How the ranking is scored

Scales

Each level describes the public evidence required. A score between two levels is allowed when the written justification explains what is present and what is missing. A dimension with no data in any public source is recorded as empty, counted as zero in the total, and listed on the score sheet.

How the data is collected

• One instruction and one source list for every company: the company site, Clutch, G2, GoodFirms, certification registries, published case studies.
• A value without a public source is left empty. Empty values are shown as ○ in tables.
• Companies with empty data in more than two dimensions are not ranked, whatever their size or reputation.
• Each score is written with a justification that cites the fields and sources it relies on. The justifications are published on the sources page.

Where these scales fall short

Four limits were found after the weights were frozen. The weights cannot be changed once company data exists, so the ranking was published on the methodology it was computed with, and the limits are published beside it rather than corrected out of sight.

• Security and delivery governance tops out at 70 for all twelve. Level 75 asks for a certificate number, an issuing body or a registry entry. Almost every company names an ISO standard; not one publishes a number for it. The only credential traceable to an issuing body's own register is a TMMi appraisal, which level 75 does not name. So the second-heaviest dimension measures how much a company discloses, not what it holds, and it is compressed into a 25 to 70 corridor.
• The pricing ladder has no rung for a company that publishes a model but no figure. Levels 25 and 50 assume a rate band on a directory profile. Two companies publish a pricing model and a floor — a minimum budget or a minimum engagement length — and no rate anywhere. That is formally below 25 and substantively above nothing; both were scored between the rungs, with the gap to the next level written out in the justification.
• Engagement flexibility separates the field by 30 points at a weight of 11. Every company scores between 60 and 90, because publishing several named ways to buy is common in this category. The dimension carries real weight but does little ordering work.
• The review ladder has no rung for a large review count on a single platform, and that gap was filled two different ways. Every level above 50 requires two platforms. Six of the twelve are on one. Two of those six are well past the volume the 50 level describes — one with 42 reviews, one with 66 — and neither fits the level above. They were scored 70 and 30. Each company is scored by itself against the written scale, with no sight of any other company's marks, which is what keeps a total from being fitted to a preferred order; the cost of that isolation shows here, where the same gap in the scale was read in opposite directions. Correcting the two marks after the totals were known would have meant setting a score with the ranking already in view, so they stand as they were given and the divergence is recorded here instead. What it costs is worth stating exactly. Score both at 50, the only level the written ladder puts them near, and the order of the twelve does not move; the order holds anywhere from about 39 to 64. Outside that it does move — near 30 the fourth and fifth places swap, near 70 the tenth through twelfth reorder. First place does not change at any value on the scale.

What the score is not

The score measures what a company documents publicly, not how a specific engagement will go. A company with a lower score can be the right choice for a buyer whose priorities differ from these weights; the ranking page names those cases in its scenario and fit sections.