Every claim traced back to a primary source, with a reliability and confidence rating.
Documents audited:
livrables/verdal-wp1-market.pptx (17 slides) and
livrables/verdal-wp1-market-model.xlsx (7 tabs, 30 inputs)
Date: 27 July 2026 · Engagement:
VG-STRAT-2026-07 · Auditor: source-verification pass,
two independent web-research streams Method: every
input on the Assumptions tab plus every deck-only factual claim traced
to its cited source. Each source reached at primary where possible.
Verdicts: Verified (identified, traceable, figure
matches) · Probable (consistent with several sources,
not directly confirmed) · Hypothesis (plausible
reasoning without a source, explicitly flagged) ·
Contradicted (source found and it says something else)
· Unverifiable (primary could not be reached).
Source tiers: T1 official statistics body / regulator /
listed-company filing · T2 IEA, industry association, major consultancy
published report · T3 trade press, vendor publication, paid
market-research abstract · T4 unattributable or secondary citation
only.
No deliverable was modified. This report recommends; it does not correct.
| Count | |
|---|---|
| Factual claims traced | 47 (30 model inputs + 17 deck-only claims) |
| Verified at primary | 14 |
| Probable | 9 |
| Hypothesis, correctly self-labelled | 13 |
| Contradicted by the source | 6 |
| Unverifiable — primary not reached | 5 |
| Confidence labels that need downgrading | 5 |
| Confidence labels that need upgrading | 1 (Savills — now settled at source) |
Headline. The self-labelling discipline in the model is strong: every one of the thirteen Hypothesis inputs is correctly flagged, and several disclose their own weakness in capitals. The problem is not the flagged unknowns — it is the five inputs labelled Verified or Probable that the primary source does not support as stated, because those are the ones a Board will assume were checked.
Three findings change the WP1 answer or the way it must be said:
| # | Claim as stated | Verdict | Tier | What the source actually says |
|---|---|---|---|---|
| 1 | McKinsey, European mobility finance: A €25 billion growth opportunity, Oct 2023 — matches figure, currency and year, measures mobility-finance revenue, five countries | Verified at primary | T2 | Published 11 Oct 2023. Verbatim: "those trends will create about €25 billion of new annual revenue for mobility-finance providers by 2030 in the countries we analyzed, which represents a 4 percent compound annual growth rate." Five countries: France, Germany, Italy, Spain, UK — "50 to 60 percent of the total European market". |
| 2 | Société Générale: ~€30bn cumulative public-charging share of an €80bn European total | Verified | T2 (secondary) | Verbatim: "according to various reports, the cumulative investment in Europe needed until 2030 could exceed €80 billion, of which €50 billion is needed for private chargers and €30 billion for public chargers." SocGen is itself citing unnamed third parties. |
| 3 | "The ACEA / McKinsey Charging Infrastructure Masterplan is of the same order" | CONTRADICTED | T2 | ACEA's own release on the Masterplan (Mar 2022): "the new research paper estimates the annual costs for public charging infrastructure at €8bn." Annual, i.e. roughly €70-80bn cumulative to 2030 — more than double SocGen's €30bn. |
| 4 | Grand View Research — European charging infrastructure US$15.1bn in 2030 | Unverifiable today | T3 | Live page now reads US5, 688.3m(2025) → * * US23,404.0m by 2033**, CAGR 19.2%. The 2030 vintage the deck cites is no longer on the source. |
| 5 | Market Data Forecast — US$25.4bn by 2034 | Verified, wrong report named | T3 | US8.76bn(2025) → * * US25.35bn (2034)** is MDF's Europe EV charger (equipment) report. MDF's Europe EV charging station report says US41.54bn → * * US671.31bn by 2034**. Two reports, same house, 26× apart. |
| 6 | Meticulous Research — US$34bn by 2032 | Verified | T3 | "the Europe electric vehicle charging stations market is projected to reach $34 billion by 2032, at a CAGR of 25.9%." Scope includes residential. The source's own HTML page title says "by 2030" — the body says 2032. |
| 7 | Falsification test: "No published source cleanly produces European EV charging market, €25bn, 2030" | Holds — Verified | — | Searched directly for the construction. Nothing produces it cleanly. The claim survives. |
| 8 | IndexBox — European charger hardware €18-24bn annual 2030 (candidate raised in the kick-off pass, not on the WP1 deck) | CONTRADICTED / misread | T4 | Live page: "€8–10 billion in 2026 to €28–35 billion by 2035… Growth will be front-loaded in the 2026–2030 period (CAGR 18–24%)." The "18–24" on the page is a CAGR percentage, not euros. No 2030 value is published. |
The report contains a sentence the deck does not use and should: "EV charging infrastructure, the newest market segment for mobility-finance providers, will remain a relatively niche segment, contributing about 1 percent of total 2030 revenue."
