Lean Canvas — market viability

Status: Draft for review · Derived analytical artifact of the solution specification · June 2026

A viability lens over the spec set. The problem/solution/customer blocks are drawn from locked decisions (problem-definition.md, solution-definition.md, README.md). The commercial blocks — revenue, cost, channels, metrics — are unvalidated leans, flagged ⚠️ where they rest on assumptions still to be tested. The point of this document is less to assert the model than to expose the riskiest assumptions (§11) before money is spent.

1. The canvas

The nine blocks in the canonical Lean Canvas arrangement — a top band of five (each tall column carrying a lower strip), over a two-box base. Cells are deliberately terse; the full treatment of each block is in §2–§12. ⚠️ marks a commercial block resting on an unvalidated assumption (see §11–§12). The layout uses plain pipe tables (no merged cells) so it renders in both PDF and HTML through pandoc.

1 · Problem	4 · Solution	3 · Unique value proposition	9 · Unfair advantage	2 · Customer segments
No single trusted asset picture — GIS, work system, SCADA, models and documents disagree · quality unknown and unprovable as regulators start to expect proof · assets findable by tag, not the names staff use	Reversible identity merge/split — conflicts retained, not overwritten · fit-for-purpose (per-skin) quality + regulator-ready evidence · audited inbound curation of messy handovers	”Know your assets — and prove it — without ripping out the systems you run.” One trusted, source-traceable picture from records that disagree; quality measured per use; stays correct as sources change	Co-equal multi-language identity (not tag-primary) · reversible runtime reconciliation with structural conflict retention — not MDM, not AI-extraction · author’s decade of domain IP + a working reference engine · warm access + believer network = a head start, not a moat	Australian regional / mid-sized water utilities — the tier below the metro majors, holding brownfield assets of unknown data quality
Existing alternatives: spreadsheet reconciliation · GIS as de-facto register · consultant data-cleanses · MDM/EAM that overwrite conflicts · live with it	8 · Key metrics: time to first gap-and-conflict view on own data · sources reconciled / conflicts resolved · pilot-to-paid conversion and expansion · demonstrable merge/split correctness	High-level concept: a reconciliation layer over your existing systems — version-control-grade provenance where corrections never destroy the original	5 · Channels ⚠️: relationship-led into SEQ early adopters · the standing advisor/tester network · WSAA + regulator/maturity pull · pre-qualified procurement panels (Local Buy / VendorPanel) to bypass full tenders · integrator partnerships later	Early adopters: SEQ — City of Gold Coast, Urban Utilities, Unitywater · buyers: asset-management lead (economic buyer), data steward (champion)

7 · Cost structure ⚠️	6 · Revenue streams ⚠️
Product R&D — reconciliation engine (TypeScript), core connectors, skin/quality tooling, UI (company-borne) · per-customer implementation & maintenance services (cost-recovered via the implementation fee) · AU-resident hosting + sovereignty/IRAP path · discovery-led consultative sales · shape: high up-front R&D, low marginal cost per added same-tier utility	Implementation fee at pilot/initiation — connectors, skin config, onboarding — creditable against SaaS licence fees · annual SaaS subscription per utility (single-tenant, AU-resident), tiered by size / sources / users · ongoing services / maintenance · paid time-boxed pilot on the utility’s own data (the wedge)

2. Problem

The top problems for the protagonist (detailed in problem-definition.md):

No single trustworthy picture of an asset. What’s known is scattered across GIS, the work/asset system, SCADA, hydraulic models, spreadsheets and as-builts — and they disagree.
Quality is unknown and unprovable. “Is our data good enough?” has no defensible answer, yet regulators increasingly expect the utility to demonstrate asset knowledge.
Records speak one language. Assets are findable by engineering tag but not by the common names, locations and SCADA points staff actually use — so knowledge stays in people’s heads.

Existing alternatives (what they do today): manual reconciliation in spreadsheets; treating GIS as the de-facto register; periodic consultant data-cleansing projects; large MDM/EAM suites that assume a single master identifier and overwrite conflicts; or simply living with it. None retain conflicting truth with provenance, and none answer quality per use.

