The State of Play

Scripted chatbots are a mature, proven automation tool for narrow customer support tasks -- and a technology whose growth cycle has ended. Rule-based and decision-tree systems that route queries through predefined flows reached good-practice status around 2021, with GA tooling from Zendesk, Dydu, and Zoho, documented ROI across verticals, and widespread enterprise deployment. The economics are settled: deflection rates of 40--70% on Tier 1 volume at a fraction of human agent cost make the business case straightforward for FAQ handling, order status, and returns. Banking alone has reached 88--92% adoption among major institutions. The defining tension is not whether scripted bots work -- they do, reliably, within well-scoped boundaries -- but whether they remain strategically relevant. LLM-powered alternatives now resolve broader query types at higher rates, and vendors including Zendesk have shifted investment toward AI agents. Scripted systems persist where predictability, compliance, and low cost matter most, but the ceiling is explicit: rigid flows, exponential maintenance costs, and inability to learn constrain further expansion. This is a practice to operate and optimise, not to bet on for growth.

Vendor support continues but strategic emphasis has shifted. Zendesk officially migrated from legacy Answer Bot to AI agents as its platform standard in mid-2025, though it still maintains Answer Bot metrics dashboards for existing deployments. Dydu secured EUR 6.3M in early 2026 to scale its rule-based and conversational AI systems, sustaining 160-plus enterprise projects across energy, finance, telecom, and public sector clients including EDF, SNCF, and Orange. Zoho maintains its SalesIQ Answer Bot for small and mid-market buyers. The installed base is large and operationally active, even as new investment flows elsewhere.

Deployment economics remain compelling within defined scope. Industry benchmarks put scripted interactions at $0.10--$0.50 each versus $5--$12 for human agents, with small businesses documenting 148--200% ROI over 12--18 months. A major French hypermarket managing 25M-plus annual calls achieved an 18% reduction in handling time through Zendesk conversational bots in production. These are real, sustained savings -- but they accrue only in high-volume, low-complexity workflows. Customer satisfaction tells the other side of the story: only 29% of banking chatbot users report satisfaction despite a 70% return rate, and 35% of small business chatbot projects fail outright due to implementation complexity. The gap between deployment and satisfaction remains the practice's persistent weakness.

The competitive case against scripted systems is now explicit. Industry analysis documents 30--40% resolution rates for rule-based bots versus 80-plus percent for conversational AI, with 78% of scripted interactions requiring human escalation. Vendors and analysts frame rule-based architectures as fundamentally constrained -- unable to learn, adapt, or handle off-script inputs -- rather than incrementally improvable. For organisations already running scripted bots in well-scoped niches, the ROI justifies continued operation. For new deployments, the strategic calculus increasingly favours LLM-powered alternatives.

• 2016: Chatbot hype reached peak with Zendesk's Automatic Answers GA, Microsoft Tay experiment, Facebook Messenger bots, and 80% enterprise adoption intent, offset by critical analysis of UX limitations and failure cases in real deployments.

• 2017: Zendesk Answer Bot GA (Aug) marked ecosystem maturity with production deployments at Dollar Shave Club (4.5K tickets/month) and GogglesNMore (70% deflation). Facebook's autonomous chatbots failed at 70% of requests, reinforcing viability of scripted approaches. Retail sector saw high abandonment rates (Everlane, 73% consumer intolerance) but successful deployments demonstrated 25–70% deflation when scoped tightly. Shift from experimental to early-production phase with clear use-case boundaries.

• 2018: Vendor ecosystem solidified with 14+ commercial chatbot solutions supporting scripted workflows. Dollar Shave Club sustained Answer Bot deployment, scaling to 9,000 monthly resolutions (14% rate). Consumer adoption reached 15% of US adults, but implementation success hinged on organizational readiness (knowledge base maturity, existing automations) rather than technology capability. Scripted bots remained confined to narrow use cases (FAQ, order status); broader expansion awaited more sophisticated AI approaches. Early-production phase continued with concentrated early-adopter base.

• 2019: Zendesk Answer Bot expanded to all web and mobile channels; customers collectively solved 1M+ tickets and saved 225K agent hours. Enterprise intent surged: 70% of organizations without bots planned deployment within 12 months. Major banking client handled 300K questions daily. Implementation challenges surfaced: Forrester predicted 60% of deployments would lack effective human handoff; 50.7% of users cited inability to reach live agents as top frustration. Scripted bots proved viable at scale for high-volume, low-complexity tasks but implementation success remained dependent on organizational process maturity rather than product capability.