So if the Board paper took its €25bn from this report, then even inside that €25bn, charging accounts for roughly €250m — and it is charging finance, not charging service revenue. The perimeter mismatch is not on one axis but on four at once: what is measured (finance revenue vs charging revenue), geography (five countries including the UK, excluding NL and BE — so not Verdal's six), the industry (financial services), and the flow (new annual revenue vs market size).
One wording correction. Slide 5 describes mobility finance as "loans, leases, insurance, subscriptions". The report's own framing is "loans and leases" to support access to mobility. Match the source's words or drop the extras.
Verdict: Verified. Primary reached and quoted.
Savills Research, Electric Vehicle Charging: Maximising Opportunity, UK Commercial — Autumn 2024, under the heading "1. CPO funded": "Rents between £3,000–£5,000 per bay per annum / Or / 5% revenue share / Or / 20% profit share / 30 year lease / Annual CPI rent review, collar and capped at 1-4%." PDF:
https://pdf.euro.savills.co.uk/uk/commercial---other/savills-ev-charging---aug24.pdf
The open decision recorded in CONTEXT.md (#4 — "read
at source once, and a second attempt could not render the page… treated
as Probable") can now be closed. Upgrade Assumptions!F37
from Probable to Verified, with the full citation.
But three qualifiers change what the figure means, and two of them cut in opposite directions:
Arithmetic defect found while verifying this. Assumptions!F37 states the range as "€13-22m a year". £3,000 × 4,400 bays = £13.2m and £5,000 × 4,400 = £22.0m — those are sterling figures presented as euros. At £1 = €1.17 the range is €15.4m–€25.7m. The multiple on slide 11 is therefore 2.3–3.9× the modelled €1,500, not "2-3x". And on the landlord-funded benchmark (£10,000/bay) the lease would earn €51.5m a year — an order of magnitude above every structure the deck models.
| # | Cell | Claim | Label in model | Verdict | Finding |
|---|---|---|---|---|---|
| 9 | Assumptions!C6 |
European electricity 2,790 TWh (2025), derived from Ember: renewables 1,331 TWh = 47.7% | Probable | Contradicted as labelled | Ember EER 2026 verbatim: "Renewables provided nearly half of EU
electricity in 2025 (47.7%, 1331 TWh)" — T2,
verified. The derivation 1,331/0.477 = 2,790 is arithmetically
right and corroborated by Ember's other pairs. But Ember's methodology
states annual data is gross generation, and demand =
generation + net imports. So 2,790 TWh is EU-27 gross
generation, not demand, and not "Europe".
1. Top-down funnel!A5 labels the row "European
electricity demand, 2030". |
| 10 | Assumptions!C8 |
IEA GEVO 2025 STEPS: EVs 1.0% (2024) → 4.3% (2030) of European electricity | Verified | Verified, with a perimeter warning | IEA publishes exactly this — T2. But the metric is "share of electricity consumption from electric vehicles relative to final electricity consumption by region", and IEA's "Europe" is the WEO regional aggregate: EU-27 plus UK, Norway, Switzerland, Iceland, Türkiye and the non-EU Balkans. Materially wider than EU-27. |
The compound problem. The funnel multiplies IEA-Europe, final consumption (4.3%) by EU-27, gross generation (2,790 TWh). EU-27 final consumption sits roughly 10–13% below gross generation, while IEA-Europe is roughly 20%+ larger than EU-27. The two errors run in opposite directions and do not cancel predictably. The resulting 129 TWh is not defensible exactly as constructed.
This is the calculation that tests the Board's €25bn. It is also the one a Board member with an energy background will catch. This is the most important technical finding in the L1 pass.
