IAB Australia publishes LLM prompting guide for marketing professionals

IAB Australia released a comprehensive guide on November 19, 2025, designed to help marketing professionals, media agencies, publishers, and advertising technology stakeholders improve their interactions with large language models. The guide, created by the IAB Australia AI Working Group, provides practical prompting techniques with vendor-agnostic instructions applicable across multiple AI platforms.

The 2025 edition addresses a growing need within the advertising industry. Marketers are conducting competitor research and mapping out market positioning, planners are brainstorming on media plans and automating workflow, analysts are summarising performance data, creatives are drafting social captions or storyboards, and publishers are matching audiences with content or suggesting ad placements.

The timing aligns with rapid AI adoption across digital advertising. IAB Europe research from September 2025 showed 85% adoption rates among companies surveyed between July and August 2025. That study, which examined 95 companies including publishers, ad tech firms, agencies, and advertisers across European markets, revealed strong investment momentum alongside significant governance gaps.

The IAB Australia guide presents a four-part framework addressing Context, Role, Objective, and Task. Context defines the background so AI understands the situation and constraints, including relevant brand, audience, challenge or market conditions. Role assigns the AI a perspective or expertise to shape the response, such as brand strategist, media planner, analyst, or creative director. Objective clarifies what outcome users want from the AI, whether insights, ideas, an action plan, or copy. Task specifies the concrete action or format for the response.

Several alternative frameworks exist across the industry. ChatGPT uses GRACE: Goal, Role, Assets, Constraints, and Expectation. Microsoft employs Goal + Context + Source + Expectations. Google uses Persona + Task + Context + Format. Anthropic applies Role → Context → Instructions → Constraints → Revision. Perplexity follows Question → Context → Action → Format.

The guide includes mandatory prompt techniques. Clear and specific goal descriptions cut down unnecessary back-and-forth interactions. Context and data provision improves LLM performance significantly when relevant background information accompanies requests. Role assignment helps guide style and depth of expertise. Format and constraint specifications make outputs easier to use, allowing requests for specific sentence counts, word limits, bullet point numbers, or table structures with defined columns.

Fact-checking remains essential. Large language models can occasionally generate convincing but inaccurate information. The guide recommends asking models to cite sources, then verifying those references against reliable data. Users should protect confidentiality by avoiding proprietary or sensitive material unless explicitly approved. Prompts should stay concise, as overly long prompts may be cut off or misinterpreted. Iteration works best through focused prompts, response reviews, and step-by-step refinement until output meets requirements.

The document provides specific marketing examples demonstrating good versus bad prompts. For campaign planning channel role maps, poor prompts requesting "a media plan for our new product" return generic channel lists without rationale, omitting KPIs, timing or risks. Effective prompts instruct AI to "act as a media strategist" and provide specific details: brand name, launch location, target audience demographics, budget figures, and requests for structured tables showing channels, KPIs, flighting schedule, test ideas and risks.

Reporting scenarios illustrate practical applications. Basic prompts asking AI to "explain why platform ROAS doesn't match sales" produce generic explanations about attribution differences without referencing specific figures. Superior prompts position AI as a measurement specialist, provide platform metrics including impressions, conversions, and reported ROAS alongside actual sales data including revenue and spend, then request comparative analysis showing why platform metrics overstate performance and suggestions for triangulation methods.

Audience creation demonstrates similar principles. Simple requests to "create an audience for my campaign" return broad demographics without consideration of data sources or compliance requirements. Detailed prompts instructing AI to act as a data strategist for specific brand types, define target audiences using particular data sources, describe audience composition, address privacy compliance considerations, and outline activation methods through advertising technology infrastructure produce substantially more useful outputs.

Optional techniques elevate quality further. Treating AI as a collaborator through iterative, multi-turn processes yields better results than one-shot queries. Users can start with base prompts, review responses, then refine by adjusting tone, adding missing details, or asking for new angles. Defining rules and guardrails by specifying words, topics, or claims to avoid helps maintain brand safety and alignment.

Providing examples or inspiration shows AI what "good" looks like. Sharing snippets of past campaigns, preferred tone of voice, or sample copy steers output toward desired styles. Letting AI interview users by encouraging clarifying questions before answering helps fill information gaps and tailor responses more precisely to needs.

Reusable variables create flexible prompts using placeholders like {brand}, {audience}, or {budget}. This approach enables repurposing and scaling prompts across different clients or campaigns without rewriting each time. Building prompt libraries by saving versions that consistently deliver strong results creates powerful reference material for future projects and training new team members.

The guide addresses practical concerns around AI integration. Amazon launched free AI prompts for sponsored ad campaigns on November 11, 2025, automatically engaging shoppers with conversational, interactive ad variations during the beta phase. That technology leverages Amazon's first-party signals from detail pages, Brand Stores, campaign data, and additional sources to surface product expertise at decision moments.

Infrastructure developments continue accelerating across the industry. Google released an open-source Model Context Protocol server on October 7, 2025, enabling AI tools to query advertising campaigns through natural language interfaces. The development marked a transition from exploration to implementation for AI integration with advertiser accounts. Sample prompts include "what customers do I have access to?" and "How many active campaigns do I have?" The system processes these conversational requests and returns structured data from the Google Ads API.

