The State of AI in Agencies Why 66 Still Have No Measurable AI Results

Elevate Your Intelligence Platform (0:00)

The digital media landscape requires clarity and unlimited intelligence across competitors, audiences, inventory, and data, all in one place. This platform helps uncover performance drivers, provides path-to-conversion reporting, media mix modeling, and offers real-time adaptive intelligence with clear revenue impact.

Webinar Introduction and Presenters (0:27)

David Burkowitz from March Media's AI Marketers Guild introduces a special edition of AI Insiders with partners at AI Digital to discuss the state of AI in agencies. David Monero, Chief AI Officer at AI Digital, leads AI Digital Labs, a dedicated AI transformation practice for agencies and brands. He is joined by Boris, head of the AI Digital Labs incubator, who will demonstrate some recently shipped tools.

AI Digital Labs' Approach and the "Adapt or Die" Mentality (1:25)

The session, "The State of AI in Agencies: Hard Data, Real Tools, and What Actually Works," will cover findings from a benchmark survey of over 100 agency and brand leaders. This survey was developed due to skepticism about existing industry surveys that asked vague questions. The focus here is on operational questions. The industry often exhibits an "adapt or die" mentality regarding AI, leading to an impulse to want AI immediately without a clear understanding of its application. AI Digital Labs aims to bridge this gap, turning excitement and ambition into action, awareness, adoption, and advantage. This need is mirrored by large enterprises like OpenAI and Anthropic, who are deploying engineers to help clients integrate AI, a service largely unavailable to the mid-market.

The Three Pillars of AI Transformation: Strategy, Training, and Engineering (5:22)

AI Digital Labs' approach is built on three connected tracks for transformation. First, AI Strategy and Advisory, which aligns leadership on ambition, risk, and investment, including roadmaps and governance frameworks. This involves critical conversations about go-to-market messaging and truly showing, rather than just telling, how an agency is advancing with AI. Second, Training and Upskilling, providing role-specific bootcamps for strategists, creatives, media, analytics, and account teams. The emphasis is on hands-on keyboard experience to facilitate "epiphany-delivering" learning, crucial for adults in organizations. Third, Forward-Deployed AI Engineering, focusing on custom agents, workflow redesigns, and tool builds, often developed in the same timeframe it previously took to scope projects. These tracks are compounding, not strictly chronological. The biggest risk in AI adoption is inaction due to the rapid pace of technological change.

Survey Findings: Leaders, Laggards, and the "Walking Dead" Agencies (11:12)

A survey of 100 agencies and brands identified three categories: "Leaders" (16%) have AI embedded across teams with measurable KPIs and client stories. "Laggards" (6%) have no AI activity but are honest about it, making them easier to help. The largest group, the "Walking Dead" (two-thirds of the field), are "drafting roadmaps, running ad-hoc experiments, forming committees, piloting tools, and making decks." While these efforts seem like progress, they often don't move the work forward or produce measurable results. Many have been engaged in these activities for two years without significant change in how work is done, essentially "checking the box" rather than truly embedding AI.

The AI "Say-Do Chasm" and Lack of Unique AI Stories (13:54)

Agencies acknowledge AI's criticality for competitiveness, rating it 8.1 out of 10 (and 9.4 for future success). However, confidence in answering client questions about AI drops to 5.8 out of 10, revealing a significant "say-do chasm" or confidence gap. A staggering 83.9% of agencies cannot articulate a unique or differentiating AI story, with 57% admitting to having only generic talking points. If an agency's AI story could be copied and pasted onto a competitor's website, it's merely "table stakes" and not a differentiator.

Key Barriers to AI Adoption: Skills Gap and Being Too Busy (15:57)

The primary barrier to AI adoption, by a wide margin, is the skills gap (61%). The second biggest barrier is being "too busy with daily client work" (52%), leading to a "side project trap." Budget and technology are much lower on the list of concerns. This suggests that the problem is largely one of leadership and prioritization, rather than a lack of available, affordable technology. The technology is rapidly evolving and accessible; the challenge lies in leveraging it effectively within organizations.

The Emerging AI Opportunity Gap and "Catch-Up Tax" (17:42)

While immediate competitive results from early AI adoption aren't drastically different yet, a significant "chasm" is emerging. Agencies that are diligently laying the groundwork – building data foundations, organizational readiness, and foundational skills – will be prepared for the rapid, "gradually then suddenly" leaps in AI technology. This creates an "opportunity gap": doing slightly more or better than competitors now will yield enormous advantages. Waiting longer will incur a "catch-up tax," making it progressively more expensive and difficult to close the gap.