3. Customer segments

Beachhead: Australian regional / mid-sized water utilities — the tier below the metro majors, holding brownfield assets already in service whose data quality is unknown.
Early adopters (SEQ, warmest access): the council-run City of Gold Coast and the distributor-retailers Urban Utilities and Unitywater. The rest of the national tier is the expansion path.
Buyers & users within the account: the asset-management lead (economic buyer — owns the “prove you know your assets” obligation), the data steward (champion — lives the handover/conflict pain), the capital/renewal planner and operations (value beneficiaries). See the S1–S6 scenarios in requirements.md.
Explicitly not (for now): metro majors (favour incumbents, long cycles) and mid-tier private industrials (no regulatory forcing function) — both examined and set aside.

4. Unique value proposition

Know your assets — and prove it — without ripping out the systems you already run.

STREAM turns records that disagree into one trusted, source-traceable picture: it shows what each source says, where they conflict, what’s missing for the job at hand, and how far the data can be trusted — and it stays correct as sources come and go.

In a phrase, four C’s: continuity, consistency, completeness — and therefore confidence. The first three are what STREAM delivers; the fourth is what the owner gets.

High-level concept (for fast comprehension): a reconciliation layer that sits over your existing systems — version-control-grade provenance for asset data, where corrections never destroy the original.

5. Solution (top capabilities that carry the UVP)

Reversible identity merge/split — one entity from many sources, transitively merged and un-merged when a source is withdrawn; conflicts retained, not overwritten (the load-bearing mechanism; solution-definition.md §2, architecture.md §2).
Fit-for-purpose (“skin-relative”) quality + regulator-ready evidence — completeness and consistency reported per use, not as one global score (solution-definition.md §3).
Inbound curation of messy handovers — bring in contractor/project data, see agreement and conflict, admit what’s trusted with a recorded reason and full audit trail (S5).

These map 1:1 to the wedge MVP in requirements.md §5 — “show them their own problem.”

6. Channels ⚠️ (lean — validate)

Direct, relationship-led into the SEQ early adopters (existing warm access).
The standing network of advisors, early testers and prospective customers from STREAM’s first life — re-engaged via the discussion-pack.md (a channel, not just a review artifact).
Industry credibility routes: WSAA and the water-industry community; regulator/maturity expectations (IPART/ESC, and the UK Ofwat maturity precedent) as a “reason to act now” pull.
Later: partnerships with GIS / digital-twin / CMMS integrators (complement-not-replace positioning makes STREAM a partner, not a threat, to those stacks).

Buying pathway — delegations of authority & tendering (go-to-market reality)

How these buyers are allowed to purchase shapes deal size, sales-cycle length and pricing as much as willingness to pay does. Two mechanics dominate, and a third is the shortcut around them.

Financial delegations of authority — who can sign at what value. Queensland councils delegate spending authority by council resolution to the CEO and, in turn, to officers, recorded in a single public delegations register (Local Government Act 2009 (Qld) ss257, 259–260). The SEQ distributor-retailers (Urban Utilities, Unitywater) are statutory bodies with their own board-approved delegation frameworks, not councils (SEQ Water (Distribution and Retail Restructuring) Act 2009). Practically, the champion (data steward) and even the asset-management lead usually hold a limited delegation; a material contract climbs to an executive, the CEO, or — for councils — a council resolution. Implication: know which delegation tier each deal size triggers, and arm the internal champion to carry it upward.

Tender thresholds — above a set value a public tender becomes mandatory, which lengthens the cycle and favours incumbents. For Queensland councils (Local Government Regulation 2012 (Qld)): below ~$21,000 ex GST — no quote/tender requirement; medium (≥ ~$21,000) — at least three written quotes (s225); large (≥ ~$280,000 ex GST) — a public written tender advertised ≥ 21 days (ss226, 228). (These Qld figures were raised in Dec 2025 and re-index by CPI each 1 July — s223E; confirm the current published figure before quoting.) For the broader tier: NSW council Local Water Utilities must tender at ≥ $250,000 (Local Government Act 1993 (NSW) s55(3)(n)); Victorian regional water corporations are state-owned and set their own board thresholds under the VGPB framework (no single figure).

