Q: Why does Developer consistently score +17 pts above programme NPS across 3 cohorts — is this audience fit, content relevance, or delivery?

Content-audience fit is the most likely driver: Developer NPS averages +52 across 3 cohorts vs. +38 programme average, but all Developer sub-groups are small (n=5–12) so no single cause can be confirmed statistically.

Developer NPS scores are +67 (C15, n=12), +50 (C16, n=6), and +40 (C18, n=5), averaging roughly +52 — approximately +14 pts above the +38 programme NPS, though the meeting context cites +17 pts. The pattern points most strongly to content-audience fit: the programme is hands-on and technically focused (exercises, agent-building, API depth), which directly maps to what Developers are there to do. Transformation Leads, by contrast, score +69 (C15), -12 (C16), and +19 (C18) — high variance driven by consistent detractor feedback that the programme lacks go-to-market, solution design, and business context. C18's Developer detractor ('Not very technical') and C15's Architect detractor ('expecting more advanced learning') suggest even technically-oriented personas can hit a ceiling, but this lands below Developer experience. Delivery and proficiency-level fit cannot be isolated as independent drivers because cross-cohort correlation data is flagged as insufficient, and no Developer-specific open text provides strong delivery attribution. Key data points: · Developer NPS: +67 (C15, n=12), +50 (C16, n=6), +40 (C18, n=5) — all above programme cohort NPS in each respective cohort · Transformation Lead NPS ranged from -12 to +69 across the same 3 cohorts, with C16 detractors explicitly citing missing go-to-market and business content · C18 Developer detractor scored 6 with verbatim 'Not very technical' — suggesting the gap narrows when content depth is perceived as insufficient even for Developers (n=5, small) Caveats: Developer n is 5–12 per cohort — below the 20-respondent threshold for confident segmented conclusions; findings are directionally consistent but not statistically robust. To confirm whether content relevance, delivery quality, or proficiency-level matching is the primary driver, you would need Developer-specific post-programme open text, facilitator ratings by persona, and AI proficiency breakdowns within the Developer segment.

2026-07-28 18:23 UTC

Q: Why do Learning and exploring AI proficiency participants rate the programme -9 pts below average — what does the D1 calibration data tell us?

The D1 calibration data does not break out confidence scores by AI proficiency segment, so the -9pt gap cannot be directly explained by D1 data alone.

The data brief states cross-cohort AI proficiency correlations are 'insufficient data,' and the 3-dimension confidence scores (D1/D2) are reported only at cohort level, not by proficiency sub-segment. What the D1 data does show is that baseline confidence was nearly identical across all three cohorts: C15=3.84, C16=3.82, C18=3.80 — a spread of just 0.04 points. This means D1 calibration cannot explain differential NPS outcomes for Learning and exploring participants specifically. The 'Learning and exploring' segment had 12 respondents in C15, 9 in C16, and 13 in C18 (totals of 34 across cohorts), which is borderline for sub-segment analysis. One C16 verbatim from a Learning and exploring detractor (NPS 6) flags the gap directly: 'It's nice on a personal level to know these things but from a business perspective I would like to learn more insights and solution design' — suggesting the programme feels insufficiently applied for this group's aspirations, not that they arrived under-confident. Key data points: · D1 confidence nearly flat across cohorts: C15=3.84, C16=3.82, C18=3.80 — no cohort-level signal distinguishing Learning and exploring participants · Learning and exploring segment n: C15=12, C16=9, C18=13 — sub-segments too small for statistically confident NPS breakdowns per cohort · Only 1 verbatim directly from a Learning and exploring detractor (C16, NPS 6) cites frustration with lack of applied business content, not foundational difficulty Caveats: To answer this question properly, D1 confidence scores need to be broken out by AI proficiency segment, not just cohort. Additionally, NPS scores per proficiency band per cohort are not provided in the brief — only the programme-level -9pt gap is cited in the meeting context, without the underlying per-cohort proficiency NPS table needed to validate or explain it.

2026-07-28 18:23 UTC

Q: What explains Deloitte's persistent below-programme NPS score across 2 cohorts, and what would change it?

