Hanzo True Savings

Your deployment

Deductions — not included until entered

Results

gross = software + infra + glue · run-rate = gross − ops · adjusted = run-rate − retained − fees − migration ÷ years

Do not present $941k as fully net until the final three deductions are entered.

Hanzo True Savings is an interactive operating-cost model, not a quote, guarantee, audit, or promise of results. Default prices are editable estimates based on representative technology stacks. Actual savings depend on existing contracts, usage, architecture, migration requirements, retained vendors, staffing, workload eligibility, and Hanzo deployment choices. Hanzo’s internal case-study figures are management-attested unless explicitly marked independently verified.

Click everything you pay for

Every vendor maps to a line in the table. Tap the ones on your bill — the totals and the vendors → 1 counter follow. select all clear

Category by category

Two axes, kept separate. Coverage carries weight and is clickable — full 100%, partial 50%, experimental 25%, roadmap 0%. Evidence carries no weight. A service can have full coverage while still management-attested. Separating the axes prevents an attestation from being presented as verification.

Prices are the analyst defaults at the reference deployment (100 employees, 30% engineers) and are editable — type over any of them. Untouched rows scale on their billing basis: per-seat with employees, per-engineer with engineers, platform fees not at all. Reset to analyst defaults

Infrastructure delta

The 60% delta above is a simplified default and a user-controlled assumption, not a universal benchmark. The egress anchor is real — DigitalOcean at $0.01/GiB against an AWS example at $0.09/GB — but it supports an egress comparison; it does not prove 60% for every workload. A full model would price provider, compute, GPU, database, storage, egress, observability, support, existing commitments, target Hanzo cost and migration overlap separately.

Glue engineering

The seams a vendor stack requires someone to own: auth bridges, webhooks, data sync, billing reconciliation, entitlements, audit exports, pipelines, vendor o11y, incident routing, schema transforms, secret rotation, API migrations, identity lifecycle, model fallbacks, security reviews. The default of 1.5 FTE at $220k fully loaded is conservative for a 30+ vendor estate, and 0.5 FTE of platform operations is subtracted on the other side.

Do not say Hanzo eliminates engineering. Hanzo redirects it from vendor seams to product.

Not just the bill — the build

Everything above prices what you stop paying. It ignores what you would have to build to own this surface yourself. Public anchors: Auth0 acquired for ~$6.5B · Datadog ~$1.55B of 2025 R&D · Zapier 800+ people and 9,000+ integrations. These are not additive valuations; they show each category is a substantial engineering domain. A replacement-cost study of the full repository estate is under analysis — no number will appear here until that report lands.

Exclusions — what this model never counts

Card interchange and network fees (the payment row counts PSP markup only) · frontier model pass-through, unless the workload is routed, distilled or self-hosted · deep vertical software (clinical, CAD, EDA, airline operations, industrial control, scientific instruments, jurisdiction-specific record systems) · migration coexistence, where both stacks run at once · retained vendors, since every row is independently deselectable.

Market context

Okta 2026: 98 applications on average, 259 at companies with 2,000+ employees, 72 below that. Zylo 2026 SaaS Management Index: a median of $9,455 of SaaS spend per employee, with 36% of licenses unused. The $613k software default at the reference deployment works out to $6,130 per employee — below Zylo’s enterprise-skewed median, so the model is conservative relative to the benchmark. None of this implies all SaaS spend is recoverable.

Second-order benefits — real, unpriced

SSO price premiums · per-vendor security reviews · DPAs · procurement cycles · duplicate observability · license reclamation · credential sprawl · integration attack surface · onboarding speed · unified policy · common metering · model portability · lock-in exposure · air-gapped deployment · incident response speed · fewer failure domains.

Material, but outside the headline until a defensible measurement exists.