The shortcut that matters — pre-qualified supplier panels. A council can buy through an approved panel without running its own tender, even above the large threshold — the “LGA arrangement” exception (Local Government Regulation 2012 (Qld) s234). The dominant panel operator for Qld councils is Local Buy (the LGAQ’s procurement company), and most councils run quotes through VendorPanel (Local Buy; VendorPanel — local government); equivalents exist in NSW (Local Government Procurement) and via state ICT arrangements.

What this means for STREAM:

Price the entry deliberately. Sizing the pilot + implementation fee under the relevant tender threshold (≈ <$280k for a Qld council, <$250k for a NSW LWU) keeps the first engagement in quotes territory — weeks, not the months a public tender adds — while the creditable-fee model (§7) still funds delivery.
Pursue panel membership early. Getting STREAM onto a Local Buy / NSW LGP / state-ICT software panel, and quotable in VendorPanel, is a channel in its own right — it converts “must run a tender” into “can buy by quote,” the single biggest timeline risk in public-sector sales.
Map the delegation chain per account during discovery (§11.1, §11.7).

Caveat: the SEQ distributor-retailers’ and individual councils’ own thresholds and delegation limits sit in internal policies that aren’t all public — confirm per account in discovery rather than assuming the statutory defaults.

7. Revenue streams ⚠️ (lean — not a locked decision)

Implementation fee (services) — charged up front at pilot / initiation. Every sale and every maintenance engagement carries a services component (connector configuration, skin/template authoring, data onboarding). It is borne by the customer as an implementation fee, not absorbed — it funds the high-touch work that lands and beds in an account. Crucially, it is creditable against subsequent SaaS licence fees, so the customer experiences it as a head start on their subscription rather than an extra charge: it eases the “yes” while still covering the cost of delivery.
Annual SaaS subscription per utility (single-tenant, AU-resident), tiered by network size / sources / users — the recurring core, against which the implementation fee is credited. Indicatively ~A$120–250k/yr (top of band for the 100k+-connection majors), deliberately set below what these utilities already pay for the enterprise GIS that depends on this data.
Ongoing services / maintenance: further connectors, new skins, data work and support beyond the subscription baseline — a continuing services line as the deployment grows.
Pilot: a paid, time-boxed proof on the utility’s own data (the wedge) — the natural place the implementation fee is first charged and the credit mechanism introduced; priced to be an easy “yes” and designed to convert.
Pricing anchor (evidence-based, triangulated — see §12). These bands are inferred from what comparable AU water utilities already spend on adjacent systems, not yet from observed sales of this category: a NSW regional utility pays ≈A$310–350k/yr for an enterprise Esri GIS agreement (Central Coast Council contract register); EAM overhauls run into the millions (Melbourne Water’s Maximo program ≈A$20M, itnews); and asset-condition/data programs run A$0.2–2.0M each, per year (same Central Coast register). Pricing a reconciliation layer below a single GIS line is defensible — but willingness to pay for this category is still unproven (§11.1).
Open pricing questions deferred but flagged: the implementation-fee size and credit ratio (full vs. partial offset, and over what period); subscription vs. consumption; whether to size the entry (pilot + implementation fee) under public-tender thresholds (see §6) to keep early deals in quote territory; self-host licence for sovereignty-strict buyers.

8. Cost structure ⚠️ (lean)

Product R&D (company-borne): the TypeScript reconciliation engine, core connectors, skin/ quality tooling, UI — the platform investment, funded by the business and not billed to any one customer. This is the genuine sunk cost.
Per-customer implementation & maintenance services (cost-recovered): the delivery work for each account — connectors to their systems, skin/template config, onboarding, ongoing data work and support. A real cost, but recovered through the implementation fee and ongoing-services revenue (§7), not absorbed — so it is a throughput line, not a margin drain.
Hosting & sovereignty: AU-data-resident single-tenant + self-host option; the IRAP / Hosting Certification path many government-adjacent buyers will require (architecture §5).
Go-to-market: discovery-led, consultative sales (long-ish public-sector cycles).
Shape: high up-front product R&D + low marginal cost per added utility once core connectors and default skins exist; per-account services are cost-recovered, so margin improves with each same-tier account (expansion thesis).