Fix, in one of two ways, both cheap: either take an
IEA-Europe final-consumption denominator so both terms
share a perimeter, or take an EU-27-specific EV
electricity share. Then re-label 1. Top-down funnel!A5 as
generation or consumption, whichever is used. The
answer will move by a low double-digit percentage — which is well inside
the funnel's own declared status as a bounding exercise, so no argument
breaks.
| # | Cell | Claim | Label | Verdict | Finding |
|---|---|---|---|---|---|
| 11 | C9 |
30% public share of EV kWh | Probable | Probable — correctly labelled | The model says plainly "No authoritative European kWh split is published." That is accurate and honest. But slide 8's bullet reads "The IEA sources the first two steps: 4.3% of power, 30% public" — the deck credits IEA with a number the model itself says nobody publishes. Deck-level defect, on the page whose title boasts about flagging judgement. |
| 12 | C10 |
22% destination share of public kWh, adjusted down from "36% of European public charge POINTS are semi-public / destination (EAFO-derived)" | Hypothesis | Hypothesis correct; the underlying anchor is T3/T4 | No EAFO-published 64/36 split could be located. The traceable source is Statzon, an aggregator citing EAFO: "Fully public… 63%… semi-public… 37%", on end-2023 EU-27 data. Attributing it to EAFO directly makes it effectively T4. Cite "Statzon, from EAFO data, end-2023" with the date, or drop the anchor and keep the judgement. |
| 13 | — | "No European source publishes a kWh split by location type" | — | Verified — the claim holds | EAFO publishes recharging points by power level, AFIR category and country; no kWh-by-location-type series exists. IEA publishes charging demand by vehicle mode, not location type. The deliverable's statement of its own gap is honest and defensible. |
| 14 | C11 |
30% grocery share of destination kWh | Hypothesis | Hypothesis — correctly labelled, no source claimed | Exemplary. F11 says outright "No European source
publishes a grocery split of destination charging." |
| 15 | C12 |
Six countries = 57% of the European EV parc | Probable | Probable, with one contradicted input | KBA verified (T1): "2,034,260 BEV are registered in Germany" at 1 Jan 2026. ANFAC contradicted (T1): ANFAC/Ideauto Parque de Vehículos 2025 (26 Feb 2026, DGT data) publishes 746,510 electrified (PHEV + BEV), not 0.69m — the model understates by ~7.6%, likely a stale vintage. And the bases are mixed: KBA's 2.03m is BEV only; ANFAC's 746,510 is BEV+PHEV. Mixing a BEV numerator into a BEV+PHEV denominator moves the 57% by several points. |
Additional definitional flag on C12, not a
sourcing issue: the 57% is a parc share and it
is applied to kWh. Countries do not consume charging
energy in proportion to their vehicle counts — charging behaviour, price
and public-network density all differ. This should be stated as an
assumption on the row, not carried silently.
Recommended fix for C12: take all six
countries from one consistent basis — EAFO's BEV & PHEV Vehicle
Fleet Size and Fleet Share file (T1, single reference date,
EU27+UK+NO+IS+CH+TR) — and use the national registries as cross-checks
rather than as the source.
| # | Cell | Claim | Label | Verdict | Finding |
|---|---|---|---|---|---|
| 16 | C21 |
DC retail €0.60/kWh blended; Eleport Feb-2026: DE 0.59, FR 0.52, NL 0.69, BE 0.69, ES ~0.50, IT 0.71 | Verified | Downgrade to Probable | Five of six verified at source. Spain is €0.47, not "~€0.50." Two further defects: these are medians of CPO list prices per country, not volume-weighted realised prices — not comparable to Fastned's realised €0.53/kWh; and "EVwire" could not be located as a co-source — produce the reference or drop it. |
| 17 | — | Deck slides 2 and 11: "a six-country median of €0.60" | — | Mislabelled | €0.60 was derived in the model as a rounded simple mean of
six country medians (F21: "Simple mean 0.62;
0.60 used"). Eleport's own published European median across country
medians is €0.54. Calling €0.60 "the median" is a
labelling error on a slide, and it flatters the Lidl comparison. Say
"a six-country simple mean of €0.60", or use Eleport's €0.54
