Leadership asks what the company is doing about AI search, or you type your own core buyer question into ChatGPT and watch a competitor get named instead of you. Either way, the question that follows is always the same: what do we do about it. If your answer is "we should invest in AI content," you will not get a budget, and you should not get one, because that is not how you make the case for AI search program funding. You make the case for an AI search program by showing where you are absent, what competitors own, what that absence is likely costing you, what a small pilot will test, and exactly what you are asking leadership to fund. This guide walks through that case step by step, from a baseline audit to a specific ask leadership can approve.
By the end you will have a baseline of your AI visibility, an AI visibility business case built on your own numbers plus outside evidence, a proposal for a bounded pilot with real success criteria, and a funding request written the way an executive actually reads one.
Step 1: Translate the AI-search problem into a business problem
Do not open a conversation with "we need an AEO tool" or "AI search is changing everything." That is a tactic looking for a budget line, and it reads that way to anyone holding the purse. If you want to pitch AEO to leadership successfully, open with the exposure instead.
Buyers may already be asking your highest-value questions inside ChatGPT, Gemini, or Perplexity before they ever land on your site. Competitors may be the ones getting named or cited when that happens. You may have no repeatable record of where you show up, which of your pages actually get cited, or which competitors are winning those spots. Content that ranks fine in traditional search can still fail to get picked up as a source in an AI answer, and your team may not have the capacity to close the gaps once you find them.
Connect that exposure to something an executive already cares about: protecting your spot in the consideration set, creating qualified demand instead of just traffic, getting more out of the content budget you already spend, reducing how much of the pipeline depends on manual work that breaks the moment someone quits, keeping the brand described accurately as AI systems summarize your category, or simply not carrying an entire channel with zero measurement on it.
One sentence should anchor the whole pitch: "We are asking for a controlled program to measure and improve our visibility for the buyer questions most connected to revenue." Everything else in this guide builds the evidence behind that sentence.
You know this step is done when you can fill in a version of this: for [buyer segment], on [priority questions], your brand appears or gets cited at [some baseline rate], while [named competitors] appear or get cited at a different rate, and that gap creates a measurable risk to consideration, demand, pipeline, or content efficiency.
Common mistake: claiming AI search will replace organic search, or that the company will lose some specific amount of revenue, without any internal data to back it up. The defensible version of this case is a measured visibility gap and a testable opportunity, not a prediction.
Step 2: Build a prompt-level visibility baseline
You cannot get budget for AEO with a handful of ChatGPT screenshots. You need a baseline, and a baseline means a representative set of prompts, checked the same way, more than once.
Build a prompt list that covers brand-specific questions, non-brand category questions, problem and use-case questions, comparison and alternative questions, decision-stage questions, questions that name your competitors, and questions from different buyer roles or industries. For a first internal audit, a narrow set works better than trying to track everything at once. Fifteen to thirty high-value prompts across awareness, consideration, and decision stages, with a handful of competitor or comparison prompts mixed in, is a reasonable starting range, though it is a starting recommendation and not an industry standard.
For every prompt, record the exact wording, the engine and date you checked it, whether your brand was mentioned or cited, which competitor got mentioned or cited instead, the exact page or domain that earned the citation, whether the description of you was accurate, and how important that prompt is to the business. Save the answers so a future check compares against the same prompt set instead of relying on memory or a new round of screenshots.
This is exactly the kind of tracking DeepSmith's AI Visibility module is built for. It checks your prompt set on a schedule and reports mention rate, citation rate, share of voice, sentiment, visibility trends, a per-platform breakdown, a competitor leaderboard, and the sources getting cited most, plus a Pages view showing which of your own pages earn citations and which prompts drive them. Discover Prompts can generate a starter set from your product, persona, and buyer-stage context if you are starting from nothing. Treat it as a way to operationalize a baseline you already decided matters, not a substitute for deciding which questions matter.
