Answer Engine Optimization has a visibility problem that search never had. A ranking could be checked by typing the query. An AI answer is assembled fresh, differs by engine, and is gone the moment it is read. Which prompts is the brand absent from? Who appears instead? And of everything published last quarter, which pages earned citations?
NOLA Marketing ran an AEO program for Single Point of Contact (SPOC) on OtterlyAI: 64 tracked prompts, reviewed daily inside the agency and monthly with the client. NOLA wrote the new content that went live on SPOC’s site (the resource hubs, the question-and-answer pages) and used OtterlyAI’s tracking to decide where to aim it before publishing, then to see what had worked. Emily Matthews of NOLA Marketing walked us through it.
NOLA Marketing is a San Francisco Bay Area technology marketing agency specializing in AEO, content and product marketing for cybersecurity, cloud, AI, and enterprise software companies.
The client
Single Point of Contact (SPOC) gives managed service providers a way to offer enterprise-grade help desk, security operations center, and IT support without building that infrastructure themselves. SPOC had already done well with traditional SEO. The newer question was AI search: buyers were asking ChatGPT, Microsoft Copilot, and Google’s AI experiences who to trust for white-label IT and security services, and SPOC had no way of knowing whether it appeared.
The prompt set is the strategy
Much of an AEO program is decided before any content gets written. If the monitored prompts do not reflect how buyers research a category, the content strategy optimises toward the wrong questions.
SPOC supplied three inputs: its solution areas, its known competitors, and the keywords already in use for SEO. NOLA did not convert that keyword list into prompts. The team worked through the buyer journey with the client first, paying particular attention to questions closer to a purchase decision.
“To build out the prompts, we first discussed the full-funnel journey with our client. In that way, we were able to uncover prompts toward the bottom of the funnel, closer to purchase, that were important to our client’s business.” — Emily Matthews, NOLA Marketing
Those prompts matter disproportionately: they are where an engine recommends a vendor rather than explaining a category.
NOLA then expanded the set twice over. OtterlyAI’s Query Fan Out Analysis identified likely variations of the questions already chosen. Then competitors’ URLs went into OtterlyAI’s AI Prompt Research, surfacing the prompts tied to competitors’ AI visibility, a way to find opportunities before deciding what to create.
The final set: 64 prompts.
Engine coverage followed the client’s buyers
NOLA focused first on Google AI Overviews and AI Mode, plus ChatGPT. Copilot became more important for a specific reason.
“Given that SPOC sells to a lot of Microsoft shops, we also started paying closer attention to Copilot.”
The competitor list became an output
NOLA set up SPOC’s OtterlyAI report with the competitors SPOC already knew about. Over the following months, other companies kept appearing in the tracked prompts, and each time NOLA brought them back to the client to decide whether they belonged in the report.
“We started with the competitors SPOC provided. Over time, as we noticed other companies being mentioned or cited, we consulted with our client to see which of those should be included. They discovered competitors they didn’t even know they had.”
The competitive set stops being limited to the companies a client already has in mind, and Emily says this became one of the findings clients responded to most strongly. Growth numbers matter; a competitor you did not know existed changes how you understand your market.
The loop
NOLA reviewed results daily internally and monthly with SPOC. At 64 prompts across multiple engines, a full manual pass means hundreds of generated responses, and doing that often enough to catch change is difficult to sustain by hand.
The daily pass caught movement; the monthly session turned it into a decision.
“We spent more time reviewing responses to figure out where competitors were mentioned and where we were not. We adjusted our monthly content strategy accordingly.”
Against those gaps, NOLA built new resource center hubs around SPOC’s core service categories, filled with content answering the questions buyers were already putting to AI engines about white-label IT, help desk, and SOC services. Monitoring was not a reporting layer bolted on afterward. It was the planning input: find the gaps, publish, measure, adjust.
Then the data showed which content was working
By June, 2,657 of SPOC’s 3,617 citations (roughly 73%) came from the new question-and-answer content built for the program.
That does not mean every AEO program should become a pile of FAQs. It does show what worked here: pages built around specific buyer questions accounted for the large majority of citations.
Reaching that conclusion required connecting citations back to individual URLs. Without page-level data, the same program could produce the same result, and nobody learns which half of the work mattered.
The evidence
Mentions grew 433% and citations 444% across the three months, with citation volume accelerating rather than leveling off. SPOC held the #1 mention rank throughout and moved from third to first in citation rank.
“We were encouraged by the growth in mentions and citations. The real clincher was when we spoke with the head of sales at SPOC, who saw a 30% increase in inbound leads.”
The 30% happened alongside the program. That is not enough to attribute the whole increase to AI visibility, but it was the business signal beyond mentions and citations.
Two numbers that make more sense together
A later snapshot, covering the most recent three months, shows how differently rank and citation share can describe the same market.
Against its tracked competitive set, SPOC accounted for roughly 73% of brand mentions — 406, compared with 102 for the next brand.
Across all domains cited by AI engines in the category, SPOC’s own domain held about 9% of citations.
The Brand Rank tells you how you compare with named competitors. Citation share tells you how much of the wider information environment you occupy: the media sites, communities, social platforms, and vendors AI engines also cite. Read together, they stop a #1 ranking from being mistaken for owning a category.
One more figure from the same snapshot: SPOC’s domain accumulated 14,112 citations in the tracked category, against 3,047 for Reddit and 2,671 for YouTube. Our conversation with Stella Rising put Reddit as the dominant social citation source inside ChatGPT; a single vendor’s resource center out-citing it is not a common outcome.
The self-test that came first
Before offering AEO as a client service, NOLA ran the methodology on its own site. Restructuring content around buyer questions took it from 3 mentions to 70 and from 30 citations to 214 in a single month, moving it from seventh to first in mention rank.
The multipliers are dramatic, 23x and 7x, but the absolute numbers are the honest version, and per NOLA’s case study, they are why SPOC signed: the agency had proven the approach on its own property rather than presenting it as a framework.
Why OtterlyAI
NOLA had tested another AI visibility platform first. The problems were not feature quality; they were agency economics.
An agency monitors paying clients, but it also wants to run exploratory research for prospects before an engagement exists. When each additional account or prompt set carries substantial marginal cost, prospecting with data becomes hard to justify. Prompt flexibility matters for the same reason: different clients need different numbers and kinds of prompts, and capacity has to follow the engagement rather than a fixed allocation.
“The new Claude skill and integration is great. It saves so much time, especially when analyzing what the competition is doing.”
What having the data changed for NOLA
“We now have real data to share with other prospective clients.”
That matters in a discipline where agencies still have to explain what success looks like. The program also gave NOLA a sense of how results develop.
“We also have a sense of timing. It takes a few months to build up citations. Mentions generally follow. Then you see the business impact. So, we can coach our clients on what to expect in the process.”
Not a universal law of AEO, but the pattern observed here, and useful precisely because it lets an agency set expectations before the work starts, instead of reading every flat week as failure.
Run AEO for your clients on OtterlyAI
OtterlyAI monitors mentions, citations, competitive rank, and share of voice across ChatGPT, Google AI Overviews and AI Mode, Perplexity, Gemini, and Copilot, with prompt research and Query Fan Out for the work that happens before publishing.