Deloitte attendees score -36 pts below programme NPS on average, driven by senior practitioners who find content too tactical and insufficiently focused on go-to-market and solution design.

In C16 London, Deloitte's NPS was exactly 0 (n=6, vs programme NPS +17, a gap of -17 pts). In C18 San Francisco, Deloitte's NPS was -20 (n=5, vs programme NPS +35, a gap of -55 pts). The verbatim evidence points to a consistent root cause: Deloitte detractors are experienced practitioners — one a Partner/MD-level with 'Delivering independently' AI proficiency — who expected go-to-market strategy, PoC failure analysis, and solution design depth, but received technical hands-on exercises they judged replicable remotely or already covered internally. The C16 detractor scoring 2 explicitly stated: 'no go to market, no what to do to not have failed PoCs' and 'no differentiator to a half day set of exercises I can do remote.' A C16 passive also flagged that Day 1 content was mandatory learning already completed at their company. To change this, content for Deloitte cohorts would need to shift toward business case framing, failure-mode analysis, and client engagement strategy — or Deloitte attendees need to be streamed into a more advanced track from Day 1. Key data points: · C16 London: Deloitte NPS = 0 (n=6), programme NPS = +17, gap = -17 pts · C18 San Francisco: Deloitte NPS = -20 (n=5), programme NPS = +35, gap = -55 pts · Average Deloitte NPS gap across both cohorts: -36 pts below programme NPS · Deloitte detractors include Partner/MD seniority with 'Delivering independently' AI proficiency — the programme's most advanced segment · Deloitte promoter in C16 cited 'hands on tech dev and understanding' (NPS 9) — suggesting the subset who benefit are less experienced practitioners Caveats: Deloitte n is critically small: n=6 in C16 and n=5 in C18. No confident statistical conclusions can be drawn — these findings are directionally consistent but require larger Deloitte samples to confirm. C15 had no Deloitte org data reported, so the 'two cohort' pattern cannot be extended further. Qualitative verbatims are the strongest signal available here.

2026-07-28 18:23 UTC

Q: What do promoters and passives say about the programme, and what specifically separates them?

Promoters cite learning and hands-on depth as drivers; passives consistently signal a gap between what was delivered and what they specifically needed — either more advanced content or better tool coverage.

Promoters (51.6% of respondents, n=48 across C15-C18) cluster around two themes: discovery of new capabilities ('learned about new possibilities in Claude I had no idea about'; 'caching and sub agents which I was looking for') and the quality of hands-on exercises and tutors ('Did a lot of exercise and learnt a lot'; 'Really nice Team of Tutors'). Passives (34.4%, n=32) express qualified satisfaction — they acknowledge value but flag specific unmet needs: one wanted more time in Claude desktop/Cowork, one noted most Day 1 content was already mandatory training at their company, and one got more value from asking presenters questions than from the structured content itself. The separating factor is specificity of fit: promoters found the programme matched or exceeded their expectations, while passives experienced a relevance or depth mismatch — not a quality failure, but a personalisation gap. Notably, two passive Transformation Leads in C15 and C18 are among the most articulate: one called it 'really packed with technical pieces' (score 8) but stopped short of a 9-10, and one found it 'useful for learning to build' but wanted different tool coverage — suggesting they valued the programme but felt it wasn't quite calibrated to their role. Verbatim volume is limited (roughly 3 passives per cohort surfaced here), so these patterns are directional, not statistically conclusive. Key data points: · Promoter rate: 51.6% (n≈48); passive rate: 34.4% (n≈32) across C15-C18, n=93 total · Passive verbatims across all 3 cohorts share a 'good but not quite right for me' pattern — content depth, tool selection, or prior knowledge overlap flagged in 4 of 7 passive quotes surfaced · No passive verbatim explicitly criticises quality; separating factor is fit/personalisation, not delivery — contrast with detractors who cite missing go-to-market content and insufficient technical depth Caveats: Only 7 passive verbatims are available across 3 cohorts (n≈32 passives total) — this is a very thin sample for confident thematic conclusions; a full verbatim export or follow-up pulse with passives would sharpen conversion targeting significantly.

2026-07-28 18:24 UTC