Insights - "Did You Prepare That in Advance?" Why a Fast Demo Can Still Lose the Tender

In almost every demo, someone asks the same question: “Did you prepare that in advance?”

We didn’t. It runs that fast.

ECHA’s own video tutorial on the Drinking Water Directive shows an expert building a notification of intention in IUCLID: one alloy, two composition documents, intended use, validation, dossier and export. It takes about 35 minutes on camera, and the expert has the reference substances prepared beforehand. We built the same dossier through the IUCLID REST API in 2.4 seconds, and it passed ECHA’s validation rules for that dossier type (16 rules, no failures).

That is one measurement on one setup, a local IUCLID 10.0.3 in September 2026, and we say so when we show it. But the gap is large enough that the setup details don’t change the conclusion.

Why people don’t believe it

The explanations come quickly. “It’s cached.” “It’s a pre-staged setup.” “Our internal data is much messier than yours.”

They are reasonable doubts. Regulatory teams have seen a lot of polished demos, and most IUCLID work they know is manual: one composition row at a time, one reference substance at a time, then a validation report that sends them back to fix two fields. When a whole dossier appears in the time it takes to click “export”, a trick is the more plausible explanation.

That is why we offer a trial on the customer’s own data. A demo on our example proves the software works. A run on their dossiers proves it works for them.

Where speed stops counting

The people in the demo usually believe it by the end. A procurement scoring sheet is harder to convince.

Recently we scored 100% on the technical evaluation of a large chemical-industry tender for managed IUCLID hosting. The team liked what they saw. Half of the score was price, though, and our recurring fee was about twice the winner’s.

Here is roughly how that plays out in a 50/50 grid. Many tenders score price relative to the lowest bid, so a bid at twice the lowest price gets half the price points. We don’t know the exact formula this buyer used, so treat the numbers as an illustration:

A perfect technical score cannot close a 2× price gap. The sheet had no column for “hours returned to the regulatory team”, so those hours counted for nothing.

What we are changing

We are learning to fill in that column ourselves.

• Saved hours go into the offer as a number. “Faster” is prose and can’t be scored. “About 35 minutes per notification by an expert, under one minute through the API, times the number of notifications you plan next year” is a figure an evaluator can put next to a price.
• The trial goes on the price sheet as a line item. A trial mentioned in a cover letter is invisible to a scoring grid. A trial with a scope, a duration and a price of zero is something procurement can compare.
• Assurance is priced separately. SLA credits, a named lead, audit support: if they are baked into the headline rate, they only make the headline rate look high. As options, the buyer can choose them.

None of this makes us cheaper. It makes the difference visible in the place where the decision is taken.

What that column could look like

Below is the kind of table we are starting to attach to our offers. The left column uses ECHA’s own tutorial as the manual baseline. The volume row is a real customer’s figure; hourly cost, migration and validation results are added per customer after a trial on their data.

A rough example of the arithmetic: a team preparing 100 notifications a year at the expert’s pace spends about 58 hours building them by hand, before any corrections. A first-time applicant, who is slower on every composition row, spends considerably more. Multiply by the hourly cost, and the figure can sit on the price sheet next to our fee.

Where this could be heading

What follows is our reading of the market, not a forecast we can prove.

European chemical companies are under cost pressure, and consolidation and restructuring are back on the agenda. Regulatory IT sits in that budget. We see three ways it could go, and they are not exclusive.

Price grids get stricter. Procurement treats IUCLID hosting as infrastructure, like any other managed server, and the cheapest compliant bid wins. In that world the only question is whether a vendor clears the technical bar. Speed is a nice-to-have.

Integration work creates new demand. Mergers and reorganisations don’t merge regulatory data by themselves. Registrations move between legal entities, two IUCLID databases with different conventions have to become one, and duplicate registrations for the same substance have to be resolved. In that world, how fast and how safely data can be moved and rebuilt matters more than the monthly fee.

Buyers start scoring outcomes. If enough vendors put measurable numbers into their offers (hours per dossier, validation errors caught before submission, time from an ECHA release to an updated instance), tenders can start weighting them. That only happens if vendors supply numbers a buyer can check.

Our bet is on the third, and the way to get there is to publish numbers with their setup and let people rerun them on their own data. Which is why we answer the “did you prepare that in advance?” question with an offer to try it.

Want to see how your own dossiers behave through the API? Book a trial on your data. We host IUCLID in the EU with the REST API enabled; see IUCLID hosting.