Index › HR › Compensation and total rewards › Compensation benchmarkin › Radford vs Willis Towers Watson
Compensation benchmarking data · November 2026 Edition
Radford vs Willis Towers Watson
Zero of fourteen models named Radford first on the direct prompt; zero named Willis Towers Watson. Radford was named by twelve of the fourteen models and Willis Towers Watson by eight and Radford carries 21 labels and Willis Towers Watson 13, so the shares are not directly comparable.
Radford
criticized challengerNamed in three categories this edition.
Willis Towers Watson
criticized challengerNamed in one category this edition.
First-choice share2%2%Of first choices across the direct, paraphrase, budget and scale prompts, 0 to 100.
Negative rate48%46%Negative labels as a share of the product's labels, 0 to 100.
Rank in category#7#8A position in a field of 9; printed, not drawn.
Labels2113A count; the two differ.
The two percentage rows are drawn on one 0 to 100 track, Radford reading right to left. Rank and label count are printed, not drawn.Pave was named alongside these two in fourteen of the fourteen direct answers. Payscale vs Radford · Payscale vs Willis Towers Watson · Pave vs Radford
Share is the count of first choices across the direct, paraphrase, budget and scale prompts over all fourteen models, for a mid-market B2B company; rank is within the category; every quote names the model and the prompt it came from. Both figures come from the compensation benchmarking data page.
By framing
How many of the fourteen models made each the first choice, per way of asking, and how many argued against it.
RadfordFirst choices, of fourteen modelsWillis Towers Watson
Direct003 against Radford · 1 against Willis Towers Watson
Paraphrase111 against Radford · 1 against Willis Towers Watson
Comparative11
Budget-constrained001 against Radford · 1 against Willis Towers Watson
Scale-constrained002 against Radford
Negative003 against Radford · 3 against Willis Towers Watson
Bars are first choices, 0 to 14 each sideModels that argued againstA model can name both, so the two sides of a row do not sum to fourteen.
Across every category in the November 2026 Edition, Radford and Willis Towers Watson were named in the same answer twenty times, of the 73 answers naming Radford and the 36 naming Willis Towers Watson. In those answers Willis Towers Watson took the first choice one time and Radford two.
Every model, every framing
The eighty-four answers behind the chart above, one cell each: where Radford and Willis Towers Watson stood in it.
| Model | DirectRAWT | ParaphraseRAWT | ComparativeRAWT | Budget-constrainedRAWT | Scale-constrainedRAWT | NegativeRAWT |
|---|---|---|---|---|---|---|
| Claude Haiku 4.5 | ||||||
| GPT-5.4 mini | RAWT | |||||
| Gemini 3.5 Flash | RA | RAWT | RA | WT | ||
| Perplexity Sonar | RAWT | |||||
| Grok 4.20 | RAWT | RAWT | RA | |||
| Mistral Small | ||||||
| DeepSeek V4 Flash | RA | RA | RAWT | |||
| Llama 4 Maverick | ||||||
| Qwen 3.7 Flash | RA | |||||
| Kimi K2 | RA | RA | WT | |||
| GLM 4.7 FlashX | RA | |||||
| MiniMax M2.5 | ||||||
| GPT-6 Luna | ||||||
| Muse Glimmer 30B | RAWT |
RA RadfordWT Willis Towers WatsonRA first choiceRA named as an alternativeRA argued againstblank: not namedEach cell is one answer, Radford on the left and Willis Towers Watson on the right.
The direct prompt
The plain question, one answer per model, grouped by where Radford and Willis Towers Watson stood in it.
