A6 · Evidence · 20 implementations
GTM Evidence Database
Documented agentic GTM implementations — with source trust scores and transferable lessons.
Source trust scoring
Every case in this database is scored for source trust — an explicit assessment of how much weight the evidence should carry. Trust is not a judgment on the company; it is a judgment on the verifiability of the claims.
| Trust level | Definition | Example sources | Cases |
|---|---|---|---|
| neutral | Independent analysis with no commercial stake in the outcome; methodology described; data verifiable | Academic studies, independent audits, RevOps team internal analyses shared at conferences, public SEC filings | 1 |
| self-reported | Data provided by the implementing company or vendor; no independent verification; results plausible but selection bias possible | Vendor case studies, customer testimonials, platform-reported metrics, conference presentations by practitioners | 12 |
| vendor-claimed | Claims made by a vendor about their own product's impact; methodology opaque; high risk of selection bias and survivorship bias | Press releases, product marketing pages, "X% improvement" without methodology, vendor-authored blog posts | 1 |
The distinction between self-reported and vendor-claimed is subtle but material: self-reported means the implementing company is reporting their own results (more credible than vendor claims about their own product); vendor-claimed means the vendor is reporting results for their customer (less credible — vendors select favorable cases).
All 20 implementation cases
Agentic email sequence personalization using company news + LinkedIn activity signals
Multi-persona sequence branching: agent detects job title, seniority, and prior engagement, routes each contac…
News-triggered outbound: agent monitors funding announcements, executive hires, and product launches for targe…
ML-based lead scoring model trained on 18 months of closed-won/lost data; scored inbound leads in real time an…
Composite PQL score combining product usage signals (feature activation, frequency, team size expansion) with …
G2 intent signal ingestion: agent processes weekly buyer intent exports, matches to ICP accounts in CRM, enric…
Technographic change monitoring: agent detects when target accounts add or remove competing technologies from …
Fully autonomous AI SDR: agent researches accounts, writes personalized emails, books meetings autonomously wi…
AI SDR as human amplifier: agent researches 50 accounts/day and drafts first-touch emails; human SDR reviews, …
ML pipeline forecast: trained on deal stage velocity, activity signals, stakeholder engagement, and historical…
Deal health scoring from activity signals: email open/reply velocity, meeting frequency, champion engagement, …
Dynamic ABM account selection: agent re-scores ICP fit weekly using 12 firmographic + technographic signals; a…
Account-specific content personalization: agent generates custom landing pages, email content, and ad copy usi…
Programmatic SEO at scale: agent generates state-by-state, industry-by-industry compliance guides (2,400+ page…
Autonomous CRM enrichment: agent runs nightly on all contacts and accounts, enriches missing fields (industry,…
Automated CRM hygiene: agent runs weekly, identifies duplicate records, stale contacts (no activity 180+ days)…
Conversation intelligence automation: agent analyzes all sales calls, identifies talk-track deviations, compet…
PQL scoring system: agent ingests product telemetry hourly, scores free users on expansion likelihood using 14…
Expansion PQL engine: agent monitors current customers for upsell signals (seat utilization >80%, feature gati…
Multi-channel ABM orchestration: agent coordinates paid ads (LinkedIn + 6sense), direct mail triggers, SDR out…
SRAL tier breakdown
| Tier | Level | What agents can own | Cases in this database |
|---|---|---|---|
| Tier 1 | Signal processing & execution | Intent scoring, enrichment, routing, behavioral optimization, monitoring | 14 |
| Tier 2 | Strategy & orchestration | Signal-to-motion decisions, sequenced outreach, CRM automation — with HITL checkpoints | 5 |
| Tier 3 | Full autonomy | Not yet viable — requires human authority at ICP, pricing, and relationship decisions | 1 |
See: C6 Agentic GTM hub →
Key findings across 20 cases
- Most production agentic GTM operates at Tier 1. Signal processing, lead scoring, enrichment, and behavioral personalization dominate — not autonomous strategy or outreach.
- Self-reported cases show 2–4× improvement claims; neutral-source cases show 20–50% improvements. The gap between vendor-claimed and independently measured results is substantial and consistent.
- HITL design is the key quality differentiator. Cases where human review is structurally embedded (not just theoretically available) consistently show better outcomes and fewer failure modes than fully autonomous implementations.
- Signal exhaustion is the most common failure mode. Automated outreach systems tend to saturate their TAM faster than human-paced SDR motion — often within 3–6 months of deployment.
- Timing beats copy in trigger-based outreach. A mediocre message within 2 hours of a relevant trigger outperforms a polished message 2 weeks later — documented across multiple cases.