You know this step is done when you have a table or dashboard showing total prompts by funnel stage, mention rate, citation rate, share of voice against named competitors, performance by engine, which pages and domains are getting cited, the high-value prompts where you are absent or losing, and any negative or inaccurate description that creates brand risk.
Step 3: Quantify the gap and the cost of doing nothing
Once you have a baseline, turn it into a scorecard an executive can actually read. There are three kinds of gap here, and mixing them up weakens the AI visibility business case you are building.
A presence gap is when you are not mentioned at all where you should be part of the conversation. An authority gap is when you do get mentioned but a competitor or a third party gets cited as the actual source. A content gap is when you have nothing that credibly addresses the topic, or what you have is too weak, outdated, or generic to serve as a source. Rank these by business importance rather than volume: a missing citation on a low-value educational question is not the same problem as a competitor citation on a high-intent comparison question.
A simple prioritization score can combine prompt importance, funnel stage, competitor advantage, the size of the visibility gap, commercial relevance, content feasibility, and any brand or accuracy risk. Score each factor one through five for a quick workshop exercise. It is worth saying out loud that this score is an internal prioritization tool, not something you compare against another company.
Alongside the visibility gap, put a number on what content already costs you: fully burdened hours and costs for research, brief creation, drafting, editing and fact checking, SEO review, internal linking, image production, formatting and publishing, and repurposing for other channels. This gives you two costs to show leadership at once, the cost of being invisible and the cost of the process you already run.
A visibility gap is your target share of voice minus your current share of voice. A competitor citation gap is the competitor's citations on your priority prompts minus your own citations on those same prompts. A content coverage gap is the priority topics or funnel stages your competitors cover minus what you cover.
You know this step is done when you have a ranked list of the five to ten most commercially important gaps, the competitor evidence behind each one, and an internal estimate of what it would cost to close them.
Pro tip: do not treat every competitor page as a threat just because it exists. A competitor publishing more on a topic does not prove the topic has demand or that their page is actually better. Use competitor gaps as hypotheses worth testing against buyer importance and content quality, not as an automatic to-do list.
Step 4: Model the referral and pipeline upside conservatively
This is the step where most pitches to get budget for AEO either overreach or underreach. Overreach looks like promising a specific revenue number from a vendor study that has nothing to do with your business. Underreach looks like refusing to model any upside because you cannot prove it with certainty yet. Neither gets funded.
Use outside research to explain why this channel deserves measurement, then use your own numbers to size the opportunity. Adobe's analysis of AI-driven referral traffic found that U.S. web traffic from AI referrals increased more than tenfold between July 2024 and February 2025, with retail traffic up twelvefold, travel up seventeenfold, and banking-site visits up twelvefold in the same window. By its January 2026 look at the 2025 holiday season, Adobe found retail traffic from generative AI tools had grown 693 percent year over year, with AI referrals converting 31 percent more than other traffic during that period. Those numbers come from consumer retail, travel, and banking categories over specific windows, not a universal AEO benchmark. The honest way to use them is as evidence that this channel is growing and worth measuring, not as a stand-in for your own conversion rate.
Build three scenarios instead of one number. A conservative scenario uses only direct AI-referred visits and the lowest defensible conversion assumptions. A base scenario adds a clearly labeled assisted-conversion estimate. An upside scenario reflects a broader visibility improvement, still subject to whatever the pilot proves. Use words like "addressable opportunity," "scenario," and "hypothesis to validate" rather than promising a result.
Where you can, run the math with your own funnel: incremental qualified AI-referred visits times your visitor-to-lead rate, times lead-to-opportunity rate, times opportunity win rate, times average contract value, gives you a modeled pipeline number, which is the core arithmetic you need to make the case for AI search program funding. Keep assisted impact as its own separate calculation. If your company thinks in margin rather than revenue, model it that way, and subtract the program's cost before you talk about return at all.
