The AI landscape doesn't move in one direction — it lurches. Some techniques leap from experiment to table stakes in a single quarter; others stall against regulatory walls, technical ceilings, or organisational inertia that no amount of hype can dislodge. Knowing which is which is the hard part. The State of Play cuts through the noise with a rigorously maintained index of AI techniques across every major business domain — classified by maturity, evidenced by real-world adoption, and updated daily so you always know where you stand relative to the field. Stop guessing. Start knowing.
A daily newsletter distilling the past two weeks of movement in a domain or two — delivered to your inbox while the index updates in the background.
Each dot marks the weighted maturity of practices within a domain — hover for a brief summary, click for more detail
AI that enriches CRM records, detects and merges duplicates, and maintains data hygiene across the sales tech stack. Includes automated field population and entity resolution; distinct from contact mapping which analyses relationships rather than cleaning data.
AI-powered CRM data enrichment is a solved technology problem with an unsolved organisational one. Every major CRM platform now ships automated deduplication, field population, and third-party enrichment as core features, and deployment frameworks document ROI in the range of 213-445% over three years. The tooling is proven and accessible. Yet 76% of CRM users still report that fewer than half their records are accurate or complete, and enterprises continue losing an estimated 20% of annual revenue to poor data quality. The binding constraint is not capability but governance: data decays roughly 2% per month, and without continuous validation workflows and cultural commitment to data discipline, even best-in-class enrichment tools degrade within a quarter. Organisations evaluating this space face a mature vendor ecosystem and clear business cases -- the question is whether they can sustain the operational rigour the tools demand.
CRM data enrichment has matured from technology frontier to operational requirement, with the constraint shifting decisively from capability to governance. Vendor ecosystem consolidation (HubSpot's Clearbit acquisition, ZoomInfo's market repricing) alongside internal repricing by AI-native competitors (Apollo, Clay) signals market transition. Waterfall enrichment (querying 4-7 sources sequentially) is now recognized as best practice, achieving 85-95% match rates versus single-source tools' 40-60% coverage and delivering measurable pipeline ROI: Anrok $300K+ in 90 days, Pylon 6,500+ contacts at 4.2X ROI, Together AI saving 30+ rep hours/month, Abacum reducing manual work 75%. Independent testing (CUFinder, 4,000-record HubSpot dataset) confirms tool maturity: CUFinder 96.0/100, Cognism 80.6, Clay 79.5, ZoomInfo 78.6 across weighted scoring. However, independent accuracy audits reveal systematic vendor claims overstatement of 10-15 percentage points: Cognism's cited 98% applies only to phone-verified subset (2.3% of database), with actual performance 62.5% on mobile/direct dials; ZoomInfo 85% actual vs claimed higher, Apollo 80%. This confidence gap is critical: OneAway's analysis of 2M+ enriched contacts shows waterfall delivering 43% higher connect rates and ICP accuracy improving from 61% to 89%, yet Gartner finds 40% of agentic AI CRM projects fail due to data quality (not AI capability), and 3x faster POC when data prioritized before AI deployment.
Adoption barriers remain fundamentally organizational rather than technical. Large-scale survey evidence (10,000 businesses, 32 countries, Q1/Q2 2026) shows 97% of organizations run AI but only 5% say their data is ready; 95% of GenAI pilots deliver zero business impact; 60% of AI projects are abandoned through 2026, with data quality cited as primary bottleneck. Within CRM teams, 76% report less than half their data is accurate or complete, 44% experience 10%+ annual revenue loss from poor data, and 40% of sales professionals still manually update CRM records. Data quality deficiencies cascade: poor CRM sync causes 62% of organizations to identify data accuracy as their primary AI adoption blocker (UserGems survey). Integration failures (missing bi-directional sync, field mapping gaps, stale data loops) cost organizations average $12.9M annually and consume 10-15 hours per week of RevOps manual reconciliation. A practical framework for data readiness exists: six dimensions with concrete targets (90%+ accuracy, 90%+ completeness, 95%+ timeliness, 98%+ validity, <2% duplication) across all enrichment approaches. Yet adoption of this framework remains sparse, with SMB CRM audits showing ~30% baseline duplicates and ~20% unreachable contacts.
