Most agencies added “AEO” to their service page in the last twelve months. The category is new enough that nearly anyone can claim expertise, and the buying signals B2B SaaS marketers normally rely on — case studies, ranking screenshots, pipeline attribution — don’t translate cleanly to answer engines yet.
This is a checklist for the discovery call. Nine questions that separate agencies running a real answer engine optimization practice from agencies running keyword work under a new label. Each question includes what a strong answer sounds like, so you can tell within thirty minutes whether the agency you’re talking to has actually done the work.
1. How do you measure AI search visibility, and which surfaces do you track?
The honest answer names specific surfaces and specific tools. ChatGPT, Perplexity, Google AI Overviews, Google AI Mode, Gemini, and Claude are the surfaces most B2B SaaS buyers are using during evaluation. Each one indexes differently and cites differently. An agency that says they “track AI visibility” without naming the surface is usually checking one of them, then extrapolating.
For tooling, expect to hear named platforms: Profound, Peec AI, Otterly, AthenaHQ, Goodie, BlueprintIQ, HubSpot, or proprietary tooling the agency built. Ask how often they pull data, how many prompt variants they test per query, and whether they track competitor citation share, not just your own.
2. What’s your methodology for getting a brand cited in an answer, not just ranked on a page?
This is the question that exposes repackaged SEO. Citation in an LLM response depends on a different set of signals than blue-link ranking: clean Q&A formatting, semantic clarity, structured data the model can parse, consistent entity definitions across the open web, and external mentions in sources the model already trusts.
A strong answer will distinguish between on-page work (rewriting pages so an LLM can extract a clean answer) and off-page work (getting your brand named in the third-party sources LLMs actually pull from — Reddit threads, G2 categories, niche publications, podcast transcripts). Agencies running a serious AEO practice do both. Agencies running rebadged SEO usually only do the first.
3. Can you show me a current-state audit of how I show up in LLMs today?
Any AEO engagement should start with a baseline. The audit should answer: which prompts in your buyer’s research journey currently surface your brand, which surface competitors instead, and where your brand gets misrepresented or omitted entirely. It should cover both your own product queries and the category queries (e.g., “best X for Y”) that buyers actually run.
At VSSL we run two audits side-by-side for every B2B SaaS engagement: a model-output audit that maps how the brand currently surfaces across LLMs, and a technical on-page audit that measures whether the site is structurally ready to be ingested. Looked at together, they show the gap between current state and what good looks like — which is what scopes the actual work.
4. Which clients have you done this for, and what changed in their LLM citations?
Specifics matter here, because the field is small enough that proof is checkable. Ask for a brand name, a category prompt, and a before/after — ideally something you can verify by running the prompt yourself in ChatGPT or Perplexity.
Useful proof looks like: “We took [client] from zero citations to consistent inclusion in ChatGPT answers for [‘category prompt’] over four months.” Less useful proof: traffic numbers, page rankings, or vague references to “AI-driven growth.” Traditional metrics may improve as a side effect of AEO work, but they’re not what you hired the agency for.
5. How do you handle the technical layer — schema, structured data, content hierarchy, llms.txt?
LLMs don’t crawl the same way Google does, and many B2B SaaS sites built for blue-link SEO have ingestion problems they don’t realize: JavaScript-rendered content the model can’t parse, long-form pages without clear answer extraction points, missing or shallow schema, ambiguous entity references.
The agency should be able to talk about: FAQ and HowTo schema where it earns its place (not stuffed everywhere), Article and Organization schema with clean entity definitions, content hierarchy that surfaces direct answers in the first 100–200 words, llms.txt implementation, and how they handle JavaScript rendering for AI crawlers. If the answers are vague or the agency defers entirely to “the dev team,” they’re not equipped for the technical track.
6. How do you build the off-page citation footprint LLMs actually trust?
LLMs disproportionately cite a small number of source types: Reddit, Wikipedia, G2 and Capterra, established trade publications, well-indexed podcasts, and the long tail of expert blog posts and LinkedIn content from named individuals. Getting your brand mentioned across that footprint is a different muscle than link building for SEO.
A real generative AI search optimization program will include some combination of: digital PR aimed at the publications LLMs cite in your category, structured presence on G2 and category review sites, founder and SME content on LinkedIn under named bylines, and Reddit engagement that’s actually engagement, not stuffing. If the agency only talks about on-site optimization, they’re missing half the work.
7. What does reporting look like, and how do you tie it to pipeline?
The weakest reporting in this category is a screenshot of “we got cited in ChatGPT” with no recurrence tracking, no competitor benchmark, and no way to tell whether anything changed for the business.
A serious AEO reporting cadence should include: tracked citation frequency across surfaces over time, competitor share of voice on category prompts, prompt-level wins and losses (which queries you’re now in, which you’re not), and an attribution model for AI-driven traffic and conversions through GA4 or your warehouse. Pipeline attribution from LLM traffic is genuinely hard right now — anyone claiming exact ROI numbers is overselling. Anyone with no plan to measure it is underdelivering.
8. Do you specialize in B2B SaaS, or is this one of many verticals you serve?
This matters more in B2B SaaS AEO services than in most categories because B2B SaaS buyer prompts have a specific shape: comparison queries (“X vs Y”), alternatives queries (“alternatives to X”), category queries (“best X for [use case]”), and integration queries. An agency that has worked through dozens of these knows which content patterns get cited and which don’t.
Generalist agencies can absolutely do good AEO work, but they’re learning your buyer’s prompt landscape on your budget. Ask what percentage of the agency’s book is B2B SaaS, and ask for category-prompt examples from clients in adjacent niches.
9. What’s the engagement structure, and what happens in the first 30, 60, and 90 days?
Short engagements rarely move the needle on AEO. LLM training cycles, citation recurrence, and off-page footprint building all compound over months, not weeks. But the first 90 days should still produce visible artifacts: a baseline audit you can hand to leadership, a prioritized list of pages to restructure, an off-page outreach plan with named target publications, and ideally the first measurable shifts in citation frequency.
Watch for two failure modes. The first is agencies that promise rankings inside 30 days — that’s an SEO promise, not an AEO one, and it usually means the work is repackaged. The second is open-ended retainers with no scoped deliverables — common in this category right now, because methodologies are still being built in public.
What this comparison should leave you with
If you’re doing an AEO agency comparison across three or four shops, the questions above will produce noticeably different answers. Strong agencies will name the surfaces, name the tools, name the clients, and talk fluently about both the technical and off-page sides of the work. Weaker agencies will lean on the category language — “AI visibility,” “LLM optimization,” “generative search” — without specifics behind any of it.
The category is new, but the diligence is the same as for any other agency hire: ask for proof, check it yourself, and prefer the agency that can show you a thing they did over the agency that can describe a thing they would do.
VSSL builds AEO programs for B2B SaaS teams across both tracks — model-output audits that map your current LLM presence, and technical on-page audits that prepare your site to be ingested cleanly. If you want to see what your current AI search visibility looks like before you start interviewing agencies, we can run the baseline.