Set up measurement before you launch. Decide what counts as a conversion (a demo request, a form fill, a trial, a qualified lead), agree on a consistent way to tag AI referral traffic in analytics, connect that website event to your CRM stages where you can, and pick an attribution model up front so you are not arguing about credit later. A citation is not automatically a website visit. Some AI answers are generated from what a model already learned rather than a live fetch of your page, so a citation can build authority without ever showing up as a session in your analytics, which is why referral traffic alone is only part of the case.
You know this step is done when you can show a clear definition for each conversion, a direct-referral scenario, an assisted-influence scenario, the assumptions behind each one, and a plain statement that these are scenarios, not promises.
Step 5: Choose a pilot that can prove or disprove the case
Do not propose an enterprise-wide rollout on the first ask. Propose one commercially meaningful content cluster that can prove or disprove the case within a defined window.
A strong pilot has a specific buyer audience, a limited prompt set, a clear funnel stage, real competitors you can benchmark against, a known content gap or a set of underperforming pages, a publishing pace your team can hit, a measurable conversion or pipeline path, and a set date when someone decides whether to continue, adjust, or stop. The sequence looks like this: capture the baseline prompt answers and cited sources, pick the highest-priority gaps, decide for each existing page whether to improve it, consolidate it, or write something new, produce a small set of genuinely useful pages, publish and give the work time to get indexed and picked up by AI systems, recheck the same prompts against the baseline, and hold the scale decision using criteria you agreed on before you started. A 90-day window is a practical length for a first pilot because it leaves room for setup, production, and early measurement, though it is a planning assumption rather than a guarantee about how fast citations or rankings move.
Set success criteria across four areas instead of one number. Instrumentation: every pilot prompt documented, assigned a funnel stage, and given an initial answer and competitor comparison, every pilot page with an owner and a conversion path, and analytics definitions agreed before publishing. Visibility: improved citation presence on priority prompts, improved share of voice against named competitors, more accurate brand descriptions, and evidence that new or improved pages get cited. Content production: pages published on schedule, review time and rework tracked, content passing brand and factual review, and a clear reason each page exists. Business: measurable AI-referred sessions where source links are clicked, AI-referred visitors compared with other channels on engagement and conversion quality, AI-influenced leads identified where the data supports it, and cost per approved page compared against your current baseline.
DeepSmith's Content Map and Opportunity Agents are useful here for turning the competitor and topic gaps from step three into specific content decisions, each carrying the evidence that justifies it, so the pilot's content list is not a guess.
Avoid grading the pilot on a single pass or fail number. Citation visibility can improve well before referral volume is large enough to move pipeline, and traffic can rise without the content getting more accurate or useful. Both are real outcomes worth reporting honestly.
Step 6: Answer the trust and governance objections before someone raises them
The most common objection you will hear when you pitch AEO to leadership is that it will damage the brand. Do not argue with that concern. Agree with it, then show the controls that address it.
Google's own guidance for ranking well is unique, useful, non-commodity content with a clear point of view and first-hand expertise, organized for people rather than search engines. It warns against publishing large amounts of content across many topics hoping something sticks, leaning on heavy automation without adding real value, mostly summarizing other sources without adding anything, writing content mainly to attract search traffic, and using AI to manipulate rankings. Meeting technical requirements does not guarantee crawling, indexing, or getting served in an answer, so the pilot should be built to measure real visibility, not assume it. Match that with your own internal controls: a documented boundary on what claims you will and will not make, human review for accuracy and judgment, a named subject-matter reviewer for high-risk topics, a requirement for real analysis rather than generic summary, a check for unsupported claims or copied competitor structure, a decision on when AI assistance gets disclosed, and a stop rule for anything that does not clear the editorial bar.
DeepSmith's Deep IQ is a useful example of what structured context looks like in practice. It stores your company's positioning, the claims you are allowed to make, product details, buyer personas, brand voice, visual guidelines, and trusted sources, and every draft the platform produces is grounded in that context, which helps keep output on-brand and accurate at higher volume. No tool replaces human review or guarantees brand safety, and the pitch should not claim it does.