• 2020: Public sector adoption validated scripted chatbot viability: UK Government Digital Service published guidance with DVLA and MoJ case studies showing measurable contact reduction. International commercial deployments achieved at scale—Väre (Finland) reached 63% automation rate; Garden Games processed 15K+ emails annually with 40%+ intent accuracy. Proof points reinforced core thesis: scripted bots excel at narrow, high-volume transactional tasks when properly scoped, but organizational readiness and human escalation design remained the primary adoption constraint.

• 2021: Market adoption reached critical mass with 65% of 2,800 European companies deploying chatbots. Harmonie Mutuelle case study demonstrated production viability at scale (500 dialogues/month, 80% comprehension). Zendesk continued product evolution with language model improvements and 17-language support, driven by 95% COVID-era request surge. Peer-reviewed research revealed design trade-offs: humanlike chatbots reduced satisfaction for angry customers when expectations were unmet. Scripted bots consolidated as proven technology with clear organizational prerequisites (data protection compliance, clean data structures, effective escalation), but fundamental limitations (no learning, no personalization) persisted. The tier remained leading-edge—adoption was now mainstream for well-scoped use cases, with clear understanding of success factors and constraints, but broader scaling awaited either organizational maturity improvements or algorithmic advances.

• 2022-H1: Vendor platforms continued maturation with Zendesk transitioning from Answer Bot usage metrics to Automated Resolution pricing and integrating Flow Builder for low-code workflow automation. Consumer adoption remained strong (82% usage, 70% satisfaction for basic queries) but user experience limitations persisted: independent survey found 32% rarely felt understood, 30.8% abandoned interactions, and 60%+ cited frustration at chatbot handoff. Production deployments confirmed enduring value in narrow, high-volume use cases (Agence Française de Développement's HR chatbots), while peer-reviewed research documented the core constraint—inability to handle unanticipated inputs. Tier transitioned to good-practice as adoption became normalized, technology accessible to business users, and success factors well understood, yet algorithmic constraints remained.

• 2022-H2: Market adoption reached mainstream scale: 54% of enterprises used chatbots for customer interactions, with 38% planning deployment. Production deployments validated ROI—Crosscard/Viabuy's Zendesk bot resolved 10K+ questions annually, SNCF handled 10M+ questions with eight languages, PwC achieved 97% comprehension on internal support. Market forecasts confirmed ecosystem maturity: $10.5B projected by 2026 from $2.9B in 2020 (23.5% CAGR). However, persistent user experience barriers remained: Zendesk research showed 60% customer disappointment, 55% reported inaccurate responses, and 50% frustrated with bots' inability to recognize limitations. Scripted bots continued to excel at narrow, high-volume use cases but broader adoption awaited organizational maturity or algorithmic advances.

• 2023-H1: Multi-industry production deployments confirmed via Dydu (BNP Paribas, Total Energies, DSAC aviation, French prefecture) and hybrid adoption in regulated industries (Switzerland study). Enterprise deployment held at ~19% with 62% planning. However, customer adoption remained limited: Gartner survey found only 8% used chatbot recently and 25% would repeat; Capterra retail study showed 50%+ negative experiences and low conversion (17% product search). Early ChatGPT disruption signals emerged. Adoption growth plateaued despite awareness; tier remained good-practice as organizational process readiness remained the binding constraint.

• 2023-H2: Enterprise adoption intent surged (71% of executives aimed for full automation by 2027) but customer-facing deployment stalled—50% of B2C support managers cited perception of chatbots as "cold and static," while industry experts documented high cost and low customer satisfaction; deployment shifted toward internal use cases (HR, IT). Dydu Observatory showed 92% consideration among professionals (up from 48% in 2021), but ChatGPT awareness reached 94% with 34% using LLMs, signaling competitive displacement. Zendesk Answer Bot remained technically limited—NLP mapping with tunable thresholds but no learning capability. Critical tension: widespread enterprise intent clashed with documented customer dissatisfaction, implementation complexity, data privacy concerns, and emerging LLM-powered alternatives. Tier remained good-practice; adoption growth plateaued as organizational readiness and scope boundaries remained the primary adoption constraint.

