Digitization vs digitalization vs digital transformation

The invoice used to arrive on paper. Now it arrives as a PDF, gets emailed to three people and waits for the same manager to approve it.

Something changed. The supplier might still wait just as long to get paid.

Digitization turns information into digital form. Digitalization uses digital tools and data to change an activity. Digital transformation describes wider changes to how an organisation operates and delivers value through those technologies.

Those differences help when deciding what a project should achieve. A searchable file, a quicker approval and a different way of serving suppliers need different work and different evidence.

The difference, using the same invoice

Imagine a distributor receiving supplier invoices by post. Finance types the details into the accounts system, walks a copy to the buyer and waits for approval. Suppliers call Finance to ask where their payment is.

Here are three possible changes. This is a fictional example, not a sequence every business must follow.

Term	Change in this example	What still needs checking
Digitization	Scan the paper invoice and capture its fields	Is the digital copy accurate and easy to find?
Digitalization	Route the invoice electronically, match records and show exceptions	Does it reduce handling and waiting, including awkward cases?
Digital transformation	Connect purchasing, receipt confirmation, approval authority and supplier support around a shared service	Does the new way of working produce reliable payments and clear ownership?

The OECD's definitions distinguish converting information, changing activities with digital technology, and wider economic and societal effects. Here we're applying that distinction to one business. The labels overlap in practice; they aren't a three-level certification.

Start with the file

Scanning makes the invoice available without passing paper around. Optical character recognition, usually called OCR, can also extract fields such as supplier, invoice number, amount and due date.

An image of a number and a correctly captured number are different things. If OCR reads €8,200 as €820, the file is digital and the payable amount is wrong. Someone needs a way to check uncertain fields against the original.

Good digitization can be the whole project. Replacing a vulnerable paper archive with searchable records may solve a worthwhile problem. It doesn't need a larger name to justify the work.

For the supplier waiting for payment, though, the unchanged approval queue is still there. That takes us from the file to the process.

Change how the work moves

Suppose the team adds an invoice workflow. It checks for duplicates, connects the invoice to the purchase order and goods receipt, then routes any discrepancy to the person who can resolve it.

Take two invoices from our fictional distributor. One is for €1,200 and matches the approved order and receipt. The other is for €8,200, but the invoiced quantity doesn't match what the warehouse recorded.

They shouldn't receive the same treatment merely because both arrived as PDFs.

The workflow can make the mismatch visible immediately. It can't resolve it by sending the same notification every morning. The warehouse may need to correct its receipt, or the supplier may need to amend the invoice.

That is digitalization: the technology changes how the activity happens. It may remove retyping, expose waiting or route work more reliably. The team still needs to test those results, including rejected invoices and manual follow-up.

My note on what to automate first goes further into choosing between integrations, workflows and AI. A rule that checks two quantities doesn't need an AI agent to make it useful.

Change the responsibilities around the process

The workflow could be technically perfect and still send every invoice to an absent manager. Approval authority and cover arrangements belong in the operating design.

For this distributor, a broader change could give authorised team leads responsibility for matching invoices below an agreed limit. Larger invoices would go to a named approver. Quantity discrepancies would go to receiving, with an escalation route when they remain unresolved.

Purchasing would keep order data usable. The warehouse would confirm receipts promptly. Finance would own the payment process, and suppliers would see status without calling several departments.

Changing those responsibilities, controls and service expectations across teams gives us a stronger reason to call the work transformation. Installing the portal by itself wouldn't establish that the change had happened.

The RACI example helps make those responsibilities explicit. Then change management addresses whether people can and do work that way. An approval route nobody uses leaves the old process running alongside the new one.

Measure the result the project actually promised

Each change needs evidence that fits its purpose.

Change	Evidence worth collecting	A weak substitute
Digital invoice records	Missing pages, field accuracy, retrieval time and access checks	Number of files scanned
Invoice workflow	Handling time, time waiting for approval, exception age and duplicate payments	Number of automated steps
Wider operating change	Use across teams, unresolved handoffs, payment reliability and supplier enquiries	Number of systems connected

Keep handling time separate from elapsed time. Cutting five minutes of typing may help Finance while leaving a three-day approval wait unchanged. Measure both from the same start and finish events, with a consistent calendar.

Any financial claim needs another check. Released time can absorb more work without reducing payroll. The benefits realisation example shows how to separate the operating improvement from an expense that actually falls.

Vrolen's manufacturing software guide applies a related question to production systems: which responsibility belongs where, and what information must move between them? Its setting is a factory, but the ownership problem is familiar.

Decide where the current problem ends

Before expanding the project, follow a few real invoices from receipt to payment. Find where information changes form, where work waits and who can release it.

If people lose time finding records, improve the records. If they retype the same fields, connect the systems. If a correct invoice waits because nobody has authority to approve it, address that authority before buying another tool.

A project can combine all three kinds of change. Describe each one clearly enough that the team knows what to build and what to check afterwards. For the supplier, the test is whether the agreed payment happens reliably, with a clear answer when something goes wrong.

Sources

• OECD, 2026. Investment Policy Framework for Digital Transformation, introduction and Box 1.1. Definitions at economy and society level; the company comparison is an applied explanation.

The distributor, invoice amounts, approval limit and proposed routes are fictional. No processing-time improvement, compliance outcome or financial saving is claimed.