Techbanq

Cost guide · data platforms

A data platform is priced by the mess it inherits

The platform itself — the warehouse, the pipelines, the catalogue — is the predictable part. What moves the figure is everything it connects to: how many source systems, how inconsistently they spell the same customer, how fresh the answers must be and who is allowed to see what. This page separates those drivers, and separates the build from the cloud bill that follows it.

Serving Saudi Arabia from our Riyadh office and our engineering hubs in Lahore and Karachi

Market
Saudi Arabia
Written for
CIO, Head of Digital
Decision
Budget a data platform across its build and its running cost
Languages
EN · AR

Information security for software delivery

  • ISO

    ISO/IEC 27001

    ISO/IEC 27001 certified

Certifications & registrations

Before the figure

The questions a data platform estimate has answered — whether or not it says so

Every estimate embodies answers to these. An estimate that does not state them has chosen quietly, and its assumptions surface later as change requests.

Which systems feed the platform?

Each source is a connector, a mapping and a reconciliation of its quirks.

What silence about it costs: Sources discovered mid-build, each arriving as an overrun.

How fresh must the answers be?

Overnight batch and always-current streaming are different architectures, not settings.

What silence about it costs: A batch platform asked to stream after it was built.

Who may see what?

Row-level rules, masking and audit trails are designed in, not sprinkled on.

What silence about it costs: Governance retrofitted under pressure from the first audit.

Who runs it on day ninety?

Your team, a managed service, or a mix — the split reshapes both quote and organisation.

What silence about it costs: A platform delivered to a team never trained to hold it.

The inputs

The six drivers hiding inside a data platform quotation

Each of these moves the figure in a direction you can reason about before any supplier names one. Together they explain why two platform quotes for the same organisation can differ honestly.

VariableEffect on priceWhy
Source-system count and conditionThe pipelines are priced by the mess, not the volumeConnecting a well-documented modern system is routine; reconciling an ageing ERP whose customer table disagrees with the CRM about who your customers are is the actual work. The effort scales with the number of sources and, more sharply, with their inconsistency — which is why profiling the data before quoting matters more than counting the systems.
Batch or streamingAn architecture decision wearing a checkbox's clothesReports refreshed overnight and dashboards current to the minute are built differently from the ground up: streaming brings its own infrastructure, its own failure modes and a higher operating discipline. The honest question is which decisions genuinely cannot wait until morning — and pricing streaming only for those flows, rather than for everything because it sounded better in the workshop.
Governance and access-control depthInvisible in the demo, decisive in the estimateA platform where every analyst sees everything is cheap to build and impossible to defend. Row-level access, masking of personal data, lineage and audit trails each add design and testing work, and regulated organisations need more of all of them. Governance is the part of the quote that buys nothing visible in a demo — and everything when a regulator or an internal audit asks who saw what.
The build-versus-managed splitDecides who you are paying, and for how longBuilding on managed services shortens the build and moves cost into the running bill; assembling more yourself lowers that bill and demands engineers who can operate what was assembled. Neither is universally right. What matters is choosing deliberately, because the split decides whether the platform's cost lives mostly in a project budget or mostly in an operating one.
The running cloud billIts own budget line, owed to a different companyCompute, storage and data movement are metered by the cloud provider and billed monthly for as long as the platform exists, and no build quote contains them. The build shapes the bill — partitioning, warehouse sizing and pipeline design all move it — but the bill itself belongs in your operating budget from the first slide, projected from expected usage rather than discovered from the first invoice.
Team enablementThe difference between a platform and an expensive shelfA platform nobody queries is a storage bill with a logo. Training analysts, establishing modelling conventions and seeding the first governed datasets people actually want are real lines of effort, and platforms that skip them plateau at a handful of users. Enablement is the cheapest driver on this list and the one whose absence most reliably wastes all the others.

Ask any supplier quoting this

  • Which of our source systems does the figure include, and what does adding one later cost as a process?
  • Which flows in this design are streaming, and what business decision justified each one?
  • Where are row-level access, masking and audit trails in the estimate — or are they assumed away?
  • What monthly cloud bill does this design produce at our expected usage, and what makes it grow?
  • Who operates the platform after handover, and what training for our team is inside the figure?

The second budget

The cloud bill is not the data platform running over budget

The most common data platform budgeting failure is not an inflated build quote; it is a correct build quote read as the whole cost. The platform then runs, the provider meters every query and every stored terabyte, and the monthly bill arrives at a finance team that approved a project, not an operating commitment. From that moment the platform is "over budget" — though nothing went wrong except the framing.

The remedy is to budget in two lines from the start: a build figure owed to the supplier, and a projected running bill owed to the cloud provider, estimated from expected usage and revisited as the platform grows. A supplier who helps you project the second line before you approve the first is pricing the platform; one who quotes only the build is pricing their own involvement in it.

Our role

What we profile before a platform carries a figure

  1. 01

    The sources, examined

    We inventory the source systems and profile the actual data before quoting, so the figure prices the reconciliation work that will really be done rather than a count of logos.

  2. 02

    Freshness argued from decisions

    We tie every streaming flow to a business decision that cannot wait, and default the rest to batch — because the cheapest architecture is the one that answers the real questions on time and no earlier.

  3. 03

    Both budget lines, projected

    The proposal carries the build figure and a projected running bill side by side, with the design choices that shape the second made explicit, so finance approves the platform's whole life rather than its first quarter.

Which questions the business needs answered, and how fresh those answers must be, are yours to rank; we translate that ranking into an architecture, a build figure and a running bill you can defend in the same meeting.

Common questions

What buyers ask

We only need a planning figure for a data platform — can you not estimate one sight unseen?+

An estimate produced without seeing your source systems would be an estimate of an imaginary organisation, and planning against it is riskier than planning without it — it hardens into an anchor the real scope is later blamed for missing. The useful shortcut is different: a brief source inventory and profiling exercise, bounded in days, that measures the two inputs which dominate the figure — how many systems and how messy. After that a planning figure means something. We would rather give you that exercise than a number that describes nobody.

Why is the monthly cloud bill not inside the data platform build quote?+

Because it is metered by a different company against your future usage, and no builder can honestly fix it in advance. What a good supplier can and should do is project it: estimate the bill from your expected data volumes and query patterns, name the design decisions that move it, and hand finance both lines before the project is approved. Treat a proposal that stays silent about the running bill as incomplete, and a proposal that promises to contain it inside a fixed build figure as something worse — a promise about another company's meter.

Does choosing streaming over batch always make a data platform dearer?+

For the flows that stream, yes in effort and in operating discipline — the infrastructure, the failure handling and the on-call expectations are all heavier than an overnight batch. The mistake is treating it as a platform-wide choice. Well-scoped platforms stream the few flows where a decision genuinely cannot wait — fraud checks, operational alerts, a control-room view — and batch everything else. If a proposal streams by default, you are paying always-current prices for answers that are read once a morning.

How much of a data platform budget should go to governance and enablement?+

Enough that both appear as named lines, and the right amount depends on how regulated you are and how new your analysts are to governed data. What we can say without a figure is directional: governance effort scales with your regulatory exposure and the sensitivity of the data, and enablement effort scales with how far your teams currently are from self-service. The warning sign is a proposal where neither word appears — that platform will be cheap to build, hard to defend in an audit, and quiet in daily use.

A scoping session that inventories your source systems, fixes the freshness and governance requirements, and puts the build figure and the projected running bill side by side before you commit.

Inventory your sources with us