Debunking Common AI Adoption Justifications (19:39)

Several common justifications for slow AI adoption are addressed. First, client confidentiality: Enterprise-safe, non-training models with zero data retention exist and have been adopted by highly regulated industries like banking for years. This is a surmountable obstacle, similar to using other SaaS tools. Second, being too busy with client work: This is akin to being "too busy rowing to notice your boat has a motor." Agencies must embrace AI or face dire consequences. Third, unclear ROI: With AI tools costing as little as $20-$30 a month, the investment is negligible, making the "R" (return) the primary issue. Many agencies dismiss AI after initial, untargeted experiments, failing to implement proper training, workflow integration, or measurement. Finally, the idea that clients don't want AI is often a misunderstanding; clients are anxious about how AI is used and want "AI-powered" solutions, representing an opportunity rather than an objection.

A Dual-Track Strategy for AI Capability Building and Immediate Impact (24:32)

Effective AI adoption requires a dual-track approach. One track focuses on building organizational capability—a long-term investment in transformation, learning, and skill development that takes time and continuous effort. The other track addresses the immediate need for results by engaging an operating partner like AI Digital. This partner leverages their AI expertise to deliver immediate impact and "cold start activation packs" (white-labeled AI solutions), providing agencies with results while they build internal capabilities. This parallel approach aims for both sustained growth and quick wins, ultimately converging for comprehensive AI integration.

Introducing AI Digital's Incubator Program and Rapid Tool Development (26:07)

AI Digital's AI Incubator program quickly builds tools and prototypes in response to common client needs, often within days. This "constant shipping cadence" has generated around 20 tools in the past two months. These tools are often built to address known client inquiries, such as how to improve rankings in AI search engines. This incubation work also informs their flagship technology, Elevate, and their forward-deployed AI engineering for partners.

Demo: AI Engine Optimization (AEO) Tool for Search Ranking (28:06)

Boris demonstrates the AI Engine Optimization (AEO) tool, built in response to agency partners asking how their clients rank in AI search surfaces (beyond traditional SEO and paid ads). The tool focuses on providing actionable outputs, not just metrics like "share of model." It accounts for the varying prioritization and constant changes in different AI models (e.g., ChatGPT, Google Gemini, Google AI Overviews). Users can prompt the tool with a product (e.g., Coca-Cola Zero) and specify target engines. The report shows share of model, sentiment, and presence in comparative or negative intents. Crucially, it provides engine-specific signals (e.g., YouTube presence for Gemini) and detailed priority actions, including specific channels, candidates, podcasts, and rewrite examples for content and web pages.

Demo: Competitor Campaign Review Tool (36:58)

The Competitor Campaign Review tool analyzes a client's marketing spend and campaigns against their peer group, specifically at the sub-brand or product level (e.g., mortgages within financial services). It shows share of voice and spend, ranks campaigns based on performance, provides specific creatives (e.g., CTV video ads for Pepsi), and analyzes publisher spend across the peer group, identifying contested versus exclusive inventory. It also scrapes landing pages associated with campaigns for comparison.

Demo: Synthetic Focus Group for Pre-Campaign Optimization (41:48)

The Synthetic Focus Group is a pre-campaign optimization tool, built in just a couple of days. Before launching an expensive campaign, this tool assesses how a creative or concept will perform. It takes defined personas, builds them out, bombards them with questions, and creates a sentiment map across these personas for the product, concept, or creative. This reveals what specific personas would say, recommend, or not recommend, which is highly valuable for smaller campaigns where a traditional focus group would be cost-prohibitive. Academic studies approximate its effectiveness to 97% accuracy for Likert ratings under specific constraints. The distinction of such tools lies not in the underlying AI model, but in how system prompts, resources, and context are set up, combining human insight with AI capabilities.

Q&A and Real-World Application Case Study (Livele Lead Voice Agent) (46:17)

During Q&A, the discussion covers measuring AI impact longitudinally by tracking subqueries and comparing "trained" versus "grounded" models to identify changes over time. The accuracy of synthetic focus groups is also addressed, noting academic studies and emphasizing the need for human review of AI outputs, comparing AI's performance against human baselines (e.g., human error rates) rather than expecting 100% accuracy. The concept of "client AI wow moments" is introduced as a qualitative measure of success, suggesting that if AI helps thrill and retain clients, it's succeeding.

A case study highlights a tool called "Livele Lead," a voice agent built for a client obsessed with direct mail. This tool transforms direct mail responses into interactive voice conversations. Users scan a QR code or visit a domain to access a screen where they can start a conversation. The AI voice agent answers questions about considered purchases (e.g., health plan co-pays, in-network doctors) by drawing from the client's sales playbook and resources. The tool captures a full transcript of each conversation, duration, questions asked, and, if provided, contact information. This provides clear brand signal about audience reactions and potential leads, creating a new, high-margin revenue stream for the agency in a matter of weeks.

Concluding Remarks and Contact Information (56:58)

The session concludes with appreciation for the audience's engagement and questions. Information on accessing the tools and research will be shared. David Monero and Boris invite attendees to connect via email (David.Monero@ai.digital or Boris's email) or LinkedIn for further questions and follow-up.