held. Errors leaned towards too much credit, mostly combinations an auditor had assembled from fields whose behaviour was never described.

Cargo-Planner got the same sample in v1.0, 30 of 31 standing, plus extra scrutiny of every pass flagged as uncertain, which removed one. Every correction is shown on the row it changed.

• The coordinator knows Cargo-Planner. The person who ran the audit proposed 6 rows after the blind author's draft, each tied to a competitor's documentation or the literature, and reviewed before scoring. Cargo-Planner passes 4 of the 6, and is the only product to pass one of them. Without these rows Cargo-Planner would document 105 of 143, and the order of the products would be Cargo-Planner 105, CubeMaster 90, Cube-IQ 86, packVol 86, MaxLoad Pro 84, Goodloading 53, EasyCargo 36, LoadOptimizer.ai 29, SeaRates Load Calculator 26. They are marked † wherever they appear.
• Two scoring rules may favour how Cargo-Planner is built. Counting combinations of general settings, and showing a plain documented or not in the matrix rather than how each product gets there, both suit a product designed around general rules. Both were flagged as such when they were decided, and both were applied to every product before Cargo-Planner was scored.
• Documentation is not software. A product that does more than it documents scores lower than it would on a hands-on test. That cuts every way, including against us.
• The market scan was in English, on a US search index, so regional products may be missing. Products whose instruction is mainly video, or behind a login, are under-read; this affects Cube-IQ most, where eight help articles require a login.
• We corrected our own documentation before publishing. Where the audit found Cargo-Planner's documentation contradicting itself or its marketing, we fixed the page before this version went live, and only where the fix could not raise our score. Each affected row says so. Additions that could raise it wait for the next version, re-scored by an auditor.
• SeaRates' release notes before 2024 were not searched, because the site could not be reached when the re-check was run.