median and note the comparison narrows. |
| 18 | C22 |
AC retail €0.45; ADAC Feb-2026, 27 operators, 8,500 test charges, DE public AC €0.39–0.59 | Probable | Unverifiable — downgrade to Hypothesis | The ADAC page reachable at primary is dated 08.12.2025, is a tariff market overview, and contains no "27 operators", no "8,500 test charges" and no €0.39–0.59 AC range. The February-2026 study description appears only in German consumer blogs (T4). The study may exist; it was not reached. Do not cite ADAC for this range until the primary is in hand. The Eleport German CPO table is a sourced, itemised alternative. |
| 19 | C23 |
Grocery-operator DC €0.42 observed; Lidl/Kaufland/ALDI €0.44 DE, Lidl €0.39 FR | Verified | Verified with a nuance that matters | Eleport verbatim: "ALDI, Lidl, and Kaufland all price at €0.44, collectively operating ~3,300 DC connectors." But for Lidl and Kaufland €0.44 is the app price; ad-hoc is €0.55. Only ALDI prices €0.44 ad-hoc. France: Lidl €0.39 both ad-hoc and app ✓. The €0.42 blend itself appears in no source — it is the consultant's blend and should say so. |
| 20 | C24 |
Electricity €0.19/kWh; Eurostat non-household band IC incl. taxes H2-2025: EU €0.1837, DE €0.2264, IT €0.2336 | Verified | Three defects — downgrade to Probable | (a) EU €0.1837 ✅ and DE €0.2264 ✅ verified at Eurostat (T1). (b) Italy €0.2336 is contradicted — at that level Italy would be third-highest in the EU, displacing Germany from Eurostat's own published top three (IE €25.52, CY €24.29, DE €22.64 per 100 kWh); secondary sources put Italy near €0.227. (c) "including taxes" is wrong — Eurostat's headline non-household series excludes VAT and other recoverable taxes, which is exactly what €0.1837 and €0.2264 are. Either the label is wrong or Italy was taken on a different tax basis from the other two, which would be worse. (d) The model uses €0.19, not €0.1837, with no stated reason — a 3.4% unexplained uplift on a €16.0m line. |
| 21 | C25 |
Grid capacity adder €0.04/kWh — "NOT PUBLISHED FOR ANY OF THE SIX COUNTRIES" | Hypothesis | Hypothesis correct; the surrounding claim is too strong | ACER publishes a biennial report on electricity transmission and distribution tariff methodologies covering every EU Member State, expressly including connection charges and capacity/power-based tariffs. The defensible formulation is narrower and still supports the argument: "no regulator publishes a comparable, site-level medium-voltage connection-cost schedule across the six countries; ACER's biennial report compares tariff design, not deliverable per-site cost." |
| 22 | C26 |
Payment/software 5% of gross | Hypothesis | Hypothesis — correct | "No EV-charging-specific disclosure found" — accurate. |
| 23 | C28 |
DC opex €19,000/charger/yr — Interpath May 2026, €18–20k | Probable | Verified as an assumption, correctly characterised | Interpath, European EV Charging, May 2026 exists. Footnote: "Assumes c. €180,000 capex per charger, c. €0.25/kWh gross margin and c. €18,000–20,000 annual cost base". The model already says "quoted as an advisory modelling assumption rather than an operator disclosure" — this is exactly right and should be the template for the rest of the tab. |
| 24 | C30 |
DC capex €200,000/charger — Interpath range €150–250k | Probable | Verified range; the point chosen is above Interpath's own | Interpath body text: "High-power charging infrastructure requires significant upfront investment of €150k–€250k per charger." But its own modelling footnote assumes c. €180,000. The model takes €200,000. The €20,000 difference is worth €1.8m a year of amortisation across 1,100 chargers — i.e. the operate loss is ~8% deeper than Interpath's own central case. Either adopt €180,000 or state why €200,000 was chosen. |
| 25 | C31 |
AC capex €6,000/point — anchored on "Spain MOVES III: up to €15,000/port for DC fast" | Hypothesis | The anchor is Unverifiable | MOVES III (RD 266/2021, as amended by RD 821/2023) grants a percentage of eligible cost — 20–35% base for publicly accessible ≥50 kW, with an €800,000 per-file cap. No €15,000-per-port ceiling appears in the national text or the IDAE guidance reached. Per-point caps exist in some autonomous-community calls; if that is the origin, name the region and the convocatoria. Otherwise the anchor should be removed and the €6,000 left as an unanchored Hypothesis. |
