The Record Speaks for Itself

A client came to us with a deceptively simple question: *How do we get accurate information about our company into ChatGPT?*

At first glance, it sounded like a content problem.

It wasn't.

We spent some time tracing the path from how AI platforms actually work backwards through every point where official corporate information either reaches them or doesn't. What we found changed how we thought about the whole problem. The question wasn't really about content at all. It was about data — specifically, the quality, structure, and verifiability of the information these systems are pulling from.

That reframe matters more now than it did even a year ago.

---

The GEO Trap

If you've been in a communications or brand role lately, you've probably heard some version of a pitch about Generative Engine Optimization — GEO — or its cousin, Answer Engine Optimization (AEO). The premise is straightforward: AI platforms like ChatGPT, Perplexity, and Google's AI Overviews are eating search traffic. People ask questions, the AI answers, and they never click through to your site. So you need a strategy to show up in those answers.

Fair enough. That's a real phenomenon, and it's only accelerating.

But here's where a lot of the current thinking goes sideways. Most GEO and AEO strategies are essentially SEO on steroids — optimized content designed to capture visibility in a world where there's nothing to click. Write the right way. Use the right signals. Get the algorithm to notice you.

That framing treats AI visibility as a content distribution problem. Get more of your stuff in front of the machines, and you win.

The problem is that approach conflates showing up with being believed. And in an AI-powered information environment, those are two very different things.

---

What AMEC Got Right

Earlier this year, AMEC — the International Association for the Measurement and Evaluation of Communication — released a white paper in collaboration with the Global Data Quality Initiative that reframes this entire conversation. The paper, focused on data quality principles for the AI era, makes a point that deserves a lot more attention than it's getting in most communications circles.

The argument, stripped of the academic framing, goes something like this: AI systems don't just crawl content. They learn from data. And if the data they're trained on is incomplete, unstructured, inconsistent, or unverifiable, the output is going to reflect all of that. Hallucinations aren't a quirk of the technology — they're a symptom of a data quality problem. The paper cites a hallucination rate of 79% in some 2025 reasoning-oriented LLMs. Eighty percent of enterprise generative AI initiatives fail to meet expectations, with data quality cited as the dominant root cause.

Those aren't fringe numbers. Those are the conditions your brand reputation is operating inside right now.

The paper introduces what it calls "context engineering" — the discipline of organizing, governing, and delivering structured data in a way that gives AI systems the grounding they need to produce reliable, accurate output. Gartner, cited in the paper, put it plainly in 2025: context engineering is in, prompt engineering is out. The organizations that figure out how to feed AI systems clean, structured, trustworthy information will have a meaningful and compounding advantage over those still optimizing headlines and meta descriptions.

That's a different game entirely.

---

It's a Data Problem. Treat It Like One.

Here's what that shift actually looks like in practice.

Legacy GEO thinking says: publish more content in the formats AI prefers, seed it across the right channels, and optimize for retrieval. That's not wrong, exactly. But it's the beginning of the strategy, not the whole thing.

The more durable approach treats every piece of official corporate communication — press releases, executive statements, product announcements, financial disclosures — as a data asset. One that should be structured, verified, traceable, and machine-readable from the moment it's created.

AMEC's framework asks four core diagnostic questions of any data feeding into AI systems: Where does it come from? Is it accurate? Does it correspond to real-world facts? And is it trusted — not just technically, but by the humans and systems that depend on it?

Most corporate communications today fail at least two of these tests. Not because the information is wrong, but because it's not structured in a way that lets AI systems evaluate its provenance or verify its accuracy. It's a press release on a wire. It's a PDF on a newsroom site. It's content — not data.

The difference matters because, as the AMEC paper notes, errors in unstructured data don't get absorbed. They get amplified. Every generative AI system that ingests poorly structured or unverifiable information about your company becomes a potential vector for inaccuracy — stated confidently, at scale, to anyone who asks.

---

The Long Game

Here's the part of this conversation that tends to get lost in the shorter-cycle thinking that dominates most communications planning.

Building a high-quality, structured data record of your organization's official communications isn't just about winning the next AI answer. It's about building the foundation that AI platforms will reference to construct an accurate model of who your company is.

Think of it as a cumulative reputation record — not a content calendar. Every structured, verified, machine-readable communication your organization publishes adds to that record. Over time, if it's done consistently and rigorously, you're not just optimizing for retrieval. You're building the underlying data profile that AI systems will use to understand and represent your brand.

That's a fundamentally different timeline than most communications strategies operate on. But it's the right one. Because the organizations that treat this as a data problem now will be the ones with accurate, trusted AI representation years from now — when the competitive gap between brands that got this right and brands that didn't will be very hard to close.

The AMEC paper closes with a point worth sitting with: the communications community is uniquely positioned to lead on this, because it sits at exactly the intersection of content production, brand knowledge, and data governance that the problem requires. That's not a stretch. It's an accurate read of where the leverage is.

The question is whether brands and their communications teams are willing to think about their information not as messaging, but as a record.

Because in the end, the record is what speaks.

---

*This piece draws on the AMEC × Global Data Quality Initiative white paper "AMEC Data Quality Principles: Tackling the Unstructured Data Quality Challenges in the AI Era" (2026). The full paper is available via AMEC at amecorg.com.*