ICC Profile Creation & Validation — Supplier Qualification Guide

What a Conforming COA Actually Has to Report — and What’s Usually Missing #

When we qualify a new colour management supplier or evaluate a profiling service for onboarding into our approved vendor list (AVL), the first document we request is the characterisation data COA, not a PDF of printed samples. A COA that says “average Delta E 1.8, pass” tells us almost nothing operationally useful. We need four fields minimum before we can gate this supplier through our QM-09 incoming profile acceptance checklist:

1. Average Delta E 2000 (dE00) across the full characterisation target (typically IT8.7/4 or ECI 2002 at 1617 patches minimum)
2. P95 Delta E 2000 — the 95th percentile value, which reveals tail behaviour the average always hides
3. Maximum single-patch Delta E 2000, with the patch coordinates identified
4. Tone reproduction curve (TRC) deviation for each channel (C, M, Y, K) at 25%, 50%, 75% values, reported against the reference characterisation data set

The P95 value is the one most suppliers omit. If a supplier sends a COA reporting average dE00 of 1.6 without a P95 figure, our default assumption is that P95 is above 3.5, which is consistent with mid-tier profiling services that do not iterate target measurement passes to eliminate spectrophotometric noise.

When a supplier cannot produce P95 data, that is a documentation problem — which usually reflects a profiling workflow that was not built for supplier qualification scenarios. We have seen this with three profiling services over the past two years, all of which produced visually acceptable output but could not support our auditable QMS records under our ISO 9001:2015-registered quality system.

Why Profiles Pass Internal Checks and Still Fail on Press — Root Causes #

This is the section most brand partners find counterintuitive, because they assume a validated profile is a validated result. The gap between profile validation and press conformance is where colour complaints actually originate.

The first failure mode is measurement condition mismatch. A profile characterised under M0 conditions (no UV filtering) will behave differently on substrates with optical brightening agents (OBAs) — which includes the majority of coated folding carton stocks at 300–350 GSM used in cosmetics and FMCG packaging. Under D50 illumination with UV contribution (M1), an OBA-treated substrate reflects 4–7% more blue-region energy than under M0. If the incoming profile was measured M0 and the press substrate is high-OBA coated board, the press operator is compensating for a phantom blue cast that the profile was never built to accommodate. The result is a consistent warm shift in neutrals, typically 2–3 dE00 in the grey balance patches, that no ink key adjustment can correct cleanly. We catch this during incoming inspection by re-measuring the supplier’s characterisation target on our X-Rite i1iO3+ under both M0 and M1, comparing their declared average against our re-measurement. A divergence of more than 0.8 dE00 between the two triggers a measurement condition query before the profile is accepted into our system.

The second failure mode is gamut boundary instability in the K-channel near shadow rolloff. Profiles built with insufficient patch density in the shadow region (below L* 20) often show K-channel TRC drift at 85–100% because the profiling software is interpolating rather than measuring through actual press behaviour in deep shadows. On a 175 lpi AM screen, shadow detail compression is already aggressive; a profile that under-samples that region will compound the problem. We specify a minimum of 1617 ECI 2002 Random v2 patches — not the 928-patch IT8.7/3 target — specifically because the ECI layout includes 90 shadow-region patches versus 45 in the older target. Suppliers who quote IT8.7/3 as their standard characterisation target for packaging work are using a tool designed for photographic proofing, not press characterisation on textured board.

The third mode is profile-to-press binding failure — using a profile characterised on one ink set and attempting to bind it to a different supplier’s ink formulation without re-linearisation. This is common when brands switch press suppliers mid-campaign but retain their existing ICC profiles to maintain colour consistency. The ink opacity and trap behaviour can shift the TRC by 3–6% at midtone even when the pigment CIE coordinates are nominally identical, because pigment load, varnish carrier and tack specification differ between ink manufacturers. Our incoming protocol flags any profile where the declared ink formulation does not match our current AVL ink supplier, and a re-characterisation is required before production release.

Does the Profiling Software Brand Matter for Packaging Qualification? #

For packaging work at commercial tolerance, the profiling engine matters less than the characterisation data quality and measurement hardware calibration. This is worth stating directly because suppliers sometimes lead with software brand as a quality signal.

The three engines most used in packaging colour management — GMG ColorServer, Heidelberg Prinect Color Toolbox, and X-Rite ColorThink Pro workflows — produce comparable results when working from the same well-measured characterisation data. Where they differ is in gamut mapping algorithm options and K-generation strategy for UCR/GCR, which are press-specific decisions, not software superiority. For qualification purposes, we care whether the supplier’s spectrophotometer is ISO 13655-certified, whether their measurement device calibration log is current within 6 months, and whether their profiling target patch count meets our 1617-patch minimum. The software brand appears nowhere in our QM-09 checklist.

Specification Notes for Brand Partners #

When you brief us on a new packaging project requiring ICC profile development or supplier qualification, the most useful document you can share upfront is your existing characterisation data and press condition declaration — specifically, which ISO 12647 press condition your current supplier is profiled against (FOGRA39, FOGRA51, CRPC6, or GRACoL 2013). Without that, we default to FOGRA51 for coated substrates, which is correct for European-standard OBA-treated boards but will diverge from legacy US-market GRACoL workflows.

The most common brief gap that causes unnecessary sample iterations is an undeclared substrate OBA level. Brands often specify the paper grade by name but do not flag whether the stock has been reformulated with higher OBA content since the last press run. A substrate OBA shift of even 2–3 FWA units can move grey balance by 1.5 dE00 under M1 conditions — enough to trigger a visual fail on a light booth comparison. Ask your current paper merchant for the current M0/M1 spectral reflectance data for your substrate before briefing any profiling supplier.

Our typical profiling and validation cycle for a new press condition runs 15–18 working days from substrate receipt to approved profile delivery, assuming one characterisation iteration. A second iteration, if the first press sheet falls outside our acceptance thresholds, adds 5–7 working days.

Frequently Asked Questions #

Do we need a new ICC profile if we switch from CMYK to CMYK + 1 spot colour printing?
Yes — adding a fifth channel fundamentally changes the ink trap behaviour and dot gain interaction on the substrate, so the existing 4-colour profile is no longer a valid characterisation of the press condition. A new expanded-gamut or spot-channel-aware profile is required.

Our current supplier says their profiles are “ISO certified” — what does that actually mean for qualification?
It depends on what they are certifying against. An ISO 12647-2 process certification confirms that their press is running within standard tolerances for density, dot gain and grey balance — but it does not certify any individual ICC profile file. A profile is a characterisation tool; the ISO 12647 certification is a press conformance certification. They are related but not interchangeable. Ask specifically for the date of the last press certification audit and the measurement instrument used, not just the certificate number.

Can we reuse an ICC profile across different packaging substrates to reduce profiling cost?
Substrate changes of more than 10 GSM, a switch between coated and uncoated finish, or any change in OBA treatment level all require re-characterisation — the cost of reusing an incorrect profile shows up in sample rejections and press makeready waste, not in profiling fees.

Planning a packaging project? Contact our team to request a complimentary specification review and sample quote.