| 26 | C32 |
Grid connection €60,000/site — "published range €10,000–500,000" | Hypothesis | The range is Unverifiable; the point may be low | The €10k–500k range traces only to an unattributed blog (T4). The best citable published figure found is Savills: £250,000–£1,000,000 to obtain 800 kVA–2 MVA for an 8–24 bay hub, "site-dependent". At four bays a site the requirement is smaller, but €60,000 sits far below the only citable published figure. Directional risk: if grid connection were €150,000 a site, the operate result worsens by €6.6m to about −€29.0m. |
| 27 | C33 |
DC asset life 12 years — Interpath | Probable | Verified | Interpath footnote: "…12-year life and 10% target IRR." |
| 28 | C34, C35 |
AC life 10 yrs; grid connection life 15 yrs | Hypothesis | Hypothesis — correct | "Modelled." Honest. |
| 29 | C37 |
Lease fee €1,500/bay/yr | HYPOTHESIS | Correct — and the benchmark beside it is now Verified | See Part 2. Upgrade F37's Savills line to Verified with
the full citation; correct the FX; add the landlord-funded £10,000/bay
benchmark; add the three site qualifiers. |
| 30 | C38 |
Revenue share 15%; "industry description puts host revenue shares at 10–25% of gross" | Hypothesis | Hypothesis correct; the 10–25% band is T4 | Checked Powerdot, Electra, Atlante and Zunder. Partnerships are announced; terms never are. Powerdot's retail partnerships (E.Leclerc, Cora, Intermarché, Carrefour, and the IMMO Mousquetaires mandate for Intermarché Belgium) are public; no percentage is. The 10–25% band traces only to vendor glossaries that themselves say percentages "vary considerably based on individual negotiations." The model's own words — "No named operator discloses its actual percentage" — are accurate. Verdal's own two agreements are the best available benchmark, and they are a data-room request, not an estimate. |
| 31 | C40 |
Board paper: €25,000m, 2030 | Client input — under test | Correct treatment | Flagged as an input, never divided into an annual margin anywhere in
the model (tab 4!A13). Methodologically clean. |
| # | Claim | Verdict | Finding |
|---|---|---|---|
| 32 | Fastned FY2025: 183.0 GWh across 406 stations; gross profit €0.53/kWh; operational EBITDA €43.7m (36%); net result −€30.3m | Verified, with a period mismatch | All FY2025 figures verified in the Annual Report 2025 (T1). But the station metrics quoted as FY2025 — 6.5 chargers, €958k invested, ROIC 17.3%, time-based utilisation 14.5% — are the Q1-2026 column of the April 2026 investor presentation. The Q1-2025 column reads 6.1 chargers / €835k / ROIC 11% / utilisation 13.9%. Presenting Q1-2026 station economics as FY2025 overstates them. Also: the audited annual report says the net loss was €30.2m; the presentation table says 30.3. Cite €30.2m. |
| 33 | "the 30% Fastned itself cites for a >40% ROIC" | Verified verbatim | "ROIC at 30% utilisation, current charge speed >40%" ✓ |
| 34 | Deutschlandnetz ≈ €250k per point | Probable, derived | No published per-point cost exists. Federal investment stated at €1.9bn (BMDV) across ~8,784 + ~1,000 points → €194k–216k/point; the €2.3bn ÷ 9,000 = €255k version is trade press. Either way it is a subsidy envelope covering construction and operation, not a capex-per-point. Present as a derived ratio with the numerator named, or drop. |
| 35 | Pod Point / Tesco FY2024: £7.672m revenue, £1.784m operating profit, 598 sites, 2.24 chargepoints/site, part advertising; group adjusted EBITDA −£20.7m; going-concern material uncertainty; sold to EDF | Verified in full at primary (T1) | Every figure ties to the FY2024 preliminary results RNS. The Owned Assets segment is defined in the accounting policy as including "a fee for provision of media screens on the chargepoints for advertising purposes" ✓. EDF takeover £10.6m at 6.5p/share, completed Aug 2025, against the £352m IPO valuation (press-reported, T3). |
Both were surfaced by this verification pass. Neither is in the model or the deck.