• 2024-Q1: Production deployments continued across internal support workflows: Agence Française de Développement scaled Dydu HR and recruitment chatbots to 3,000 employees with 90%+ qualified interaction rate, while Dydu added nine new enterprise deployments (tax advisory, energy, public finance sectors) by March. Research validation affirmed technical maturity: peer-reviewed hybrid rule-based system achieved 98% accuracy and 97% precision on 1,684 e-business queries, with response time optimization (41 seconds vs. 20 minutes). Market metrics held steady at 92% enterprise adoption consideration, with 8/10 satisfaction but "reliability of responses" persisting as primary barrier. Vendor analysis showed 300-400% average ROI with 65% ticket automation in successful deployments. Critical barriers remained: documented limitations (incorrect responses, inability to handle complex queries, empathy deficits, security concerns) continued to constrain customer-facing expansion. Tier remained good-practice; LLM disruption signals intensified with ChatGPT adoption reaching 34% among professionals, pressuring scripted systems from above while organizational process maturity continued limiting broader adoption from below.

• 2024-Q2: Vendor platforms continued incremental improvements (Zoho's Answer Bot 2.0) alongside sustained enterprise adoption patterns. Market adoption metrics held steady (74% FAQ preference, 39% B2C chatbot involvement), but growth had stalled; customer satisfaction barriers remained (74% found chatbots useful for specific tasks but uptake constrained by perceived limitations). Real-world deployment challenges persisted, with platform configuration and integration complexity documented as practical barriers to expansion. Competitive displacement from large language models accelerated, shifting strategic focus from rule-based expansion toward integration choices and narrowly-scoped use-case optimization. Tier remained good-practice as core value proposition (high-volume, low-complexity automation) remained proven and accessible to business users, yet economic and competitive pressures intensified.

• 2024-Q3: Production deployments validated scripted chatbot viability in specialized domains: Mass General Brigham's rule-based return-to-work chatbot handled 5,575 users with 71.6% meeting criteria, reducing daily OHS calls from 633 to 115 (82% reduction) with wait times dropping from 28 to 6 minutes—demonstrating operational impact in healthcare. Peer-reviewed evidence confirmed persistent maturity constraints: healthcare chatbots offered delivery and administrative benefits but faced ethical, technical, and UX limitations; consumer adoption remained limited (only 30.1% likely to use health chatbots despite interest in voice-based support). Vendor platform evolution accelerated with Zendesk repositioning knowledge base chatbots toward AI agent integration (promoted as "next evolution"), signaling incumbent shift toward LLM-powered alternatives. Documented chatbot failures (DPD poetry incident, Chevy price manipulation) highlighted deployment risks from inadequate guardrails and human oversight gaps. Tier remained good-practice—scripted bots delivered proven value in narrow, well-scoped use cases with measurable operational savings—but the competitive landscape shifted decisively toward LLM-powered approaches, constraining further growth of rule-based systems.

• 2024-Q4: Scripted chatbot adoption remained stable in internal support workflows with evidence of sustained production deployments: Renault's IT helpdesk chatbot (deployed 2021, Teams integration Oct 2023) achieved 72% traffic increase and 94% comprehension with 137K visitors, demonstrating long-term viability of rule-based automation in employee support. However, market headwinds persisted: Zendesk shifted strategic messaging to AI agents (Oct-Nov 2024), positioning them as next-generation chatbots with 80%+ autonomous resolution; practitioners documented continued customer frustration barriers (redirected to knowledge bases without solving problems) and ROI measurement challenges limiting expansion. Critical insight: scripted bots' role shifted decisively from customer-facing growth engine to stable, well-scoped internal automation tool. The tier remained good-practice—proven, accessible, economically sound in narrow use cases—but growth ceiling was explicit as vendor innovation and market attention flowed toward LLM-powered alternatives capable of broader scope.

• 2025-Q1: Scripted chatbots established a stable niche in 2025 with quantified economics and defined use-case boundaries. Market research confirmed rule-based systems holding ~35% of the chatbot solutions market due to cost-effectiveness and ease of deployment, handling 65% of routine inquiries in utilities and retail. Zendesk maintained support for legacy automated article response features with ongoing analytics dashboards. Industry benchmarks quantified the economic case: setup costs $3k–$20k, development $3k–$7k, deflation rates 20–40%, with operational savings of up to 40% AHT reduction. E-commerce adoption metrics showed sustained customer acceptance (31% for order management, 28% for returns). However, critical limitations persisted and were well-documented: ML6 consultancy analysis highlighted scripted systems' manual maintenance requirements, rigid flows, and inability to extract context from prior interactions, contrasting them with GenAI approaches. Independently, 75% of chatbot users reported failures on complex issues, with root causes including training data limits, classification failures, and lack of reasoning capability. The strategic picture remained clear: scripted chatbots delivered proven ROI in narrow, well-defined use cases (FAQ, order status, returns) but faced ceiling for broader adoption. Tier remained good-practice as the technology proved accessible, economically sound, and operationally viable in defined scopes, with clear understanding of both strengths (cost, predictability, compliance) and constraints (rigidity, context limitations, complexity handling).