Neither was the first choice, one was named
1 of 14 modelsThe answer put something else first and named one of the two as an alternative.Kimi K2Payscale alternatives: Mercer Total Remuneration Survey, Pave, Radford, Salary.com CompAnalyst
Neither was named
13 of 14 modelsThe answer made no first choice from these two in this category.Claude Haiku 4.5Pave, SalaryCube alternatives: CompUp, OpenComp, Payscale Ascent
GPT-5.4 miniSalary.com CompAnalyst alternatives: Compensation-IQ, Pave, Payscale Ascent
Gemini 3.5 FlashPave alternatives: Carta Total Compensation, Payscale, Salary.com CompAnalyst
Perplexity SonarPayscale alternatives: Pave, Ravio, SalaryCube
Grok 4.20Payscale alternatives: Pave, Ravio, Salary.com CompAnalyst
Mistral SmallPave alternatives: Mercer Total Remuneration Survey, Payscale
DeepSeek V4 FlashPayscale alternatives: Mercer Benchmark Database, Pave, Ravio, SalaryCube
Llama 4 MaverickSalaryCube alternatives: Mercer Total Remuneration Survey, Pave, Payscale
Qwen 3.7 FlashComprehensive.io alternatives: Figures, Pave, Payscale, Salary.com CompAnalyst
GLM 4.7 FlashXFigures, SalaryCube Comp Planning alternatives: Pave, Salary.com CompAnalyst
MiniMax M2.5CompAnalyst, Payscale alternatives: Pave
GPT-6 LunaPayscale alternatives: Pave
Muse Glimmer 30BPayscale alternatives: CompAnalyst, Comprehensive.io, Pave, Salary.com CompAnalyst, SalaryCube
Bold names in an answer are the products the judge labeled a first choice; a model naming several gives each of them that label. The full answer text for every row is in the record.
By buyer segment
The same question asked on behalf of a different buyer. Each standing is computed within its segment and they are never added together. The figures above are the mid-market standing, which is the one the category orders by.
Small business
Level: the same share of first choices.
Radford0%#10 of 10
Willis Towers Watson0%#– of 10
The full small business standing →Mid-marketThe figures above
Level: the same share of first choices.
Radford2%#7 of 9
Willis Towers Watson2%#8 of 9
The full mid-market standing →Enterprise
Level: the same share of first choices.
Radford5%#4 of 9
Willis Towers Watson5%#6 of 9
The full enterprise standing →What the models said about Radford
Every negative label with a quote, up to three, then the highest-weighted positives, up to three. Six of seven in this category shown.
“Providers to avoid or use carefully include: - Mercer, Willis Towers Watson (WTW), and Radford when the budget is tight” Muse Glimmer 30B · negative prompt · hard negative
“Avoid for Mid-Market: Radford (Aon), Mercer, or Willis Towers Watson” Grok 4.20 · direct prompt · hard negative
“Large vendors (like Radford) rely on hundreds of massive companies to build benchmarks... avoid the heavy enterprise platforms unless you have a dedicated Compensation Manager” Qwen 3.7 Flash · scale prompt · soft negative
“Radford provides deep traditional global enterprise data; Pave offers real-time API integrations” Gemini 3.5 Flash · comparative prompt · first choice
“It is widely considered the gold standard for tech and SaaS.” Gemini 3.5 Flash · paraphrase prompt · first choice
“Radford excels in private company data (including VC/PE-backed firms) and has massive participation” Grok 4.20 · paraphrase prompt · alternative
What the models said about Willis Towers Watson
Every negative label with a quote, up to three, then the highest-weighted positives, up to three. Six of seven in this category shown.
“explicitly advise skipping these providers unless you have a significant budget and need deep integration capabilities” Muse Glimmer 30B · negative prompt · hard negative
“avoid traditional enterprise survey providers (like Radford/Aon, Mercer, or Willis Towers Watson” Gemini 3.5 Flash · budget prompt · hard negative
“Avoid for Mid-Market: Radford (Aon), Mercer, or Willis Towers Watson” Grok 4.20 · direct prompt · hard negative
“Need HR, compensation, or talent data? | Mercer, Korn Ferry, Aon/WTW” Kimi K2 · comparative prompt · first choice
“Best all-around: Mercer or WTW” GPT-5.4 mini · paraphrase prompt · first choice
“Highly trusted by board compensation committees and HR executives worldwide.” Gemini 3.5 Flash · paraphrase prompt · alternative
Also compared
Payscale vs RadfordPayscale vs Willis Towers WatsonPave vs RadfordPave vs Willis Towers WatsonComprehensive.io vs RadfordComprehensive.io vs Willis Towers WatsonSalary.com CompAnalyst vs RadfordSalary.com CompAnalyst vs Willis Towers WatsonGlassdoor vs RadfordGlassdoor vs Willis Towers WatsonMercer Total Remuneration Survey vs RadfordMercer Total Remuneration Survey vs Willis Towers WatsonAll twenty-eight head-to-head pages and the full compensation benchmarkin standing
Comparisons are drawn for the top eight products in each category, each against each. The output is the models' output; nothing here is a recommendation by the index.