# Recast vs HockeyStack: which do AI models recommend for Attribution & MMM, September 2026

GTM AI Recommendation Index, September 2026 Edition, Attribution and marketing mix modeling. Two of twelve models named Recast first on the direct prompt; one named HockeyStack. Page: https://gtm-ai-index.com/marketing/attribution-mmm/recast-vs-hockeystack/

| | First-choice share | Rank | Negative rate | Labels | Models naming it |
|---|---|---|---|---|---|
| Recast | 9% | #3 of 22 | 9% | 35 | 10 of 12 |
| HockeyStack | 3% | #6 of 22 | 10% | 29 | 11 of 12 |

## The direct prompt, model by model

- Mistral Small: recast first (first choices: Dreamdata, Recast) (alternatives: Cometly, HockeyStack, Improvado, Measured, MixModeler, Ruler Analytics)
- MiniMax M2.5: recast first (first choices: Dreamdata, Recast, SegmentStream) (alternatives: Analytic Edge, HubSpot Marketing Hub)
- DeepSeek V4 Flash: hockeystack first (first choices: Dreamdata, HockeyStack, Rockerbox) (alternatives: Improvado, Recast, Ruler Analytics, SegmentStream, Sellforte)
- Gemini 3.5 Flash: neither first, one named (first choices: CaliberMind) (alternatives: Dreamdata, HockeyStack, RevSure)
- Perplexity Sonar: neither first, one named (first choices: Dreamdata) (alternatives: Factors.ai, HockeyStack)
- Grok 4.1 Fast: neither first, one named (first choices: Dreamdata) (alternatives: HockeyStack, Recast, RevSure)
- Qwen 3.7 Flash: neither first, one named (first choices: Rockerbox) (alternatives: Alviss AI, Recast)
- Kimi K2: neither first, one named (first choices: Dreamdata) (alternatives: Factors.ai, HockeyStack, Recast, Rockerbox)
- GLM 4.7 FlashX: neither first, one named (first choices: Dreamdata, SegmentStream) (alternatives: Heeet, HockeyStack, Lifesight, Measured, MixModeler, Northbeam, Recast, Ruler Analytics)
- Claude Haiku 4.5: neither named (first choices: Dreamdata) (alternatives: CaliberMind, HubSpot Marketing Hub, Northbeam)
- GPT-5.4 mini: neither named (first choices: Dreamdata) (alternatives: Adobe Marketo Measure, CaliberMind, MixModeler)
- Llama 4 Maverick: neither named (first choices: Improvado) (alternatives: Dreamdata, Rockerbox)

## What the models said about Recast

- "Far too expensive for limited budgets" (DeepSeek V4 Flash, budget prompt, hard negative)
- "Built for Statisticians, Not Marketers ... Requires a data scientist to translate Bayesian posteriors" (Kimi K2, negative prompt, soft negative)
- "Recast/Funnel are solid but more digital-heavy/enterprise (~$50K+)" (Grok 4.1 Fast, paraphrase prompt, soft negative)
- "You likely want to focus on Tier 2 (SaaS-Enabled UMM): Recast, Measured, Lifesight, Mutinex, LiftLab, Haus" (Gemini 3.5 Flash, scale prompt, first choice)
- "Recast and Measured being more suitable for mid-market digital teams and enterprises, respectively" (Llama 4 Maverick, scale prompt, first choice)
- "Recast is the standout — transparent Bayesian models, weekly refreshes, and a fast setup" (DeepSeek V4 Flash, paraphrase prompt, first choice)

## What the models said about HockeyStack

- "Verdict: Avoid for any situation requiring accountability or regulatory compliance" (Mistral Small, negative prompt, hard negative)
- "Multi-Touch Tools (e.g., Triple Whale, Northbeam, Rockerbox, HockeyStack): Frequent "trust issues"" (Grok 4.1 Fast, negative prompt, soft negative)
- "Attribution isn't the core product ... so attribution depth is limited." (DeepSeek V4 Flash, negative prompt, soft negative)
- "Dreamdata or HockeyStack (attribution) will deliver more practical value faster" (DeepSeek V4 Flash, direct prompt, first choice)
- "is Dreamdata, HockeyStack, or Factors.ai, according to the search results" (Llama 4 Maverick, paraphrase prompt, first choice)
- "Alternatives if needed: HockeyStack (~$25K-$30K/year, stronger on AI/predictive GTM intelligence but pricier" (Grok 4.1 Fast, paraphrase prompt, alternative)

Share is the count of first choices across the direct, paraphrase, budget and scale prompts over all twelve models, for a mid-market B2B company; rank is within the category. Comparisons are drawn for the top eight products in each category. Published under CC BY 4.0; the output is the models' output, and nothing here is a recommendation by the index.