- Security shortlists form on narrow questions: frameworks, managed services, named alternatives and category fit, each with its own set of vendors named.
- 51% of B2B software buyers in G2’s 2026 survey said they start research with an AI assistant more often than with Google.
- On the 65 US buyer questions we measured, reddit.com was cited on 59 of 65 questions and youtube.com on 53 of 65 questions, far more often than any vendor site.
- No agency can guarantee a mention: OpenAI states that placement in ChatGPT search is not guaranteed.
What do CISOs ask AI engines before building a vendor shortlist?
Security buyers ask narrow questions: which tool helps meet a specific framework, whether to buy MDR or staff an in-house SOC, and which credible alternatives exist to a vendor they already know. Each of those questions produces its own short list of vendors.
AI assistants are now a routine first step in software research: in G2’s 2026 survey, 51% of B2B software buyers said they start research with an AI assistant more often than with Google [1]. In security, the questions that build a shortlist fall into four families. An agency should measure each family separately, because the vendors named in one rarely match the vendors named in another.
| Question family | Example buyer question | What gets a vendor named |
|---|---|---|
| Compliance frameworks | Tools to prepare for a SOC 2 Type II audit; how to meet CMMC Level 2 as a defense subcontractor | Pages that map product features to specific controls, with the framework version stated |
| Managed services | Best MDR provider for a mid-market company; managed security provider for HIPAA-covered entities | Clear scope (coverage hours, response model, who acts on alerts) and third-party descriptions that match it |
| Alternatives and comparisons | SIEM alternatives for a company with no SOC; identity security platform comparison for Okta shops | Factual comparison pages with sourced criteria, plus presence in the comparisons engines already cite |
| Category fit | Which EDR works best for a lean IT team; vulnerability management tools that prioritize by exploitability | A precise answer to the fit question, consistent across your site, review profiles and documentation |
Generic questions such as “best cybersecurity company” attract the largest vendors and say little about your odds. The fit and compliance questions are where a mid-sized vendor or an MSSP can realistically be named.
Why does generic GEO content fail with security buyers?
A CISO reads with a checklist in mind: framework version, deployment model, integrations, evidence. One wrong technical claim costs more trust than several good articles earn, and AI engines compare what you say with other sources before repeating it.
- Framework precision. SOC 2 Type I and Type II are different reports, and CMMC levels carry different requirements. Content that blurs them is ignored by buyers and contradicted by the documentation engines also read.
- Service versus product. MDR and MSSP providers sell people and process as much as software. Their questions (“MDR vs in-house SOC”, “managed security provider for…”) need their own pages and their own measurement.
- Named alternatives. “Alternative to [vendor]” questions are won with factual comparison pages: sourced criteria, a date, links to the competitor’s own pages, and no disparagement.
- Expert review. Every technical statement should be read by someone with security experience before it goes live. Ask any agency who that person is and what they check.
Which sources do AI engines cite, and where should a security vendor be present?
Most citations point to sites other than the vendor’s own: community threads, video, professional networks, reference works and trade media. A security vendor has to be described accurately there, not only on its own blog.
In one study of B2B software recommendations, the recommended tool’s own site was cited only 11.6% of the time [2]. Our own measurement shows the same pattern. On the 65 questions US buyers ask about GEO itself, including the question on this page, we ran ChatGPT, Claude, Gemini and Perplexity 5 times each (1,300 answers + 325 Google searches) and logged every cited domain [7]:
- reddit.com: cited on 59 of 65 questions (251 citations).
- youtube.com: cited on 53 of 65 questions (235 citations).
- en.wikipedia.org: cited on 36 of 65 questions.
- linkedin.com: cited on 34 of 65 questions (98 citations).
- searchengineland.com: cited on 26 of 65 questions (108 citations), on all five surfaces measured.
These are questions about GEO, not about security, so do not copy the list. Security answers also lean on analyst-style comparisons, vendor documentation and review platforms. The method transfers: ask your own buying questions several times on each engine, log every cited domain, and count how many questions each domain appears on. A domain present across many questions is where your facts must be right first.
How can a security vendor show up on Reddit and review sites without breaking the rules?
By participating in the open: named accounts that disclose their link to the company, genuine customer reviews, and useful answers to questions people are already asking. Security communities spot astroturfing quickly, and the platforms’ rules forbid it.
