Validate - Effort Model (Internal)¶
Never client-facing. Two line families sit in this chapter: the population lines - the least-calibrated model in the process, every rate carrying a low-confidence flag and a capture instruction - and the cycle line, fitted alongside the Build model on the same five shipped sites.
Assumptions¶
- Conventions: website-pricing.md § Effort-Model Conventions.
- Expected mix: population lines 15:85 P:M - scripts and the migration report are lead work, manual population and QA are mid work. Cycle line 25:75 P:M - triage and the client-facing moments are senior work; amends implementation is mid.
- Pricing basis: computed estimate at rate card, counts as the contract, as Design and Build. Inputs come fixed from the Cornerstone: migration tiers per content type (2.2 - scriptable vs complex, with volumes), content volumes, the readiness gate for client-supplied content, and the cycle allowance.
- Population covers MAJOR-executed work only: scripted migration and our manual population. Client population (where sold) runs through the shared triage workflow and is not priced on these lines; content creation is entirely the client's.
The Population Lines¶
| Line | Rate | Scales with |
|---|---|---|
| Migration script | ~4–6 hrs per source × content type | Sources and types - not volume; runs are nearly free |
| Script re-run at testing close (the two-pass pattern) | ~1–2 hrs per source | Near-fixed - the cheap insurance against content drift |
| Manual population (by us) | ~0.25 hrs per item | Item volume × complexity (calibration observed ~0.2; cushion kept) |
| Content QA | sampled: ~0.05 hrs per item, floor ~4 hrs | Volume, per the 2.2 QA-at-volume approach |
| Beta pass (full-site review, reports, release) | ~8 hrs near-fixed | - |
The scripting threshold: a content type clears it at roughly 30–40+ items. Below that, manual is cheaper - the decision is ROI, made per type in the 2.2 migration tiers.
Indicative Grid (Population)¶
Representative shapes - real projects price from their migration tiers and volumes:
| Size | Shape | Hours | Price |
|---|---|---|---|
| S | Mostly manual (~15 items) | ~15 | £2,000 |
| M | 2 scripted types + ~50 manual items | ~40 | £4,500 |
| L | 3 scripted types + ~60 manual items | ~45 | £5,000 |
| XL | 4–6 scripted types + ~200 manual items | ~90 | from £10,500 |
All + VAT.
The Cycle Line¶
12 hrs per cycle · S 1 · M 2 · L 3 · XL 5. At the 25:75 mix (P3 · M9) a cycle prices at £1,500, so the allowance is S £1,500 · M £3,000 · L £4,500 · XL £7,500. Cycle counts are not the size step k: XL carries 5 so the smaller unit does not quietly cut the provision at the top.
The cycle is deliberately a small unit: a shorter window, a tighter amends batch, and more of them. That makes the allowance saleable in finer steps and makes a variation cycle a £1,500 conversation rather than a £2,500 one. A cycle = client tests in the agreed one-week window → triage through the rubric → amends batch → re-release. The line covers client review support, the internal test track, triage, and amends implementation; it is sold at the sales moment, flexed by judgement (the client, complexity, historical learning), and confirmed in the Cornerstone. 5.4's close is near-fixed on top (~4–6 hrs + re-run). Cycles beyond the size allowance are priced variations at £1,500 each - or the Momentum route.
Calibration Flags (the Important Part)¶
Population lines:
- The reference actuals are rough (owner's own caveat) and the
basis varies per site:
- Concurrent - scripted fixed templates + manual bloks: 32h
- CDSDS - entirely manual, because volume was low: 22h
- Into Games - scripted + manual mix: 15h
- Source EV - population interleaved with Build as scaffolding, plus ~15h tail (the scaffolding pattern, now named)
- Tracsis - scripted fixed templates + manual bloks: 100h The rates above reproduce these within tolerance, but treat this model as provisionally fitted, not calibrated.
- Capture properly on the next project: script hours per source × type, manual items per hour, QA sample time - the same line-level discipline as Build, owned by the lead engineer.
- The two-pass pattern applies to scripted content only - manual re-population is duplicated effort and is never planned. The re-run pairs with the legacy content freeze at testing close (a 5.4/chapter 6 moment).
- The scaffolding pattern moves hours, not cost: population done during Build (where content is ready early) shifts effort backward and validates the model sooner - encourage it, and log the hours to this chapter's lines regardless of when they happen.
- Prices are hours × rate card rounded clean, as Design and Build. Margins and internal figures live here and nowhere else.
Cycle line:
- The cycle allowance is deliberately below the historical actuals (30 / 60 / 90 / 70 / 130 hrs across the five reference projects), which contained uncapped amends and pre-date the window/rubric discipline. The severity floor is unchanged - critical/high fixed before launch regardless. Capture per cycle and per severity to prove the allowance holds. The move to 12-hr cycles holds that gap rather than widening it. Against the tiers as they now stand, provision runs S 12 hrs · M 24 · L 36 · XL 60. XL's count was set to 5 rather than its size step of 4 precisely so the smaller unit did not cut the top end - at 4 it would have lost 20% against an allowance already below the evidence, in the chapter whose own flag 8 records that loops balloon. The five reference actuals (30 / 60 / 90 / 70 / 130 hrs) were all at or above today's XL shape, so XL's 60 hrs is the figure to check against the first captured actuals.
- Overservicing is logged, never silent. Critical/high fixes beyond the allowance are a deliberate quality choice - log the hours against this line even when unbilled, so the flex is visible at chapter close and the judgement factors calibrate.
- Loops balloon. The rare project that needs passes beyond the plan historically blew the line - the kickoff warning and the rubric are the controls; if a third window opens, the PM flags it to the Decider as a cost conversation then, not at invoice time.
- The Momentum route is the margin release valve: medium/low residue carried into the partnership converts overservicing pressure into recurring-revenue onboarding. Track how often each route is taken - it is a sales metric as much as a delivery one.