A few other objections are worth having answers ready for. If someone says you already invest in SEO, the answer is that this is not a replacement: foundational SEO, crawlability, indexing, and page experience still matter to Google's AI features, and the pilot tests whether your content shows up in AI answers as well as it shows up in conventional search. If someone says there are not enough clicks to justify this, the answer is that clicks are only one signal; citations, mentions, competitor share, and assisted influence matter too, and Pew Research found that people click a link inside a Google AI summary in only about 1 percent of visits where a summary appears, which is why direct clicks cannot be the only thing you measure. If someone says you cannot prove ROI in advance, the honest answer is that you are right, which is exactly why you are asking for a bounded pilot with pre-agreed criteria and a decision date rather than an open-ended budget. If someone worries this is another tool the team will not use, build adoption into the pilot's success criteria: time to baseline, time to pick an opportunity, time to produce an approved page, and whether the team follows the workflow. If someone asks why not just hire an agency, compare both options on the same terms: time to launch, internal effort, quality controls, knowledge transfer, content volume, and total cost, using your real fully burdened numbers rather than assuming either option is cheaper.
Step 7: Package the executive funding ask
This is where executive buy-in AEO work turns into paper. Put the whole case on one page. State the exact decision you are asking for. Lay out the business problem in terms of buyer prompts, the visibility gap, and the competitor evidence behind it. Show the baseline: current mention rate, citation rate, share of voice, and the priority gaps you found. Explain why now, using outside evidence on AI referral growth, clearly labeled by market and date. Describe the pilot scope: prompt count, content cluster, which engines you are tracking, how many pages, how long it runs, and the decision gate at the end. State plainly what the pilot will prove or disprove. List success criteria across visibility, content production, traffic, and pipeline. Break down the investment across software, content production, internal review time, and analytics. Show your three scenarios with their assumptions laid out. Name the real risks (generic content, inaccurate claims, low early traffic, slow indexing, attribution ambiguity, low adoption, competitor response) and how you plan to manage each. Say what happens if the pilot meets, partly meets, or misses its criteria. Finish with the actual ask: the budget amount, who needs to approve it, the start date, and the date you will come back with results.
This is the format that works when you pitch AEO to leadership: lead with outcomes rather than tactics, back the case with your own data plus competitor evidence, tie KPIs to something the business already cares about, move in defined phases, and ask for something specific instead of an open-ended line item. Show the cost of your current process alongside the cost of the pilot, and compare both against doing nothing, adding headcount, or hiring an agency, using the same full-cost basis for each.
Product mentions belong where the product genuinely does the work: the baseline in step two, turning gaps into content decisions in steps three and five, and executing the pilot in steps five and seven. DeepSmith's Content Studio takes an idea from New Ideas through Planned Content to a finished, brand-grounded article, researched and linked, with a cover image and publishing metadata built in, and Autowrite can produce and land a scheduled piece in Produced Content without anyone touching it. Pricing runs from Pro at 99 dollars a month tracking ChatGPT, up through Grow, Scale, and a custom Enterprise tier covering all ten tracked engines, with a 7-day free trial and no long-term contract, so a pilot does not require a large upfront commitment to test.
What to do next
Start with the baseline. Pick the buyer prompts that actually matter to revenue, run them, and write down exactly where you stand today. Turn that baseline and the competitor evidence into the one-page proposal outlined above, and take it to leadership as a bounded pilot with a real decision date, not an open-ended request to "do more AI content." That is the whole loop behind executive buy-in AEO programs need: baseline, gap, scenario, pilot, ask. If you want to see what a real baseline and a real draft look like before you build the proposal, DeepSmith's free trial gives you actual visibility data and actual output from your own site to bring into the conversation.