9. Key metrics ⚠️ (lean — instrument early)

Activation (the “aha”): time to a user’s first gap-and-conflict view on their own data with a fit-for-purpose quality answer — the MVP’s whole reason to exist.
Engagement/value: sources reconciled per deployment; conflicts surfaced & resolved; fit-for-purpose quality lift over time; regulator-ready evidence produced.
Commercial: pilot → paid conversion; deployments live; net revenue retention / expansion within a utility (more skins, more sources) and across the tier (second utility).
Trust (existential): demonstrable merge/split correctness — the precondition everything else rests on (requirements.md §2.4, architecture.md §7).

10. Unfair advantage

The block hardest to copy — assessed honestly (✅ durable / ⚠️ erodes):

✅ Co-equal multi-language identity. Most tools privilege the engineering tag and treat common names/locations as second-class; STREAM is built tag-agnostic. Competitors are structurally committed the other way — hard to retrofit.
✅ Reversible runtime reconciliation with structural conflict retention. Architecturally distinct from MDM (last-writer-wins) and from AI document-extraction plays (DIGATEX) and class libraries (Datum360/Autodesk). Copyable eventually, but it’s a different product centre.
✅ Founder–market fit & a working reference engine. The original author’s decade of domain depth, the v1.2 engine (incl. the now-confirmed merge/split test suite) as a proven blueprint, and the documented lessons — not reproducible by a new entrant from cold.
⚠️ Warm beachhead access + a pre-existing believer network. Genuinely valuable early, but relationships aren’t a moat at scale — it buys a head start, not permanent defensibility.

11. Riskiest assumptions to validate (the viability teeth)

Ordered by kill-the-business-if-wrong. Each needs a cheap test before heavy spend.

Willingness to pay & budget authority. The §7/§12 band is now anchored to verified adjacent spend (Esri GIS, EAM, condition programs) — but that proves these buyers spend at this scale, not that they’ll pay for this category. Do they have (or can route) a line item for an assurance/reconciliation layer at that band — and does the asset-management lead actually control it? Test: pricing & budget conversations in the SEQ discovery round; a paid pilot is the real signal.
Acuteness of the pain vs. “good enough” status quo. Is unknown data quality a funded priority now, or a known-but-tolerated irritation? Test: do early adopters fund a pilot on their own data, or only nod along?
The re-education cost. The category (“reversible reconciliation / per-use quality”) isn’t one buyers shop for. How long/expensive is it to make them want it? Test: does the discussion-pack + own-data demo shorten the conversation, or stall it?
Reachability of the rest of the tier. SEQ is warm; the national tier (council LWUs, regional corporations) is mostly cold and procurement-heavy. Test: one referral/expansion beyond the warm three.
Sovereignty/IRAP as a gate, not a footnote. If buyers require certification before a pilot, the up-front cost and timeline shift materially. Test: confirm the actual bar in SEQ discovery (architecture §5 spike).
Connector reality. Can 2–3 of a real utility’s sources be ingested at acceptable effort? If every account is a custom integration project, margins (§8) break. Test: the MVP itself.
Procurement pathway & delegations. Can an early deal be done by quotation (priced under the tender threshold, or bought through a pre-qualified panel) rather than a full public tender, and is the champion’s delegation enough to drive it? If every sale triggers a months-long open tender, the cycle and incumbent advantage rise sharply. Test: in SEQ discovery, map each target’s delegation tiers and panel memberships, and price the entry accordingly (see §6).

12. Market sizing (sourced — June 2026)

Entity counts are now sourced; the revenue figures still depend on the §7 pricing band (anchored to adjacent spend, not yet observed for this category — so treat ARR as a ceiling, not a forecast).