• 2025-Q2: Scripted chatbot market remained stable with sustained enterprise deployments alongside strategic vendor shifts toward LLM-powered alternatives. Zendesk officially migrated from legacy Answer Bot to AI agents as the platform standard (June 2025), signaling strategic deprecation of rule-based scripted approaches within major incumbents. However, real-world deployments continued: a major French hypermarket managing 25M+ annual calls achieved 18% reduction in handling time and 17% agent productivity gains through Zendesk conversational bots in production. Market context showed continued industry adoption (58% B2B, 42% B2C companies deploying chatbots) with broader market growth ($15.57B in 2024 projected to reach $46.6B by 2029), but customer-side barriers persisted—50% of customers reported frustration, and 70% of consumers signaled willingness to switch brands after bad chatbot experiences. Scripted bots remained economically viable in defined niches, but competitive pressure from GenAI approaches and persistent customer satisfaction gaps continued limiting expansion beyond narrow, high-volume transactional use cases. Tier remained good-practice as the value proposition (cost-effective, predictable automation in narrow scopes) held steady, yet the strategic landscape shifted decisively as vendors repositioned toward broader AI agents.

• 2025-Q3: Scripted chatbot deployments remained operationally viable but faced mounting evidence of fundamental limitations and competitive displacement by LLM-powered alternatives. Zendesk continued supporting Answer Bot with updated metrics documentation (Sept 2025), including suggestion rate, resolution rate, and ticket-assist measures for production monitoring. However, peer-reviewed research affirmed persistent technical constraints: rule-based systems provide low cost and operational transparency but fall short in scalability and flexibility—core findings from Aug 2025 comparative analysis of chatbot development methods. Critical evidence emerged documenting real-world failure modes in production deployments: leading chatbots delivered inaccurate or misleading answers 27% of time on complex queries; 67% of users abandoned chatbots stuck in instruction loops or generating irrelevant responses; one major UK bank's scripted chatbot misinformed over 140,000 customers about overdraft policies, resulting in lost customers and regulatory scrutiny. Infrastructure failures imposed quantifiable costs: retailers documented $4.2M annual revenue loss from chatbot failures with no human escalation path. Hybrid deployment models gained traction as pragmatic mitigation: analysis of 2025 implementations showed organizations increasingly adopting human-bot collaboration (bots handling 24/7 routine inquiries, agents managing complex problem-solving) with implementation costs of $2k–$150k, though integration complexity remained a persistent adoption barrier. Market analytics revealed ROI measurement challenges: companies focusing solely on activity metrics (interactions, messages) achieved 60% lower ROI than those measuring business outcomes; a Fortune 500 retailer with 2M monthly chatbot interactions experienced 25% customer churn due to poor resolution rates. Tier remained good-practice as scripted bots delivered proven value in narrowly-scoped, high-volume transactional workflows (FAQ handling, order status, returns), but the empirical case for broader expansion had weakened substantially—critical limitations on accuracy, context awareness, and escalation transparency now documenting the ceiling of rule-based automation viability.

• 2025-Q4: Scripted chatbot market approached closure of growth cycle with consolidating adoption evidence and explicit technical constraint documentation. Consumer adoption reached mainstream scale: Pew Research (Dec 2025) survey of 1,458 U.S. teens found 64% use chatbots, including 30% daily—demonstrating broad acceptance across demographics. Banking sector adoption reached 92% of North American banks with $2B+ market value, though satisfaction gaps persisted (29% satisfaction despite 70% return rate), revealing persistent implementation-execution challenges. Small business ROI remained proven (148-200% returns over 12-18 months) but adoption barriers persisted: 35% of projects failed due to implementation mistakes. Critical technical analysis surfaced in Q4: AIQ Labs documented fundamental rule-based limitations—requirement for hundreds of thousands of hand-tuned rules yet inability to learn or adapt, with maintenance costs scaling exponentially. NICE projection held at 70% of interactions managed by AI agents by 2025. Industry analysis emphasized that isolation, governance gaps, and accuracy failures prevented broader expansion. By year-end, the strategic picture remained stable: scripted bots maintained proven niche value in narrow, transactional use cases (FAQ, order status, returns) with clear ROI and compliance advantages, but technological ceiling was explicit and well-documented. Tier remained good-practice as adoption plateaued within defined scope boundaries; further growth awaited either organizational process maturity improvements or fundamental algorithmic advances beyond rule-based paradigm.

