Sources & verification notes

#How to read this module

This is the same discipline this platform applies everywhere it makes a claim it can't fully stand behind: name the source, tier it honestly, and — where a specific widely-repeated figure could not be traced to anything resembling a primary disclosure — say so explicitly rather than let it stand unchallenged. This is the same treatment Income Playbooks gives a fabricated insurance-industry failure-rate stat and AI Agency gives the "95% of AI agencies dead by 2026" line: named, checked, and flagged rather than quietly repeated.

#Established — confirmed directly against a primary source

• mckinsey.com/careers/interviewing — McKinsey's own description of the Problem-Solving Interview's evaluation dimensions (analytical problem solving, conceptual thinking and creativity, quantitative skills).
• bain.com/careers/hiring-process/interviewing — Bain's own description of its case interview ("think in a logical and structured way... precision, but also creativity, and there isn't necessarily a 'right' answer") and its Experience Interview ("we'll discuss your past experience — personal, academic, and otherwise").
• careers.bcg.com/global/en/case-interview-preparation — BCG's own case-interview description, evaluated qualities, do's/don'ts guidance, and its free official resources (the OneDay@BCG simulation and an associate-authored "what BCG looks for" blog post).
• McKinsey's statement to Business Insider (2024) that it planned to hire roughly 1% of applicants, consistent with 2023 — a specific figure attributed directly to the firm via a named outlet, distinct from the many unattributed acceptance-rate figures circulating elsewhere in this space.
• The Imbellus → Roblox → McKinsey Solve lineage — Imbellus was founded in 2015 to build assessments measuring how people think rather than what they've memorized; Roblox acquired Imbellus's IP in November 2020; McKinsey rebranded the assessment as Solve around 2022. This corporate history is corroborated consistently across multiple independent sources.

#Directional — a consistent pattern across multiple independent secondary sources, not confirmed against one primary disclosure

• The five-stage funnel shape (resume/GPA screen → online assessment → first round → final round → offer) and the general behavior of each stage, synthesized from hackingthecaseinterview.com, roadtooffer.com, casestar.io, myconsultingcoach.com, preplounge.com, igotanoffer.com, and casebasix.com.
• Current online-assessment formats: McKinsey Solve's reported module names ("Redrock," "Sea Wolf," "Sustainable Futures Lab") and invitation-length patterns; Bain's SOVA test delivered via TestGorilla as of 2026, its ~75-minute length and section coverage; BCG's "Casey" chatbot format (25–35 minutes, 6–10 questions, a video response) — none of these three current-format descriptions are confirmed against a firm-published primary source; BCG has explicitly not published an official description of Casey, per the aggregator sources describing it.
• The GPA-screening pattern (fast, heuristic, no published firm minimum, a practically observed ~3.5–3.6 floor at target schools, offsettable by other signals) — consistent across mconsultingprep.com, hackingthecaseinterview.com, casebasix.com, and strategycase.com, none of which cite a firm-published policy.
• The non-target/referral pattern (non-target interviews happening almost exclusively via warm connections) and the general H-1B/OPT visa-bridge mechanics for international hires, including the Bain-specific "files once, doesn't re-file" characterization — the latter specifically unconfirmed against a Bain primary source in this pass.
• The case-type taxonomy (profitability, market sizing, market entry, M&A/growth as the four dominant archetypes) and the "generic/memorized framework is the largest driver of rejections" pattern, corroborated across multiple independent case-prep platforms and instructor writeups.
• The McKinsey PEI's "Drive" and "Leadership" dimension names; Bain's Experience Interview characterized as focused on "Results Orientation" and "Teamwork & Collaboration."
• Mental-math technique guidance (percentage-anchor decomposition, rounding, the Rule of 72) and the recommendation to drill timed and narrated out loud rather than silently.
• The paid-resource comparison (Case in Point/Victor Cheng's LOMS, CaseCoach, PrepLounge, RocketBlocks) — pricing and feature claims sourced to third-party comparison sites (roadtooffer.com, managementconsulted.com, cuttothecase.com, prepmatter.com), not verified directly against each vendor's current pricing page in this research pass.
• The MBB Offer Machine's own self-description of its service scope (application strategy, networking guidance, pre-test/case/fit coaching) and its Trustpilot standing (4.8/5, 150 reviews, ~94% five-star) — the review platform and rating are genuinely third-party, though the review content itself reflects self-selected, opt-in reviewers.

#Speculative — single-source claims, no stated methodology, or claims this research flags rather than repeats

• Any specific stage-by-stage pass-rate percentage — an overall MBB acceptance rate of roughly 0.5–1.5%, a first-round-to-final-round advance rate of 10–30%, a final-round-to-offer rate of 20–30%, and the claim that ~90% of US MBB hires come from target schools. The most detailed version of this table traced in this research (casestar.io/data/pass-rates) discloses its own methodology as a mix of unnamed "firm disclosures," unnamed "recruiter interviews," and a self-run "alumni survey" (stated n=847) — and the same page explicitly states that "firms don't publish stage-by-stage data" and that its figures "are estimates." No firm-published funnel-conversion data was found in this research. Treat every specific percentage in this space as a plausible planning input at best, never as a verified figure.
• McKinsey's total annual application volume — cited anywhere from roughly 200,000 to over 1,000,000 depending on the source, with no reconciled figure found. The hire count (~2,000/year) and the ~1% figure are better attested (see Established, above); the applicant denominator underneath it is not.
• The claim that roughly 50% of McKinsey cases cannot be solved with a standard framework — traced to a single source with no stated methodology. The broader, better-corroborated point it's illustrating (some cases require a genuinely custom structure) is not in question; the specific "50%" figure is.
• The claim that ~35% of case-math questions specifically involve percentage calculation, and the claim that ~70% of candidates who fail MBB math do so because they trained without a timer — both single-source, no stated methodology, no sample size disclosed.
• A "2–4x" referral-lift multiplier on screening odds — appears in some secondary content with no traceable methodology behind the specific number; the qualitative pattern (a real referral meaningfully helps) is well corroborated, the multiplier is not.
• The MBB Offer Machine's "350+" to "400+" placed-clients claim — self-reported by the vendor on its own marketing pages, with no independent audit, no disclosed denominator (total paying clients, so no real success rate is computable from it), and no third-party verification of the founders' stated MBB employment history in this research pass. This is named specifically so a reader encountering it in the program's own marketing recognizes it as an unverified vendor claim rather than an audited outcome statistic — the same treatment this course gives every other unverifiable success-rate figure it names.

#What this means practically

Where this course states a specific number as fact, it's in the Established list above or explicitly caveated inline as Directional. Every figure in the Speculative list appears in the wider MBB-prep content ecosystem framed as settled fact; this research could not confirm any of them against a primary source, and this module exists so a reader who encounters one of these numbers elsewhere — in a prep platform's marketing, a coach's pitch, a forum post — recognizes it rather than treating it as more solid than it is.