Consumer behavior shifts accompanying AI adoption create urgency around effective prompting strategies. Equativ's October 2025 survey of 4,000 North American and European consumers revealed 67% use AI weekly, with 38% searching less and 30% visiting fewer websites as LLMs reshape advertising. Previous research from MediaLink conducted between August and September 2025 found that 84% of marketers observe consumer behavior shifts away from conventional search and web browsing.

Technical accuracy challenges accompany growing adoption. LLM tracking tools face accuracy problems from personalization features, according to analysis shared on November 3, 2025. ChatGPT and similar systems rewrite queries based on individual user data, producing responses that differ substantially from what tracking tools report. The same query generates different results for different users based on stored information about location, interests, and previous interactions.

Industry standardization efforts address infrastructure needs. IAB Tech Lab released its Agentic RTB Framework version 1.0 for public comment on November 13, 2025, introducing standardized specifications for deploying containerized agents within real-time bidding infrastructure. The public comment period extends through January 15, 2026. The framework establishes requirements for implementing agent services that operate within host platforms.

The IAB Australia AI Working Group was established in May 2025 to build awareness of key topics, trends and best practices in relation to the use of AI in digital advertising. The group operates through five workstreams: Education & Upskilling, Standards & Governance, Innovation & Use Cases, Infrastructure & Readiness, and Measurement, Impact & Sustainability.

Current members include representatives from Adobe, Bench Media, Criteo, Dentsu, IVE Group, Google, Integral Ad Science, Microsoft, MiQ, News Corp Australia, OMD, Paramount, PHD, Prophet, Reddit, Scope3, Seven Network, Spark Foundry, Strike Social, Suncorp, WPP Media, Yahoo, Youi, and Zenith Media. Daevid Richards from News Corp Australia and Kellyn Coetzee from Zenith Media serve as co-chairs.

The guide includes links to additional prompting resources. Anthropic provides prompt engineering documentation covering overview topics and building effective agents. Microsoft Learn offers prompting guidance through AI Builder. A GitHub project called Prompt Buddy provides a free Microsoft Teams Power App using Dataverse for Teams, creating spaces where teams can share favorite AI prompts and upvote prompts from others.

The document concludes with a practice recommendation: commit 10 minutes daily to crafting and refining prompts. Trying one new prompt each day for two weeks builds skills applicable across multiple advertising tasks, whether building channel mix, reconciling ROAS, defining audience segments or planning content calendars. Regular practice demystifies AI interactions and builds prompting habits that deliver tangible results.

Timeline

• May 2025: IAB Australia establishes AI Working Group to build awareness of key AI topics in digital advertising
• July 2025: IAB Europe releases AI whitepaper addressing growth, guardrails and policy for European digital advertising
• July-August 2025: IAB Europe surveys 95 companies across European digital advertising markets, finding 85% AI adoption rate
• September 2025: IAB releases AI in Advertising Use Case Map organizing 84 AI use cases across six categories
• October 2024: IAB Tech Lab releases AI in Advertising Primer providing comprehensive framework for understanding AI applications
• October 7, 2025: Google releases open-source Model Context Protocol server for Ads API integration
• October 14, 2025: Adobe launches LLM Optimizer enterprise tool for AI visibility optimization
• October 15, 2025: Ad Context Protocol launches for advertising automation amid industry debate
• November 3, 2025: LLM tracking tools face accuracy crisis from personalization features
• November 11, 2025: Amazon launches free AI prompts for sponsored ad campaigns
• November 13, 2025: IAB Tech Lab releases Agentic RTB Framework version 1.0 for public comment
• November 19, 2025: IAB Australia AI Working Group releases comprehensive guide to prompting with LLMs for marketing professionals

Summary

Who: IAB Australia AI Working Group, comprising representatives from Adobe, Bench Media, Criteo, Dentsu, IVE Group, Google, Integral Ad Science, Microsoft, MiQ, News Corp Australia, OMD, Paramount, PHD, Prophet, Reddit, Scope3, Seven Network, Spark Foundry, Strike Social, Suncorp, WPP Media, Yahoo, Youi, and Zenith Media, with co-chairs Daevid Richards from News Corp Australia and Kellyn Coetzee from Zenith Media.

What: A comprehensive, vendor-agnostic guide providing prompt engineering techniques for large language models, featuring a four-part framework (Context, Role, Objective, Task), mandatory techniques including clear goal specification and fact-checking requirements, and optional advanced techniques like iterative collaboration, guardrail definition, example provision, and reusable variable implementation.

When: Released on November 19, 2025, following the working group's establishment in May 2025.

Where: Australia, targeting brand marketers, media agencies, publishers, and advertising technology stakeholders across the digital advertising ecosystem, with applications extending globally through vendor-agnostic guidance applicable to multiple AI platforms including ChatGPT, Microsoft, Google, Anthropic, and Perplexity systems.

Why: Addresses growing need for effective AI interaction skills as generative AI reshapes advertising industry workflows across competitor research, market positioning, media planning, performance analysis, creative development, and audience targeting, while 85% adoption rates demonstrate rapid market penetration requiring standardized best practices to maximize value from large language model investments.