Building Your Reporting Narrative

TL;DR

• AI search is probabilistic, not deterministic. There's no single rank to report, so your narrative needs to center on trends across mention rate, share of voice, and sentiment.
• Use the four-question framework to structure every leadership readout: Are we in the conversation? Are we growing? Are we positioned correctly? What are we doing next?
• When sentiment drops, look at third-party sources first. 85% of top-of-funnel brand mentions come from content you didn't write.
• The AirOps Model Context Protocol (MCP) lets you query your analytics in plain language through Claude, so you can answer ad hoc questions without building new dashboards.
• Every report should end with a specific recommendation tied to data. That's what gets programs approved.

Why AI search needs a different story

You've learned to read the AEO dashboards. You can spot what moved and diagnose what didn't. But your leadership team hasn't been on that journey with you. They don't know what mention rate means, and they don't care about citation share yet.

The core problem: organic search reporting was deterministic. You rank number three, then number one. Traffic goes up. Done. AI search is probabilistic. Run the same prompt five times and you get different results each time. There's no single rank to report.

Reporting now centers on how often you appear, in what context, and with what sentiment. That requires a completely different narrative structure when you're presenting to leadership.

The four-question framework

Justina introduces a repeatable structure for every AEO report you give to leadership, regardless of the time period or audience. Each question maps to metrics your stakeholders can act on.

Question 1: Are we in the conversation?

This is the baseline. For the topics and prompts that matter to your business, are AI engines mentioning your brand at all?

The headline metrics live on the visibility tab in AirOps: mention rate and share of voice.

Check mention rate by topic to see where you're ahead and where you're lagging against competitors. For example, a brand might lead on one topic but trail on another within the same category.

Question 2: Are we growing our visibility?

Leadership wants a trend line, not a snapshot.

• Filter on the tags you set up in the previous lesson to measure whether specific actions worked.
• Track mention rate, share of voice, and average position against competitors to capture full context.

This is where you tell the story of whether you're gaining ground or falling behind.

Question 3: Are we positioned correctly?

Appearing isn't enough. You need to appear in the right context and with the right sentiment, described the way your Brand Kit defines your positioning. Justina covers two approaches here.

Bottoms-up approach:

• Pick a handful of prompts in the prompts tab that you really care about. Your enterprise feature set, a head-to-head against a specific competitor, or the category-defining query in your space.
• Click into each prompt to see every brand mentioned in the AI response and how each was referenced.
• Read the actual response. Look for positioning gaps, then note which prompt surfaces the issue and which sources are shaping it.

Top-down approach:

• Start with overall sentiment scores by topic in the sentiment tab.
• Identify which topic has the lowest score, then go to the citations view.
• Look at the third-party sources AI engines are pulling from. A two-year-old review mentioning a feature gap you've since closed, or an outdated competitor comparison page, can drag sentiment down.
• Justina notes that 85% of top-of-funnel brand mentions come from third-party sources. When sentiment is off, the fix is almost always off-site. Update what you can control and create new content that corrects what you can't.

Question 4: What are we doing next and why?

This is what separates a report from a narrative. Leadership doesn't want a dashboard. They want a recommendation.

Use your prioritized opportunities and internal priorities to connect the data you presented to the actions you're proposing.

Using the AirOps MCP for ad hoc questions

Everything you need for the four-question narrative lives in the Insights dashboards. But when you need to ask a more specific question, compare this week to last week or break out a single topic by provider, you don't need to build a new dashboard.

The AirOps MCP connects your AirOps data directly to an AI assistant like Claude.

Ask a question in plain language and it queries your analytics on the spot. No code or API calls required.

Justina demos asking how mention rate changed week over week. The response returns mention rate broken out by time period, including specific callouts like a noticeable dip on a particular date.

You can continue the conversation to investigate further. For example, asking for the top three topics driving a change and which prompts to pay attention to within each.

Putting it into practice

The next time you're preparing for a leadership check-in, skip the spreadsheet. Open the four questions, fill in the data from your Insights dashboards (or ask via the MCP), and add your recommendation. A recommendation backed by data your leadership team can act on is what gets budgets approved and programs expanded.