Reddit’s rules prohibit manipulation and spam [3]. In the US, the FTC rule on consumer reviews and testimonials bans fake reviews [4]. Beyond the legal risk, a security brand caught gaming a forum loses exactly the trust it sells. The safe work is slower and more durable: keep review profiles complete and current (deployment model, integrations, certifications, segments served), answer technical threads with a disclosed account, and give trade media data they can check.
Can a GEO agency guarantee that ChatGPT will name my security product?
No honest agency can promise it. OpenAI states that placement in ChatGPT search is not guaranteed, and the same question asked twice rarely returns the same list of brands [5][6].
With search on, ChatGPT rewrites the question into targeted searches, retrieves results from third-party search providers and ranks them on several factors [5]. In a SparkToro study, AI tools almost never returned the same brand list twice across 100 runs of the same prompt [6]. What an agency can commit to is the means: the questions measured, the content produced and published, the protocol, and raw answers you can check yourself.
How does ZivRank work with cybersecurity companies?
ZivRank runs a dedicated offer for security vendors and service providers, measured with the same protocol before and after the work. Read this section as our own description, checkable against the linked pages.
- Sector link. ZivRank’s founder also founded Resilium, an affiliated cybersecurity company.
- Measurement. 50 buying questions per program, each asked 5 times on ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews: 1,000 answers + 250 Google searches per measurement. A noise threshold of ±15%, set before any comparison, separates real change from normal variation. You receive the raw answers.
- Production. Our team writes, fact-checks and publishes the content, on your site and on third-party sources. Nothing technical goes live without an expert read.
- Timing and price. Observed pickup time of 3 to 8 weeks after publication. Programs run $2,300 to $10,000 per month over 6 months, with a baseline audit at $2,000, billed at the start and included in your 6-month program. The Quick Scan is free: up to 5 of your buying questions, with a video walkthrough within 48 hours.
The sector page lists the questions we measure for cybersecurity vendors, and the methodology page documents the protocol.
Frequently asked questions
Can a GEO agency cover MSSPs and MDR providers, not only security products?
It should. Service providers are chosen on questions about coverage, response model and compliance scope, which differ from product questions about features and integrations. Ask the agency to measure service questions as a separate family, with their own list of competitors, so a gain on product questions does not hide a gap on the questions that sell your service.
How long does it take to start appearing in ChatGPT answers?
Count in weeks. A new page must be crawled, indexed by the search providers ChatGPT draws on, then picked up in answers. Across ZivRank programs the observed pickup time is 3 to 8 weeks after publication. A correction on a third-party page that ChatGPT already cites can show sooner than a new page on your own site.
Do security comparison pages hurt relationships with other vendors?
Not if they stay factual. Compare on criteria you can source, link to each vendor’s own documentation, date the page and correct it when a product changes. Pages that disparage a competitor tend to backfire with security buyers, who check claims, and with engines that read the competitor’s pages too.
What should a security vendor fix first on its own site?
Make sure the AI search crawlers can read it: OAI-SearchBot, Claude-SearchBot, PerplexityBot and Googlebot, including in CDN and firewall rules, which security companies often set strictly. Then give each buying question a page with the answer near the top, framework versions stated plainly, and a named, qualified author.
Sources
Every figure in this article comes from the sources below, checked on November 9, 2026. Bracketed numbers in the text refer to this list.
- [1] G2 (Tim Sanders), In the Answer Economy, Don’t Win the Click — Win the Answer, April 15, 2026.
- [2] DerivateX, B2B SaaS AI citation study, May 31, 2026, updated September 22, 2026.
- [3] Reddit, Reddit Rules, checked November 9, 2026.
- [4] Federal Trade Commission, Trade Regulation Rule on the Use of Consumer Reviews and Testimonials, posted August 21, 2024.
- [5] OpenAI Help Center, Searching the web with ChatGPT, checked October 11, 2026.
- [6] SparkToro (Rand Fishkin), NEW Research: AIs are highly inconsistent when recommending brands or products, January 27, 2026.
- [7] ZivRank, US measurement: 65 buyer questions about GEO, ChatGPT, Claude, Gemini and Perplexity × 5 runs (1,300 answers) + 325 Google searches, November 8, 2026.