TAM — asset owners with fragmented brownfield data needing reconciliation (water + adjacent utilities/infrastructure, AU and beyond): large but diffuse; not the near-term play.
SAM — the regional / mid-sized water tier below the metro majors: ~177 entities. NSW 92 council Local Water Utilities (89 council-run, serving ~1.85M people outside Sydney / Hunter — NSW Audit Office); Victoria 12 regional urban water corporations (DEECA); Queensland ~72 urban water/sewerage providers (qldwater); TasWater 1. The commercially-addressable subset is ~150 (the Qld count includes ~15–20 very small remote/Indigenous-council providers below practical deal size).
Sizing by network (BoM National Performance Report; participation expected above 10,000 connected properties; bands 100k+ / 50–100k / 20–50k / 10–20k — BoM NPR). The beachhead skews large: Urban Utilities ≈677,000 connected properties / ~1.55M people / 9,828 km mains; Unitywater 800,000+ customers — both top-band, supporting top-of-band pricing.
SAM revenue ceiling — ~150 addressable × a blended ~A$150–200k/yr (§7) ≈ A$22–35M/yr ARR at full penetration. A ceiling, not a forecast.
SOM (3-year, beachhead) — SEQ first, then nearest-tier expansion: a handful of accounts (low single digits → ~10), i.e. ~A$0.5–2.5M ARR — enough to prove the model and fund the expansion motion, not enough to rest on.
Implementation fees are not additive to these ARR figures. Being creditable against the subscription (§7), they mostly smooth early cash flow and fund delivery rather than add to steady-state revenue — so ARR remains the headline number, not ARR-plus-services.

Read: the beachhead is small but real and reachable — viable as a wedge and proof, with the thesis depending on (a) converting warm SEQ access and (b) the §8 cost shape letting each additional same-tier utility land cheaply. The §11 assumptions are what determine whether that holds.

13. Competitive landscape & positioning (scan — June 2026)

Grounded in what the beachhead utilities actually run and what the named competitors have actually deployed in AU water. “No AU-water footprint” below means none could be verified in tender awards, case studies or customer lists.

The GIS substrate is Esri — and its modernisation is STREAM’s opening. ArcGIS is the de-facto GIS system-of-record across all five SEQ targets (Queensland Urban Utilities), and SA Water, WaterNSW, Melbourne Water, Hunter Water, Icon Water. The active wave is migration to ArcGIS Utility Network, where “data of unknown quality” pain peaks — Hunter Water engaged a specialist for data validation during its 2025 migration: exactly STREAM’s adjacent, funded pain. (Notable exception: Sydney Water runs GE Smallworld — Esri is dominant, not universal.)
The asset/EAM layer is fragmented — and that fragmentation is the wedge. Across the five SEQ targets every major utility runs a different asset/works system: Urban Utilities on ABB/Hitachi Ellipse, Unitywater on IBM Maximo (itnews), Seqwater on TechnologyOne, City of Gold Coast on Hansen / Infor Pathway, Logan and Redland on Assetic (trending to TechnologyOne). Nothing unifies them — precisely the multi-source reconciliation problem STREAM solves, and why the rip-and-replace EAM model (Melbourne Water’s ~A$20M Maximo overhaul) is the incumbent approach STREAM positions against.
The ‘single source of asset truth’ contenders have no AU-water foothold. Cognite Data Fusion, Bentley iTwin and Datum360 (now Autodesk) pitch a unifying data/twin layer with ambition overlapping STREAM’s — but none has a verifiable AU water deployment (Cognite has no AU presence at all; its only “Unitywater” hit is a naming collision with Cognizant). AVEVA AIM, Hexagon’s spun-out Octave/SmartPlant and DIGATEX are oil-&-gas/process-centric with no AU water penetration.
Complement vs compete. STREAM complements (ingests-from, sits beside, does not replace) Esri ArcGIS, the heterogeneous EAMs (Maximo / Ellipse / Assetic / Hansen / TechnologyOne) and AVEVA PI (SCADA historian) — its value is strongest where these meet. It competes on ambition with Cognite / Bentley iTwin / Datum360, none of which is entrenched in AU water. And every enterprise rival here is quote-only / pricing-opaque — a clear, value-anchored price (§7) is itself a differentiator.

(Scan corrections, to avoid repeating errors: Octave was a Hexagon spin-off distribution, not an IPO; the council “Hansen” is Infor Pathway, not Advanced; no DIGATEX–EPIC lineage exists.)

