Evaluations Guide

*Provenance (checked 2026-09-29): this file is a verification-and-rebuild pass on a raw, unverified draft — THE DEFINITIVE GUIDE TO EVALUATIONS — WBS12.md, supplied from ~/Downloads/FINSTO/ — which carries no confirmed provenance of its own and belongs to a material family with a documented history of a low verbatim-quote hit rate. Every specific claim in that draft was checked, not transcribed. Checked against research/veridian/WBS12-verified-facts.md in full (16 series, 15 mark schemes, 13 readable examiner reports, June 2019–October 2024). Independently re-verified against primary PDFs for the specific paper the draft's own worked examples are drawn from — October 2023 WBS12/01 (question paper, mark scheme, examiner report) at ~/Documents/SOL/Old/P2 Materials/Business/October_2023/, text-extracted directly with pdftotext -layout for this pass, page by page, not trusted from the draft's own quotations. Additionally grepped all 14 readable WBS12 examiner reports (Jan/June/Oct 2019–2024) for the draft's specific claimed phrasing, to check claims of the form "confirmed every series." Read content/veridian/technique-hub/wbs12/forge-exam-technique-guide.md in full first, so this file is additive to it rather than a duplicate — that file already covers tariff-by-tariff timing and general technique; this file covers evaluation (AO4) specifically. Verification result: of roughly 22 distinct checkable claims in the source draft, 17 verified cleanly, 3 needed a flag or a correction, and 2 — both claims of the form "confirmed every series in examiner reports" — were checked directly and did not check out anywhere in the archive. See §5 for the full ledger.

#WBS12 — Evaluations: What Genuine Conditional Judgement Actually Looks Like

Evaluation is AO4. It is the mark component that separates a Level 3 answer from a Level 4 answer on every levels-based question this paper sets — Discuss, Assess, and the 20-mark Evaluate. This file is about what genuinely earns that mark, built only from claims that check out against real mark schemes and real examiner reports, using the actual October 2023 WBS12/01 paper (Arditi Tours, Grupo Tamazula, Amazon's Kindle withdrawal from China) as its worked material throughout, since that is the paper the unverified draft was itself built from.

#1. Where AO4 exists on this paper, and where it doesn't

• Define (2), Explain/Calculate/Construct (4), Analyse (6): zero AO4. Confirmed independently in two different series: the October 2022 examiner report states plainly that advantages "were not rewarded as 'analyse' questions do not have any AO4 (evaluation) marks," and the October 2023 report repeats the identical rule on the same paper's own Q1c: "'Analyse' questions do not have any AO4 (evaluation) marks." Writing an evaluative "however" on a 6-marker earns nothing and costs time that belongs to the second reason's own chain.
• Discuss (8): evaluation is implicit, through genuine two-sidedness — no conclusion required. The top band (Level 3, 6–8 marks) asks for balance and "an awareness of competing arguments/factors" — not a stated preference, not a condition. Pearson has restated "a conclusion is not required for an 8 mark discuss question" verbatim or near-verbatim in at least 10 of the 13 readable examiner reports spanning 2019–2024. The October 2023 report's own Q1d tip repeats it for this exact question: "A conclusion is not required for an 8 mark discuss question."
• Assess (10): Level 4 requires "a supported judgement." That is the exact phrase in the Level 4 descriptor (8–10 marks) — a stated, reasoned preference between the competing arguments, not just an awareness that they exist.
• Evaluate, the 20-mark Q3 (worth 25% of the paper): "significance of competing arguments" plus "an effective conclusion." The October 2023 examiner report's tip for Q3 states this directly, in language specific to the evaluate command word: "ideas needed to be developed and presented with understanding of the significance of competing arguments. To achieve the top level... an effective conclusion is sought."

