Cohelion

In this industry, software alone does not survive without the experience behind it

Anyone can train a model. Knowing which delay code hides revenue leakage, what a de-icing shift does to cost per turnaround, and how a handling contract pays. That takes twenty years on the ramp. That's what you're buying.

20
years in aviation services
3
of the top 5 global networks
500+
stations live on one platform
95%
forecast accuracy, back-tested
Where we come from

It started with one question about cargo volumes

  1. 2003 Cargo Service Center, Badhoevedorp
  2. — Acquired by Swissport, rolled out network-wide
  3. — Ground handling, lounges, cleaning: 312 stations
  4. — WFS and Unilode
  5. Today SATS, Singapore

In 2003 Cargo Service Center (CSC) in Badhoevedorp contacted us with a straightforward-sounding request: “we need a solution that helps us consolidate our cargo operations across the six stations in our network. Oh, and every station runs a different ERP and cargo management system.” We took on the challenge, and we succeeded. Following that we added reporting and forecasting, and without quite knowing it we had created a master data management system that works well in the aviation sector, with all its semi-standards.

When Swissport acquired CSC, our platform was rolled out across their entire network. From there we integrated and combined their ground handling stations, lounge, cleaning operations and the rest, until we covered all 312 stations they operated globally. WFS and Unilode followed, and after Covid we signed SATS in Singapore.

The platform grew the way that history suggests: product by product, each one built with an operator, on real network data, for a decision somebody had to make. We are deliberately niche and deliberately specialist, so you will not find us selling to hospitals or supermarkets, and when you call, the person who answers has stood on an apron.

In the AI era that matters more rather than less. Models are cheap, but knowing what to ask them is not. Our forecasting reaches 95% accuracy because of two decades of experience in feature engineering around seasonality, contract structures, and disruption patterns.

How we work

Your own numbers first, a commitment second

No operator should have to sign a multi-year commitment to find out whether the numbers work. So phase one puts the full application in front of you, configured for your operation and running on your own data. You decide go or no-go against something your team has used. Only then do we connect it to your source systems and go live. We build on top of the software you already run, so there is no migration, no platform switch, and no SAP-style timeline or budget.

Phase 1 Six weeks ideal, three months typical

Proof of value

This is the full application, configured for your operation and running on your own data. The only thing missing is the live connection to your systems, so data is loaded manually rather than refreshed daily. You decide whether to go further against something your team can use.

Intake Configure the model Process your data Working application
What you get
  • The full application, not a cut-down demo
  • Insights across a selected timeframe
  • Your own data, loaded manually rather than through live interfaces
  • Hosted by Cohelion, nothing to provision
  • Ten or more dashboards your team can open
Phase 2 Two to five months

Full implementation, live

Only after a go decision. This is where the application connects to your source systems, so the numbers refresh daily instead of being loaded by hand. Around 80% of phase one carries straight over, so this extends the work rather than restarting it.

Intake Configure the model Integrate your applications Run and support
What you get
  • Insights refreshed daily, not on request
  • Automated ingestion from your business applications
  • Hosted in your own environment
  • The data model fully exposed to your team
  • New product features as they ship
Go / no-go The decision point between the two phases, taken on the deliverables in front of you. If it is a no, you stop having seen your own numbers. If it is a go, roughly 80% of phase one is re-usable.
Inside a phase

Each phase runs through the same four stages, with the lead time known before we start.

  1. Intake
    1-2 weeks

    Requirements, an inventory of your IT and business landscape, our data delivery specification, and a draft commercial proposal.

  2. Design
    1 month

    An on-site visit, analysis of your business and processes, the allocation mechanisms, and what you need on a dashboard.

  3. Build
    2-5 months

    Interfaces to your applications, the costing and reporting solution configured, data quality checks, and user testing.

  4. Run and support
    Ongoing

    Users and roles set up, business users trained, manual and data documentation handed over, with support if you want it.

The honest comparison

What you get from a specialist

Generic enterprise platforms can store your data. They don't know what a turnaround is. The difference shows up in three places: time to value, cost of change, and whether the numbers survive contact with your controllers.

  • Aviation-services data model out of the box Stations, turnarounds, SLAs, cost objects: predefined, not consulting deliverables.
  • Weeks to first value, months to full rollout Because we’ve done this at 500+ stations, the path is known. No discovery year.
  • A team that answers, not a ticket queue You work with the same specialists who built the platform, on a first-name basis.
  • Priced for a rollout, not a transformation program On top of your existing stack. The investment you’ve already made keeps earning.
Let's talk

Twenty years of answers, and your own in six weeks

Book a 30-minute conversation. We will show you Cohelion running on a dataset close to yours.