Data Mi
Better data. Ready for what’s next
Data Mi cleans, structures and transforms business data through customised, automated and repeatable processes. Whether you are changing systems, fixing unreliable master data or rebuilding the same spreadsheet every month, next month it runs again without anybody rebuilding it.
One customer, four ways
SOUTHERN FOODS GRPSouthern Foods Group P/Lsthn foods grp — ACCT 4471SOUTHERN FOODS GROUP PTY LTD4 customer IDs · 4 credit limits · 2 payment termsOne customer
Southern Foods Group Pty LtdACN 004 112 887One customer ID · one credit limit3 former spellings kept as search aliasesRule written down and re-runnableThe problem
Bad data creates work everywhere else
Duplicate customers. Inconsistent product codes. Missing fields. Different classifications between sites. Files that need hours of work before anybody can use them. Businesses work around these for years.
How it works
Clean. Map. Transform. Validate. Repeat
Five steps. The first one is done once. The other four become a process that belongs to you and runs on demand.
Steps two to five run again every time, unchanged
What it looks like
The same customer, before and after
A small example of a very large problem. Multiply it by every customer, supplier, item and site you have, and you have most of what stalls an ERP migration.
One customer, four ways
SOUTHERN FOODS GRPSouthern Foods Group P/Lsthn foods grp — ACCT 4471SOUTHERN FOODS GROUP PTY LTD4 customer IDs · 4 credit limits · 2 payment termsOne customer
Southern Foods Group Pty LtdACN 004 112 887One customer ID · one credit limit3 former spellings kept as search aliasesRule written down and re-runnableMapping
Every rule written down, and signed off before anything runs
This is what a mapping document looks like at the level that matters. Your people agree each rule. Nothing is decided by us on your behalf, and nothing is decided twice.
CUST_NMCustomer nameTrim, title case, strip trading-name suffixADDR1 + ADDR2 + ADDR3Street / Suburb / State / PostcodeSplit and validate against Australia PostPROD-CD (legacy)Item code (new)Cross-reference table, 4,182 rowsTYPE = G / C / IGeneral / Commercial / IndustrialLookup. Unknown values escalatedDATE_STR (12 formats)ISO 8601 dateParse by pattern. Ambiguous dates escalatedFix the process, not the file
You keep the process
A contractor cleans the file and leaves. Data Mi writes the rules down, automates them and hands them over, so the same work happens again next month without a person and without a debate.
Automate the obvious. Escalate the judgement calls.
Not every data problem needs a person, and not every data problem should be settled by a rule. Anything the rules cannot resolve is listed and put in front of somebody in your business. Nothing is silently guessed.
A real run
What happened to every record
The honest version of a migration status report. A vendor reporting only the first number is not telling you about the second, and the second is the one that decides whether you go live on time.
The rule was unambiguous, so the record was transformed and logged.
Two rules could both apply. A person in your business decides which one wins.
The source record is not viable. Listed with the reason, never silently dropped.
Use cases
Where this earns its place
Data migration
ERP, TMS, WMS and billing migrations. Profiling, mapping, cleansing, trial loads, reconciliation and cutover.
Data migration servicesData transformation
The monthly file somebody rebuilds by hand, turned into a process that runs itself and belongs to your team.
Data transformationData quality
Customer, supplier, item and site records reconciled to one version, with standing checks that report drift early.
Data qualityAcquisitions
Bringing an acquired business’s customers, products and history into your systems without losing what matters.
AI readiness
AI will not fix bad data. If the source is duplicated or inconsistent, the quality of what comes out is capped before anybody starts.
Your data
Processed here. Kept only as long as you want
Data Mi handles some of the most sensitive information in your business, so the rules about it are simple and stated.
Worth saying plainly
Your data is not sent to external AI services as part of this work. Where we use automation, it runs inside Mattingly’s own environment in Australia.
How an engagement runs
Nothing runs until you’ve signed off the rules
Profile what you have
We read the actual data and show you what is in it. This is usually the uncomfortable part.
Agree the rules
Every decision is written down and signed off by your people before anything runs.
Build and trial
The process runs on real data. You get an exceptions list, not a claim of success.
Hand it over
The process is documented and yours. It runs again next month without us.
Questions
Data Mi, answered
Data Mi is a data preparation service for Australian businesses. We clean, match, map, transform and validate business data through customised, automated and repeatable processes — for system migrations, master data clean-up, recurring data preparation, consolidating acquisitions, and getting data ready for AI.
A contractor cleans the file. Data Mi writes down the rules, turns them into a process and hands that process to you, so the same work runs again next month without anybody rebuilding it. The output is a working process you own, not a tidier spreadsheet.
They are listed as exceptions and put in front of a person in your business. Nothing is silently guessed. That exception list is usually the most valuable thing produced in the first pass, because it shows precisely where the business disagrees with itself.
No. Data Mi work is processed in Australia, in Mattingly’s private cloud environment, and your data is not sent to external AI services. Transfers use encrypted methods including HTTPS and SFTP, and source files are deleted on schedule or on request.
Profiling usually takes days rather than weeks, and it is where most of the surprises appear. Building and trialling the process depends on how many rules the business needs to agree, which is a business question more than a technical one. Once it is built, running it again takes minutes.
Talk to us
Send us the file everyone complains about
We will tell you what is actually in it, what it would take to fix, and whether that is worth doing. No obligation either way.