Fastned's own supermarket business case. The
April 2026 investor presentation carries Fastned's published 2030
comparison of a motorway service area against a
supermarket site: 1,000 daily passing traffic, 20%
capture, 40 sessions/station/day, 30-minute dwell
(sourced by Fastned to Eurostat shopping-time data), 69 kW max session
speed, and maximum annual revenue per station of €252k against
€1,417k for a motorway site. This is a listed competitor's
published, sourced view that a grocery car park is worth roughly
one fifth of a motorway site. It is a direct, T1 test
of C18 (80 kWh/bay/day) and it belongs in WP1's
sensitivity, not only in WP2.
Pod Point's Tesco estate as a per-site reality check. £7.672m and £1.784m across 598 sites = £12,829 revenue and £2,983 operating profit per site per year, or £5,738 revenue and £1,334 operating profit per chargepoint. The whole operating profit per chargepoint is below the bottom of the Savills £3,000–5,000 per-bay rent. On a mature UK grocery estate, being the landlord out-earned being the operator.
This second point is the strongest sourced answer the
engagement has to the Board's own question — "is value
being left on the table?" — and it is a primary filing, not a model
output. It currently sits in
notes/kickoff-grounding-research.md and appears nowhere in
WP1.
| Cell / claim | Current label | Recommended | Reason |
|---|---|---|---|
Assumptions!F37 — Savills benchmark |
Probable, "CONTESTED" | Verified | Read at primary; full citation available. Open decision #4 in
CONTEXT.md can be closed. |
Assumptions!E21 — DC retail price €0.60 |
Verified | Probable | Spain misquoted; list-price medians, not realised prices; co-source "EVwire" not locatable. |
Assumptions!E22 — AC retail price €0.45 |
Probable | Hypothesis | ADAC primary not reached; the cited study parameters appear only in T4 blogs. |
Assumptions!E24 — electricity €0.19 |
Verified | Probable | Italy contradicted; tax basis mislabelled; €0.19 ≠ the cited €0.1837 without a stated reason. |
Assumptions!E6 / 1. Top-down funnel!A5 —
2,790 TWh |
Probable | Probable, re-labelled | The figure is right; the label (demand vs generation, Europe vs EU-27) is wrong. |
Assumptions!E8 — IEA 4.3% |
Verified | Verified, perimeter-flagged | The figure is verified; the perimeter it is multiplied into is not the same one. |
Assumptions!E12 — 57% parc share |
Probable | Probable, basis-flagged | ANFAC input contradicted; BEV vs BEV+PHEV bases mixed; parc share applied to kWh. |
Thirteen Hypothesis labels were checked and all thirteen are
correct. C7, C10, C11,
C15, C16, C17, C18,
C19, C25, C26, C29,
C31, C32, C34, C35,
C38. Several disclose their own weakness in capitals. This
is the part of the package that most directly answers the client's
stated scoring criterion, and it should not be softened when the
corrections above are made.
| Item | Why | What would settle it |
|---|---|---|
Eurostat nrg_pc_205 raw cube |
Dissemination API returned empty | Pull Italy H2-2025 from the databrowser directly, and confirm the tax basis (X_TAX vs I_TAX) |
| ADAC February-2026 study | presse.adac.de not reachable; only T4 blog
corroboration |
Obtain the ADAC press release or replace the AC price with the Eleport German table |
| IEA WEO Annex C regional definition | Annex not opened | Confirm the exact "Europe" country list; that it exceeds EU-27 is certain |
| "EVwire" as co-source for the price report | No such source located | Produce the reference or remove it from F21 |
| MOVES III per-port ceiling | Not in the national text or IDAE guidance reached | Name the autonomous-community convocatoria, or remove the anchor
from F31 |
| Any host revenue-share percentage | No CPO discloses one | Interview programme item; and Verdal's own two agreements (DR-02) |
| Grand View 2030 vintage | Source rebased to 2033 | Cite with retrieval date, or drop |
The model's honesty layer is the best part of the package and
it is doing exactly what the client said it would score on.
Thirteen Hypothesis labels, all correct, several disclosing their own
omissions by name — F17 admits a €19.6m driver "belongs
in the sensitivity and it is not in it"; F25 says the
largest cost hole is published nowhere; F37 discloses that
the firm's own two research passes disagreed. That discipline should
survive every correction below.
Six claims are contradicted by their sources and six confidence labels need moving. The ones that matter, in order:
None of these breaks the WP1 argument. The perimeter finding stands, the falsification test on €25bn holds at source, and the McKinsey candidate is stronger than the deck claims. What changes is the confidence with which several individual numbers can be said out loud.