14. Regulatory timeline & value evidence (sourced — June 2026)

Grounds the “why now” (§6, problem-definition.md) and the “why it’s worth paying” behind the price.

The regulatory tide is moving toward “prove you know your assets.” ISO 55001:2024 adds an explicit “managing data and knowledge” requirement; transition from the 2014 edition ends 31 July 2027 (ISO). AS 5488.1:2019 already classifies buried-asset data into quality levels QL-A → QL-D with provenance (SAI). Economic regulators tie price to a credible asset case — IPART’s Sydney/Hunter determinations run to 2030 and test capex prudency/deliverability (IPART); Victoria’s ESC assesses forecasts under PREMO, whose “Management” criterion demands data-supported plans (ESC). (NSW council LWUs sit under the state Best-Practice Management framework + BoM reporting, not IPART.)
Leading indicator — regulation travels. UK Ofwat is making asset-management maturity an enforceable licence condition: PR24 asset-health roadmap (Dec 2024), decision to impose (Nov 2025), licence modifications in 2026, sector-wide maturity assessment in 2026 (Ofwat). The clearest signal of where AU expectations are heading.
The cost of not knowing (what licenses the spend): median >12.7 water-main breaks per 100 km/yr nationally (BoM NPR 2023–24); ~A$8B to rehabilitate Australia’s ~40,000 km of asbestos-cement main (RNZ), with one utility’s AC maintenance forecast to rise ~A$10M → ~A$40M/yr (AWA); an urban-water asset base of ~A$170B with ~A$4.5B/yr capex rising toward >A$10B/yr by ~2027 (Aust Govt) — capital largely committed on today’s untrusted data.

The pitch this licenses: ~A$120–250k/yr is small against a multi-million-dollar annual condition/data wallet that exists because the data is untrusted — and against the regulatory cost of being unable to prove asset knowledge.

15. Discovery & validation plan

Turns the riskiest assumptions (§11) into the SEQ discovery round — what to ask, whom, and what would falsify the assumption (a pre-mortem, not a sales script). Desk research has taken the researchable items as far as practical; these are the ones only live conversations can close.

# (→§11)	Question to put	Who	What confirms vs falsifies
1 WTP & authority	”For a layer that reconciles your GIS/EAM/SCADA and proves data quality per use, what’s a fair annual figure — and from whose budget?”	Asset-mgmt lead; finance/portfolio owner	Confirms: names a band overlapping A$120–250k from a routable line. Falsifies: “no line / needs a business case we can’t fund.”
2 Pain is funded	”What do you spend now to find the true state of your assets — condition surveys, data clean-ups?”	Asset/capital planner	Confirms: live, funded condition/data programs. Falsifies: “we live with it; nothing funded.”
3 Re-education cost	Show the own-data gap-and-conflict view. “Would you act on this?”	Data steward + asset lead	Confirms: unprompted recognition / asks for more. Falsifies: needs repeated explaining to see value.
4 Reach beyond SEQ	”Who else like you should see this?”	Any beachhead contact	Confirms: a warm referral converts. Falsifies: no path beyond the warm three.
5 Sovereignty/IRAP	”Your bar for a system holding asset data — AU-resident? IRAP? self-host?”	IT / security / architecture	Confirms: a pilot can run pre-certification. Falsifies: certification required before any pilot.
6 Connector reality	”Can we ingest 2–3 of your real sources in the pilot?” (then do it)	Data steward / IT	Confirms: 2–3 sources in at acceptable effort. Falsifies: every source a custom project.
7 Procurement path	”Could this start as a quote, or via a panel (Local Buy / VendorPanel)? Whose delegation covers it?”	Procurement + champion	Confirms: a quote/panel route within the champion’s reach. Falsifies: a full public tender is unavoidable.

Run these in the SEQ beachhead (City of Gold Coast, Urban Utilities, Unitywater) before committing to the build. The cheapest decisive test for #1–3 is a paid pilot on their own data — it converts opinion into signal.

This canvas is a living viability hypothesis, not a plan of record. Update it as the SEQ discovery round (the §15 plan; requirements.md §4) returns evidence — especially on §11.1–3, which gate everything else.