AI agents can only act on what your system tells them. If organisation records are duplicated, codes are inconsistent or milestones are missing, an agent will make the same mistakes your team quietly fixes every day, only faster. The good news is that data readiness is practical work you can start this quarter.

Why data quality decides AI results

Your experienced staff are very good at working around messy data. They know that “ABC Ltd” and “A.B.C. Limited” are the same customer, or that a missing milestone usually means the carrier hasn’t updated yet. An AI agent doesn’t have that background knowledge. It takes the data at face value.

That means clean data is no longer a nice-to-have. It is the foundation that decides whether automation saves time or creates extra work.

The usual suspects

In almost every system we review, the same four problems come up:

• Duplicate organisations: the same customer, supplier or carrier set up two or three times, each with slightly different details.
• Inconsistent codes: different teams using different codes for the same port, package type or service level.
• Free-text fields: important information typed into notes or comments instead of structured fields, where a system can’t reliably read it.
• Missing milestones: key events like departure, arrival or delivery never recorded, so the shipment history has gaps.

None of these are unusual. They build up naturally over years of busy operations. The key is to find them and fix them before you automate.

A simple 30-day data health check

You can get a clear picture in a month without disrupting daily work:

• Week 1: Pick your scope. Choose one branch, one trade lane or one customer group. Agree which data matters most for the workflow you want to automate.
• Week 2: Measure. Count duplicates, check how often key fields are empty, and list the codes in use. Simple reports are usually enough.
• Week 3: Fix the biggest issues. Merge duplicates, standardise the most-used codes and fill in missing data for live shipments.
• Week 4: Prevent it coming back. Agree simple rules for new records, assign owners and set up a regular check.

At the end of the month, you’ll know how ready you are and where to focus next.

Who owns master data (and why it can’t be “everyone”)

When everyone can create and edit master data, no one is truly responsible for it. That is how duplicates and inconsistencies creep in.

Good practice is to name a small group of data owners, for example one person for organisations, one for rates and one for codes. They approve new records, handle change requests and keep standards up to date. Everyone else can still request changes, but there is a clear path and a clear decision maker.

Quick wins inside CargoWise

If you use CargoWise, there are several practical steps you can take straight away:

• Use the system’s own duplicate checks before creating new organisations
• Review and tidy code lists so teams pick from a short, agreed set
• Make key fields mandatory where they matter for automation
• Use workflow templates so milestones are created and tracked consistently
• Set up simple reports that flag incomplete or inconsistent records each week

These changes are small on their own, but together they make a big difference to how reliably AI can work with your data.

How Zinnovate supports data clean-up and governance

Zinnovate helps forwarders get their data AI-ready, from a one-off health check to a lasting governance model. We can:

• Run a data health check and give you a clear, prioritised list of fixes
• Clean up duplicates and standardise codes inside CargoWise
• Set up ownership, rules and regular checks so the improvements last
• Connect data readiness to your AI and automation plans

If you’d like to find out how ready your data is, talk to our team.