The pattern that actually matters: the word significance is genuinely Pearson's own word — but it belongs to the 20-mark Evaluate question specifically (and to the level below top on Assess and Discuss, describing what a weaker answer is "unlikely to show"). It is not printed inside the Assess Level 4 descriptor itself, which uses "awareness of competing arguments/factors, leading to a supported judgement" instead. That distinction matters practically: an Assess answer needs a stated winner; a 20-mark Evaluate answer needs the winner justified by weight, not just named.

#2. A claim that does not check out: there is no Pearson "trigger phrase" rule

The source draft's central mechanism is a set of required "trigger phrases" — "only if," "provided that," "conditional on" — which it presents twice as verified fact: "Confirmed from examiner reports across every series: 'Trigger phrases... are required for Level 4.' Not suggested. Required." It makes the identical claim of "confirmed every series" for the idea that "it depends" scores exactly zero.

Both claims were checked directly. Every readable WBS12 examiner report from June 2019 to October 2024 (14 files, all cross-checked, including the October 2023 report covering the draft's own three worked examples) was text-extracted and searched for depends, condition, unconditional, genuine, trigger, "only if," "provided that". None of these words or phrases appears in any of them — not once, in any series, on any question. The specific "confirmed every series" claim is unsupported by the archive it claims to come from.

What Pearson's own examiners actually say about how these marks are awarded is different, and better evidenced: best-fit, whole-response marking against the level descriptor's own language, not phrase-matching. The October 2023 report states the mechanism plainly for Q1d: "Examiners read the whole response and decide which level is the best match. If a response is lacking certain characteristics, examiners move towards the bottom of the level... this approach is used for all levels of response questions on the paper." forge-exam-technique-guide.md independently confirms the identical mechanism from the January 2025 report and adds the mark scheme's own header warning that indicative content is "examples, not a checklist." No fixed vocabulary is rewarded or required anywhere in the primary sources — a response can reach Level 4 without ever writing "only if."

A second, smaller inaccuracy in the same part of the draft: it prints a "Level 4 descriptor" reading "...shows an awareness of the SIGNIFICANCE of competing arguments" and presents this as one verbatim quote applying to "all extended questions." The real Level 4 Assess descriptor contains no such wording — it says "shows an awareness of competing arguments/factors, leading to a supported judgement," with no instance of the word "significance." The word "significance" only actually appears, verbatim, in the 20-mark Evaluate's Level 3 and Level 4 descriptors. The draft's quote conflates two different tariffs' wording into one and inserts a word that isn't in the descriptor it claims to be quoting.

#3. What genuine evaluation looked like, on the real paper, per the real examiner report

#Arditi Tours — Q1d (Discuss) and Q1e (Assess)

The real extract gives a break-even point of 17 passengers per journey and a diesel price of €1.58 a litre in April 2022, both rising. The mark scheme credits two different margin-of-safety readings depending on how the data is read: 2 passengers on a per-service basis, or 8 passengers "spread over the whole day" — a genuinely useful nuance the draft's own worked example dropped, keeping only the more comfortable 8-passenger figure.

The examiner report describes what a genuinely strong Discuss answer did — with no trigger phrase in sight: "Stronger responses presented chains of reasoning... calculating a daily margin of 8 passengers, thus enough to be reasonably comfortable... This was countered well with... the rising cost of diesel and/or the fluctuating demand." That is the real shape of Discuss-level evaluation: state a comfortable position, then genuinely counter it with a second, extract-based factor that cuts the other way — no conclusion, no named condition, just real two-sidedness.

For Q1e, the mark scheme's own indicative content makes exactly the move the draft's model answer borrows: "if demand is price elastic which is more likely due to there being several competitors, revenue will decrease with an increase in price." The examiner report confirms this was genuinely rewarded: "Analysis of price elasticity of demand was often effective, especially when considered with evidence such as there being a number of competitors." This is Assess-level evaluation done right — a stated position (price rise unlikely to be the best method), built directly on a named extract fact (several competitors), not a hedge.

