The pattern
One controlled drop point sits between you and the partner, and every file passes through it with an owner, a schedule and a log.
How it works
Your system writes the file to an outbound folder. The transfer service picks it up, encrypts it in transit, delivers it to the partner’s endpoint, and records the result. Inbound files land in a quarantine folder, get checked for shape and size, and only then move to the folder your system reads from. Failures alert the named owner, not a shared inbox.
Start with what you already license: a locked-down SharePoint site with external sharing, or the SFTP capability in your existing cloud or ERP platform, with logging turned on.
When it fits
The partner can only work in files. Volumes are batch, not real time. A few minutes or hours of delay is fine.
When it doesn’t
The partner offers an API and you need near real time, or the same data goes to many partners and you are copying files to each of them. Look at API management or event streaming instead.
NZ reality check
Most NZ organisations run this with one person who also does three other jobs, so the design has to survive that person going on leave.
- Scale: two or three partners do not need a platform. Twenty do.
- Privacy Act 2020: files with personal information going offshore need a cross-border disclosure check first.
- Critical infrastructure: location data for lines, substations, pipes and plant needs a deliberate decision about who sees what level of detail. Share what the requester needs for the job, not the whole network.
- Local estate: many NZ organisations already have SharePoint and an Azure or AWS subscription. Use them before buying a dedicated product.
- Cost: the main cost is staff time chasing failed files, not licences. Count that before comparing products.
Trade-offs and traps
The main cost is that files are always behind reality, so anything downstream works from a snapshot.
- New platform as reflex: buying managed file transfer to fix what is really an ownership gap.
- Familiar over durable: keeping the old SFTP box because one person knows it.
- Measuring activity, not outcomes: counting files sent instead of files the partner could actually use.
The saved-time test
Time saved chasing missing files should go back to the people who were chasing them. If nobody can say where it goes, the saving is on paper only.
Worked example
A South Island electricity distribution business shared asset data with a steady stream of outside parties: contractors planning work near the network, councils and developers scoping new subdivisions, and other utilities wanting to share trenches. Each needed to know what was in the ground and what they could work with.
Requests went to whoever in the GIS team picked them up. That person pulled an extract, saved it as a shapefile or spreadsheet, and emailed it or sent a shared link. Nobody could say later who held which version. Some extracts carried more detail about critical assets than the requester needed.
The first step cost nothing. The team agreed what each type of requester could see, named one owner for external data requests, and wrote a short set of terms covering use and refresh.
The second step used tools they already had. Standard extracts came out of the existing GIS platform on a schedule, landed in a controlled share with access per partner, and every download was logged.
Partners worked from a known version, routine requests no longer needed an analyst’s time, and a dedicated data-sharing platform was left until volumes justify it.