Birchy
Outsourcing & BPO
Research
What does a provider's rank position mean in this ranking?
The tiers mean something and the neighbouring positions mostly do not. Tier structure is stable: across 24 weight perturbations, no provider changes tier except one, and 10 hold their exact position. But 25 of 29 providers sit inside a cluster where adjacent scores are within 1.5 points of each other, and inside a cluster the order is not a quality signal.
This is a statement about this board, scored on this instrument. It is not a claim about rankings in general; no other ranking was examined.
What the instrument separates, and what it does not
A 100-point score has a resolution limit, and being honest about it is more useful than pretending otherwise.
It separates tiers reliably. Leader, Strong Performer, Specialist and Emerging are divided at published thresholds, and the sensitivity analysis moves 5 of 24 runs across a band — always the same provider, Alorica, which sits 0.2 points above the Strong Performer floor at rest.
It does not separate neighbours. No provider moves more than 2 positions under any perturbation, which sounds like stability and is also the problem: if a provider can move two places because a weight shifted 20%, then two providers two places apart are not reliably ordered by these scores.
The clusters
Two providers inside the same cluster are, on this evidence, indistinguishable. Only 4 providers stand alone: Teleperformance, HGS, Everise, Covisian.
Why 1.5 points
The figure is derived, not chosen. The rating bands this framework uses for its evidence-graded criteria are 1.5 points wide, so a gap narrower than one band cannot place two providers in different positions on any single criterion. A difference smaller than the instrument's own resolution is not a difference the instrument can see.
What to do with two providers in one cluster
Read the axis split, not the total. Two providers on the same score can reach it differently — one strong on Delivery Proof and thin on Capability Fit, the other the reverse — and that difference is real where the ranking gap is not. Each data sheet publishes both axes and the sub-criterion scores behind them.
Then read the gap line. Every provider record names its largest remaining gap, and for two providers a point apart the gaps are usually not the same gap. That is the comparison worth making.
What not to do is treat the higher position as the better provider. Where the scores are inside the window, the ranking is telling you they are level and the order is an artefact of having to print them in some sequence.
How this was established
Every score derives at build time from published sub-criterion anchors; nothing is transcribed. The clusters are computed from the scores rather than curated, so they cannot drift from the board. The sensitivity analysis re-runs the whole field 24 times, and its results are published in full on the ranking itself and in the methodology.
The tier thresholds, the anchors and the arithmetic are all published: the scoring anchors, and both scoring tables as CSV.