How much should a SaaS company budget for AEO at $10M, $20M and $30M ARR?

A B2B SaaS company should budget about $42,000 a year for AEO at $10M ARR, $42,000 to $66,000 at $20M ARR, and about $66,000 at $30M ARR. That is roughly 2 to 4 percent of the median marketing budget at each size. The figures are Salienz's published rates for a full program, the work and the monthly measurement together, set against SaaS Capital's 2026 survey of more than 1,000 private B2B SaaS companies. 1 4

Those numbers sit well below what AI assistants tell you to spend. Asked the same question, their typical answer was $5,000 to $15,000 a month at $10M ARR, which is $60,000 to $180,000 a year, or 6 to 18 percent of a median marketing budget. Trace the larger figure back through the answers and it rests on very little. 2

AI assistants say $5,000 to $15,000 a month for a $10M ARR company, and almost none of them show where that comes from. Salienz tracked the question "How much should a $10M ARR SaaS company budget for AEO each month?" once a day from 2 to 8 October 2026 on ChatGPT, Claude, Perplexity, Microsoft Copilot, Gemini, Google AI Mode and Google AI Overviews, and read all 47 answers. Four were off topic: three Perplexity answers and one Copilot answer were about running a $10M ARR company or paying salespeople, not AEO. The other 43 each gave a monthly range. 2

The typical answer builds its range from three stacked percentages: marketing as a share of revenue, organic and content as a share of marketing, and AEO as a share of that. Gemini's answer on 2 October put marketing at 12 to 18 percent of revenue for a $10M ARR company; SaaS Capital's survey puts the median at 10 percent for that size. The two later steps carried no source at all, and the pages the assistants cited most often were agency blogs selling the service. 1 2

The advice inside the answers has the same problem. 37 of the 43 recommended schema markup or an llms.txt file as part of the budget. Salienz's crawl of 99 cybersecurity vendors' websites found no link between either and how often AI assistants recommended the vendor: vendors with an llms.txt file were named in a median of 0.8 percent of answers, the same as vendors without one. 3 Money spent there does not show up in the answers.

A private B2B SaaS company spends a median of 10 percent of ARR on marketing at $10M to $20M ARR, and 9 percent above $20M, according to SaaS Capital's 2026 survey of more than 1,000 companies, run in March 2026. Across all sizes the median is 8 percent, because companies under $3M spend 5 percent. Funding matters as much as size: equity-backed companies spend a median of 10 percent of ARR on marketing, and bootstrapped companies 5 percent. 1

Larger companies spend a similar share. Gartner's 2026 CMO Spend Survey of 401 marketing leaders, most at companies with more than $1 billion in revenue, put marketing budgets at 7.8 percent of revenue, up from 7.7 percent in 2025. 5 So at $10M ARR the working figure is about $1.0M a year, at $20M about $1.8M to $2.0M, and at $30M about $2.7M. A $10M ARR company spending the bootstrapped median of 5 percent has about $500,000, which makes the same AEO program about 8 percent of its marketing budget.

Measuring AEO properly costs $455 to $800 a month on published software plans, and only one plan covers every major assistant. Proper means a fixed set of about 100 buyer questions, checked every day on ChatGPT, Claude, Perplexity, Microsoft Copilot and Google's three AI surfaces. Fewer questions cannot tell a change from noise: in Salienz's Audit Size Test, two readings of an unchanged brand built on 15 questions differed by 2 points or more in up to 44 percent of draws, against 7 percent or less for 90 questions. 6

Salienz priced that standard on every AI visibility tool with a public pricing page, using each vendor's own allowances, on 18 October 2026. One hundred questions a day on seven surfaces is about 21,000 answers a month. 7

At $10M ARR, $455 to $800 a month is $5,460 to $9,600 a year, about 0.5 to 1 percent of the marketing budget. It is the floor, and it only tells you where you stand. None of these plans places your name on the pages AI assistants read.

Most of an AEO budget should buy presence on pages other people publish, because that is what AI assistants cite when they recommend vendors. In Salienz's AI Security Shortlist Study of 7,385 answers from seven AI assistants, roundups and comparison pages took 40 percent of the citations in the answers that recommended vendors, and the vendors' own websites took 22 percent. 3 The details are in how cybersecurity companies get recommended by AI models.

• Placements in the roundups and comparison pages that already rank for your category's buyer questions.
• Pages on your own site that answer a narrow question your product wins, which is how smaller vendors got named in that study.
• Guest posts, digital PR and community answers under real names, in the places your buyers read.
• A fixed question set re-run every month, counting mentions of your name and citations of your site separately.

What it should not buy is a schema project or an llms.txt file sold as the strategy. Both are cheap to add for other reasons, and neither showed any link to being recommended. What each part of the work costs is broken down in what an AI visibility audit costs.

As ARR grows, the program has more ground to cover, while each piece of work costs the same. A company at $10M ARR usually sells one product into one buyer conversation, so 100 questions and 40 targets cover it. Past $20M there are usually several products, more competitors writing comparison pages about you, and older third-party pages describing you wrongly, which doubles the questions worth tracking. 4

At $20M the product decides it. One product in one category is still a 100-question job. Two product lines, or a category where rivals already publish pages about leaving you, is a 200-question job. Above $30M the question becomes how many product categories need their own question set, since each one is a separate buyer conversation.

Before any retainer there are two cheaper steps. Salienz runs a free audit, 15 buyer questions asked three ways on five engines, for the brands it accepts, and an $800 audit of 100 questions across five engines with a 90-day roadmap, delivered in seven business days and credited against the first month. 4

Spending more on AEO makes sense when there are more products or markets to cover. It does not buy faster results on the same questions. Enterprises already spend far more: in Conductor's survey of more than 250 enterprise digital leaders, published in January 2026, companies put an average of 12 percent of their digital marketing budgets into AEO and GEO in 2025, and 94 percent planned to increase it in 2026. 14 Conductor sells AEO software, which is worth knowing when reading that number.

The case for spending something is no longer in doubt. In G2's survey of 1,076 B2B software buyers in March 2026, 51 percent said they now start software research in an AI chatbot more often than in Google. 15 The case for spending a lot is weaker. Citations come and go: in the Audit Size Test, 86 percent of the places a brand showed up appeared on some days of the week and not others. 6 A bigger budget on the same 100 questions mostly buys more placements chasing the same few places in an answer.

You know the budget is working when the same fixed set of buyer questions, re-run monthly, names you in more answers than it did at the start. Count mentions of your name and citations of your site as two numbers: one company in the AI Security Shortlist Study had its website cited in 199 answers and was named in 9. 3 Judge the program over its full term rather than month to month, because single readings move with noise. How Salienz turns that into one number is set out in the Salienz Score, and why analytics show so little of the effect in why GA4 shows almost no ChatGPT traffic.