Critical negative signals persist despite tool maturity. ZoomInfo's Q1 2026 guidance cut ($62M) and 600-person restructuring reveal traditional vendor economics under pressure from API-native competitors commoditizing databases at lower cost. Clearbit discontinuation forced 1,000+ non-HubSpot teams to migrate; Breeze Intelligence's credit-based pricing ($1,184-$4,135/month) and reported coverage gaps (30-40% of SMB/non-North American records) signal customer friction. Production incident reports document specific AI enrichment failure modes: hallucinated data sources (404 URLs), malformed records from incomplete migration data, integration drift when CRM fields change—requiring mandatory guardrails (retrieval grounding, citation validation, dead-letter queues, human-in-loop checkpoints) to operate reliably. AI-only approaches hit a structural ceiling: ZoomInfo's research confirms LLMs alone cannot reconcile conflicting data without proprietary validation layers; 56% of vendors embed AI in enrichment, yet 55% of companies adopting AI-powered data profiling find AI breaks down on context-dependent rules and compliance nuances, with 20% revenue loss persisting. Regulatory deadline pressure emerges as accelerant: EU AI Act (August 2026 main application) and Data Act (December 2026) create hard compliance requirements for data governance, shifting CRM data management from operational best practice to regulatory necessity. Organizations unable to implement continuous enrichment, cleanse-before-enrich sequencing, and audit data lineage will face regulatory exposure alongside operational drag. Data decay (2-3% monthly, 22-30% annually per $12.9M Gartner annual loss benchmark) remains non-negotiable driver of enrichment cadence; batch quarterly approaches obsolete in high-turnover sectors.
— HubSpot Q2 2026: Data Agent reached 16,000 customers (+80% QoQ), fastest-growing agent; Customer Agent 72% resolution rate; Prospecting Agent 17,000 activations; however, net customer addition missed guidance (7k vs 9-10k expected), revealing buyer budget constraints and extended sales cycles as adoption barriers despite strong feature momentum.
— Production deployment at scale: Grou deployed Clay across 23 client accounts processing 2.4M enrichments with 96% find rate and 1.4% bounce rate; churn analysis documents 4 accounts returning to Apollo, revealing credit consumption realities (4-8 credits per record vs 10K claimed) and ROI trade-offs by customer ACV tier.
— Named customer deployments document enrichment ROI: Redgate +42% BDR pipeline improvement post-Cognism, Mollie +80% connect rate gain from phone-verified data; Fujitsu reps save 4 hrs/week, Seismic team 11.5 hrs weekly from faster prospecting; demonstrates data quality directly drives sales productivity.
— Independent 500-record accuracy test across 10 B2B vendors (ZoomInfo, Cognism, Apollo, UpLead, Lusha, LeadIQ, LakeB2B, RocketReach, Lead411) over 7-day controlled period; Saleshandy 91.2%, ZoomInfo 84%, Cognism 91%; methodology highlights 22-70% annual data decay and email deliverability gaps causing early campaign failures.
— Autonomous CRM hygiene agents operating at scale: 12B Salesforce record analysis reveals 45% duplicates overall, 80% via API integrations; Validity 2026 data confirms 76% of orgs report <50% accurate/complete data with 37% revenue loss; continuous maintenance approach (not quarterly cleanup) now required; 5-15 day time-to-production documented for production deployments.
— Critical analysis with independent data: Salesforce Agentforce adoption stall at 34% (23k/150k customers), market cap loss $200B+; Fivetran survey (n=400): only 15% report data ready for agentic AI despite 60% investing, 42% cite data quality/lineage as primary barrier; establishes CRM data readiness as prerequisite, not AI capability gap.
— VentureBeat Pulse survey (n=101 enterprises, June 2026): 57% observed AI agents confidently delivering wrong answers due to CRM data issues; 78% of orgs running governed context layers report failure patterns vs 20% with no plans; documents AI reliability risk from source-data gaps as operational constraint.
— Synthesis of 2026 enrichment benchmarks: 25-30% annual B2B data decay; waterfall enrichment progression (1 provider 51% → 5 providers 66%, diminishing returns after 3-4); verified email 51% vs phone 30% coverage; documents multi-provider waterfall as industry best practice with economic optimization curve.