07 — Rollup to SROI Row 1
Audience: Impact Lead compiling the Q1 board report (Sprint 3 §03). Use: Defines exactly how dashboard data becomes the headline SROI ratio.
The Row 1 sentence
"For every £1 invested, this cohort returned £X.XX in value — driven by [top 2 drivers]."
The number is calculated. The drivers are evidenced. Both are auditable.
Inputs (all from §05 data model)
| Input | Source | Used for |
|---|---|---|
| Engagement fee | Client contract (CRM) | Denominator |
| Workshop attendance % | v_cohort_engagement | Quality weighting |
| Coaching delivered count | v_cohort_engagement | Confidence in behavioural data |
| Pulse responses (Day-30) | pulse_responses (n≥5) | Behavioural delta evidence |
| Confidence delta | v_cohort_sentiment_deltas | Driver: wellbeing/confidence |
| Self-reported conversations opened | Pulse Q4 (Sprint 1 §06 / Sprint 3 §05) | Driver: manager capability |
| Self-reported workload adjustments made | Pulse Q5 | Driver: line management quality |
| Sector wage data | ONS public dataset (cached annually) | Productivity proxy |
| Avoided turnover (modelled) | UK CIPD turnover cost data × estimated retention lift | Driver: retention value |
| Avoided absence (modelled) | HSE workplace absence cost data × estimated absence delta | Driver: presenteeism value |
Formula (simplified)
total_value = (
avoided_turnover_value +
avoided_absence_value +
productivity_uplift_value
) × confidence_factor
row1_ratio = total_value / engagement_fee
Where confidence_factor is bounded by:
confidence_factor = min(
pulse_response_rate / 0.6, # 60% response rate = full weighting
workshop_attendance / 0.8, # 80% attendance = full weighting
1.0
)
This means a cohort with poor engagement cannot game a high SROI ratio — the confidence factor down-weights everything.
Driver definitions
Top 2 drivers shown in the headline are the two with the largest contribution to total_value. Always cite the mechanism, not just the value. Examples:
- "driven by managers opening earlier wellbeing conversations and a measurable reduction in modelled absence days"
- "driven by avoided modelled turnover in the frontline cohort and confidence gains in financial conversations"
What NOT to claim
- No clinical outcomes. Never claim reduced depression, anxiety, or any DSM/ICD concept. We are not a clinical service (see project memory: no-clinical-services).
- No causal language without confidence factor. Always: "associated with", "coincided with", "self-reported" — never "caused" or "led to" unless a controlled comparison was contracted.
- No individual attribution. Never: "Sarah moved from amber to green." Always cohort-level.
- No cross-client benchmarking in v1. "Top quartile" claims need the v2 benchmark dataset.
Audit trail
Every sroi_rollups row stores calculation_jsonb with:
- All input values used
- All public dataset versions (e.g. "ONS_ASHE_2025_Q3")
- Confidence factor breakdown
- Driver contributions
- Calculation timestamp + script version
If a client challenges the number, Impact Lead can reconstruct it from calculation_jsonb alone — no recalculation needed.
Sign-off chain
| Step | Owner | What they check |
|---|---|---|
| 1. Auto-calculation | Cron job (weekly from W6) | Formula runs without error; writes sroi_rollups |
| 2. Driver review | Impact Lead (W12) | Top 2 drivers make narrative sense vs Row 4 stories |
| 3. Quality gate | Delivery Lead (W12) | Confidence factor ≥0.6; if lower, hold publication |
| 4. Board report compile | Account Mgr (W12-13) | Pulls Row 1 sentence into Sprint 3 §03 template |
| 5. Client review | HR sponsor (W13) | 48h window to query before final delivery |
| 6. Final delivery | Account Mgr (W13) | PDF + dashboard archived snapshot |
When to suppress Row 1 entirely
Do not publish Row 1 if any of:
- Pulse response rate <40% and workshop attendance <70%.
- Cohort size <10 (statistical reliability floor).
- Material change to client context mid-engagement (restructure, acquisition) that confounds the measurement.
In those cases, the board report shows Rows 2-4 only, with an explicit note:
"Row 1 SROI calculation suppressed for this cohort. See Section [X] for context. Engagement evidence is presented in Rows 2-4."
This protects the credibility of every other Row 1 we publish.
End of Sprint 5. This closes the loop: Sprints 1-4 curriculum runs inside the Sprint 5 hybrid playbook, surfaces on the dashboard, and rolls up to the Sprint 3 board report.
Next: Sprint 6 — Sales enablement + pricing calculator. Say "ship 6".