• 2026-Feb: Scripted chatbot market consolidated around stable, narrowly-scoped deployment patterns with explicit competitive displacement signals from AI-powered alternatives. Vendor support remained: Dydu announced €6.3M funding in Feb 2026 to scale rule-based and conversational AI systems, with 160+ enterprise projects handling millions of monthly conversations across verticals (energy, finance, telecom, public sector). Production-scale deployments validated continued viability in defined use cases—banking sector achieved 88-92% Tier 1 automation with scripted systems, processing up to 90% of routine interactions. Market economics held steady: 2026 benchmarks confirmed $0.10-$0.50 per-interaction cost for scripted systems deflating 40-70% of Tier 1 volume, with ROI remaining strong ($8 return per $1 invested in some verticals). However, critical competitive pressure from LLM-powered alternatives intensified: industry analysis declared rule-based chatbots obsolete due to fundamental limitations (inability to handle off-script inputs, 30-40% resolution vs. 80%+ for conversational AI, rigid decision trees, high escalation rates of 78%). Customer satisfaction remained constrained (29% reported satisfaction despite high return rates), highlighting persistent implementation-execution gaps. The tier remained good-practice—scripted bots delivered proven, economically sound automation in narrow, high-volume transactional tasks—but the strategic position continued narrowing as vendor innovation and market attention flowed decisively toward LLM-powered autonomous and agentic systems capable of broader scope.

• 2026-Apr: Scripted chatbot market faced explicit vendor deprecation and competitive obsolescence signals. Zendesk announced sunsetting of legacy scripted bot builder by August 2026, consolidating all bot functionality into advanced AI agents—signaling the major incumbent's formal exit from rule-based approaches. Industry research (Gartner, Forrester) confirmed: 70% of enterprises have deployed some form of automated conversational system, yet 54% of consumers report chatbot interactions as frustrating due to inability to handle complex semantics, rigid scripting, and context loss. Independent study of 50+ mid-market companies (travel, mobility, events) showed 50% adoption with mixed outcomes: 40% achieved response time improvements, but cost reductions and resolution rate gains remained limited, suggesting organizations are "still in early stages of learning curve." Technical analysis documented core limitations driving organizational abandonment: rigid linear conversation flows that crash on out-of-sequence inputs; poor NLU causing high escalation; and lack of learning capability preventing improvement over time. Real-world practitioner finding: organizations recognize scripted bots "work, but don't scale"—solving specific FAQ problems but failing to improve autonomously or adapt to new conditions. Banking sector maintained high automation rates (88-92%) in narrow Tier 1 tasks, validating continued economic viability in well-scoped, high-volume transactional workflows. However, the strategic ceiling was explicit: no major vendor invests in rule-based architectures as primary platform; all incumbent innovation and market messaging has shifted to agentic AI. The tier remained good-practice due to proven, accessible, economically sound operation in defined scopes, but further expansion of scripted systems was constrained by documented technical limitations and explicit vendor deprecation signals.

• 2026-May (Apr-May scan): Late April and early May 2026 evidence consolidated the narrative of scripted chatbot obsolescence and implementation challenges. UC Berkeley research (California Management Review) documented five frustration sources specific to rule-based systems: understanding failures, inability to solve complex problems, poor handover integration, lack of humanization, and lack of personalization; Gartner survey showed 64% customer preference against AI chatbots entirely. TIMEWELL analysis of major 2026 implementations across Klarna, Intercom, and others concluded: "Almost no companies are getting results from a chatbot in isolation"—Klarna's 2.3M conversations/month shifted to hybrid model in 2025 due to quality erosion. Oxford peer-reviewed research found chatbots trained for warmth paradoxically made 10-30% more errors on critical topics (medical, false-belief correction), highlighting design constraints. Practitioner benchmarks remained narrow: ecommerce deployments cost $0-29/mo and achieve 5-15% conversion lift with 20-30% deflection, but only for stores under $500K annual revenue with stable catalog. Real-world deployment failures documented: automotive dealership case study across 41 franchises showed scripted decision-tree bots zero phone captures and lost sales-critical conversion window; SMS-first alternative achieved 92% new customer contact rate. Market data confirmed: $15.12B AI customer service market in 2026 with 80% adoption intent, but only 25% fully integrated; 79% customer preference for human agents persisted despite cost investments in automation. The tier remained good-practice—scripted bots delivered proven, economically sound automation in well-defined, high-volume transactional use cases—but the evidence base reinforced a narrowing ceiling: customer frustration, implementation fragility, and vendor exit signals constrain any expansion beyond the current scope.