#Grupo Tamazula — Q2e (Assess, waste minimisation)

The extract establishes real, checkable facts: hot sauce made with fresh chillies, exports "especially to the US and Canada," a 26,000 m² factory using "the most modern equipment available." The mark scheme's own indicative content weighs a genuine difficulty against a genuine mitigation: perishability and lengthening export transport time on one side; chilled/frozen storage, stock rotation, and computer-managed ordering on the other. The examiner report's Q2e commentary is a caution worth keeping for its own sake, distinct from anything about phrasing: "On occasion, candidates scored zero marks because they showed no understanding of waste, as referenced in the specification." Evaluation cannot rescue an answer that has misunderstood the underlying concept — genuine subject knowledge comes first.

#Amazon's Kindle withdrawal from China — Q3 (Evaluate, 20 marks)

The extract's real figures check out precisely: "China's digital reading market generated more than $6bn... an increase of more than 18% from 2020"; domestic rivals named as "Xiaomi, iFlytek and Huawei"; "more than 500 million Chinese users" reading digitally in 2021; the Kindle's "Harry Potter" gap. All of these appear in the draft's worked example and are accurate.

One place the draft overreaches: it builds a condition around "Chinese government regulation... a genuinely insurmountable external barrier," presented as if extract-anchored. Neither Extract E nor Extract F mentions any government regulation affecting Amazon's China operations at all — "government regulation" appears only in the mark scheme's generic list of external-failure categories ("market conditions, competition, supplier problems and government regulation"), which is spec-level content, not something evidenced about this business specifically. Treating it as a named, business-specific condition is exactly the "unanchored condition" failure mode the draft itself warns against elsewhere.

What the examiner report actually names as the route to the top level is worth keeping in place of the dropped framework: "higher level attainment came from an awareness of competing arguments such as contrasting the likely impact of internal causes with external causes but also, by evaluating the likelihood of the causes analysed." Concretely, for this extract, that means: Amazon's marketing and localisation failures (the Harry Potter gap, poor customer support, no hardware adaptation) were within Amazon's control; the $6bn/18% growth figure shows the external market was not contracting, which weakens competition as the primary cause; the genuine, evidence-supported comparison is which set of causes is more likely, given what the extract actually shows Amazon could have controlled — not a search for a magic phrase, and not an invented regulatory barrier.

#4. Failure patterns that are genuinely confirmed (kept from the draft, re-sourced)

• Listing without linking to the actual question. Confirmed directly on this paper's own Q3: "Rewriting the extracts to state the information provided to candidates in the first place, rather than answering the question, did not enable the candidate to progress through the levels."
• One-sided Discuss. Confirmed on this paper's own Q2d: "Some candidates only look at one side, thus restricting their marks due to not providing an awareness of competing arguments."
• Extract quoted standalone, not used in reasoning. Confirmed on this paper's own Q2c: "a part of the extract was simply stated separately, rather than used in the analysis. This does not allow access to the application marks."
• Guessing at an unfamiliar topic. Confirmed on this paper's own Q2e (above) — genuine subject-knowledge gaps score zero regardless of evaluative technique.

The draft's "it depends is zero marks" and "generic advantages/disadvantages list" patterns are plausible in spirit — a vague hedge cannot show the specific, extract-based awareness every descriptor asks for — but neither is attributable to an examiner report as stated. Treat them as reasonable inference from the level descriptors' own language (which consistently rewards "developed chains of reasoning" and "effective use of context," neither of which a placeholder hedge can supply), not as a separately confirmed Pearson rule.

#5. Verification ledger

Result: 17 of 22 checkable claims verified cleanly, 3 flagged (incomplete, overreaching, or internally contradictory), 2 dropped as fabricated. The two fabricated claims are load-bearing for a large share of the original draft's prose — its "Type 1/2/3" trigger-phrase framework, its formula templates, and its phrase-checklist — none of which is reproduced in this file. What is kept here is what actually checks out: real extract facts, real mark-scheme indicative content, and real examiner-report commentary on the genuine mechanism (best-fit, whole-response marking against descriptor language) that determines these marks.