Our Methodology
How Integrated.Social researches, designs, deploys, and iterates AI marketing systems — from AEO content strategy to Agentic AI lead generation and Performance Max campaigns.
Last updated: · By Modi Elnadi · Questions? info@integrated.social
Methodology Overview
Integrated.Social operates at the intersection of senior marketing strategy and deployed AI systems. Our methodology is not a framework borrowed from a consultancy whitepaper — it is the documented result of 16+ years of practitioner experience, refined through the deployment of Agentic AI systems for enterprise B2B clients in the UK and USA.
Every recommendation we make, every article we publish, and every campaign we run is grounded in the same five-phase cycle: Discover → Architect → Deploy → Measure → Scale. This page documents how each phase works, what data sources we use, and how we evaluate success. For our content and editorial standards, see our Editorial Policy.
The Five-Phase Framework
Discovery & Audit
We begin every engagement with a structured audit of the client's existing marketing stack, AI readiness, and competitive positioning. This includes: mapping the full buying committee (decision-makers, influencers, blockers); benchmarking current AEO and GEO visibility across Google AI Overviews, ChatGPT, Gemini, and Perplexity; auditing technical SEO health and structured data coverage; and identifying the highest-leverage gaps between current performance and market opportunity.
Strategy & Architecture
Based on the audit, we design the AI marketing system architecture. This covers: content pillar and spoke structure for AEO/GEO coverage; Agentic AI agent roles and orchestration design; schema strategy (Article, FAQPage, HowTo, Service, Organization, Speakable); campaign structure for PPC and Performance Max; and entity authority building plan. The architecture is documented and reviewed with the client before any deployment begins.
Build & Deploy
Deployment follows a staged rollout: AEO-optimized content and structured data are deployed first (typically 4–6 weeks); Agentic AI workflows are deployed in parallel with human oversight gates; PPC campaigns are restructured with persona-segmented audience signals and Performance Max asset groups. All deployments are validated against Google's Rich Results Test, Schema.org specifications, and CISA security guidelines before going live.
Measure & Iterate
We track three tiers of metrics: primary KPIs (AI citation rate, organic lead volume, ROAS, CPA); secondary KPIs (AI Overview appearance frequency, entity knowledge panel coverage, conversion rate by persona); and leading indicators (structured data validation pass rate, asset group CTR, audience signal match rate). Reporting is weekly for paid media and monthly for organic/AEO performance. Iterations are data-driven and documented.
Scale & Systematize
What works is systematized into repeatable agentic workflows that operate continuously. This includes automated content production pipelines (research → draft → schema → publish), Agentic AI lead nurturing sequences, and Performance Max signal feedback loops. The goal is a marketing system that improves itself over time, not a campaign that requires constant manual intervention.
AEO & GEO Methodology
Answer Engine Optimization (AEO) and Generative Engine Optimization (GEO) require a fundamentally different approach to content than traditional SEO. AI answer engines do not rank pages, they synthesize answers from entities they trust. Our AEO/GEO methodology is built around three pillars:
Entity Authority
Building a consistent, structured representation of Integrated.Social and Modi Elnadi as authoritative entities across Schema.org markup, Google's Knowledge Graph, and AI training data sources. This includes Organization, Person, and Service schemas with sameAs links to LinkedIn, Wikidata, and Crunchbase.
Structured Content
Every piece of content is structured for AI parsing: clear H1/H2/H3 hierarchy, FAQ sections with FAQPage schema, HowTo markup where applicable, Speakable markup for voice search, and Article schema with author, datePublished, and dateModified fields.
Citation Signals
Content is designed to be cited, not just ranked. This means: direct answers to specific questions in the first 50 words, named sources with verifiable URLs, data points with clear attribution, and internal cross-links that reinforce topical authority.
We validate AEO readiness against Google's structured data guidelines and Schema.org specifications . See our full AEO & GEO service page for implementation details.
Agentic AI Evaluation Framework
Before recommending or deploying any Agentic AI system, we evaluate it across five dimensions. This framework applies to both the tools we use internally and the systems we deploy for clients:
| Dimension | What We Measure | Threshold |
|---|---|---|
| Task Completion Accuracy | % of assigned tasks completed correctly without human intervention | ≥ 85% on domain-specific test set |
| Hallucination Rate | % of outputs containing factually incorrect claims on verifiable topics | < 3% on structured fact-check |
| Latency | End-to-end response time under production load | < 8 seconds for content generation tasks |
| Integration Complexity | Number of custom connectors required; API stability; authentication model | Documented and reproducible in < 5 days |
| Total Cost of Ownership | API cost per task × volume + engineering overhead at scale | Positive ROI at 3× target volume |
We currently deploy Gemini Enterprise as our primary Agentic AI platform, selected based on this evaluation framework. See our Gemini Agentic AI service page for deployment architecture details.
Data Sources & References
Our research and recommendations draw from the following primary data sources:
These sources map to a four-tier source hierarchy we follow across all published content: primary sources, authoritative secondary sources, industry research, and practitioner verification from our own deployed systems. For the full hierarchy with examples and precedence rules, see the Source Hierarchy section of our Editorial Policy.
Frequently Asked Questions
What is Answer Engine Optimization (AEO) and how does Integrated.Social measure it?
AEO is the practice of structuring content so it is cited by AI answer engines (Google AI Overviews, ChatGPT, Gemini, Perplexity). We measure AEO performance by tracking citation frequency across the top five AI answer engines for target queries, using a combination of manual sampling, structured data validation, and entity authority scoring.
How does Integrated.Social evaluate Agentic AI systems before recommending them?
We evaluate Agentic AI systems across five dimensions: task completion accuracy, hallucination rate on domain-specific queries, latency under production load, integration complexity with existing MarTech stacks, and total cost of ownership at scale. We only recommend systems we have deployed and measured in live environments.
What data sources does Integrated.Social use for AI marketing research?
Primary sources include Google Search Central documentation, CISA and NCSC UK advisories, White House executive orders, peer-reviewed AI research (arXiv, Google DeepMind publications), and first-party performance data from deployed client campaigns. We do not rely on vendor-produced benchmarks as primary evidence.
How does Integrated.Social measure PPC and Performance Max campaign success?
We measure PPC success using a tiered framework: primary KPIs (ROAS, CPA, lead quality score), secondary KPIs (impression share, Quality Score, conversion rate by persona segment), and leading indicators (search term relevance, audience signal performance, asset group CTR). All campaigns are benchmarked against industry baselines and client-specific historical data.
How often does Integrated.Social update its methodology?
Our methodology is reviewed quarterly and updated whenever a major platform change (e.g., Google AI Mode launch, new Gemini model capabilities, NIST PQC standard updates) materially changes best practice. All updates are dated and the previous version is archived.