Starting AEO and GEO does not require a big budget. It requires a baseline, a short list of fixes, and a way to measure change. Here is the first 30 days, in order.
Ask ChatGPT, Gemini, Claude, and Perplexity the questions your buyers actually ask, phrased the way real people phrase them. Note who gets recommended and who gets ignored. The free automates this in about a minute. If you are new to the terms, AEO vs GEO explains the difference in plain language.
Assistants recommend brands they can read and verify. Answer your buyers questions directly on your pages, in the first paragraph, not the fifth. Add FAQ and organization schema. Make sure facts about your business are consistent everywhere they appear. Then go where assistants source opinions: reviews, comparisons, and community threads in your category. Our guide on how to get recommended by AI covers each input.
Re-run your baseline questions and compare. Model refreshes take weeks, so expect early movement, not victory. Then decide how to scale: keep it manual, or let a platform run the loop continuously. That decision is covered honestly in do I need software for GEO and AEO.
Your entire program lives or dies on the quality of the question basket you measure, so build it carefully in week one. Pull five questions from how customers describe their problem before they know your category exists, the situational phrasings, five from active evaluation (comparisons, alternatives, is X worth it), and five from your sales team's actual call notes, the objections and use cases that come up when money is on the table. Phrase them the way people talk, not the way marketers write keywords, because assistants respond to conversational context. Fifteen questions across four assistants is sixty answers a month: enough signal to see patterns, small enough to actually maintain.
Then freeze the basket. The temptation to swap questions every month destroys your trend line, and the trend line is the whole point. Add questions when you enter new markets; retire them only when a product line dies. If a question never names any vendor at all, keep it anyway, unclaimed moments are where early movers win cheapest.
Four mistakes eat most first months. Publishing before baselining, so you can never prove what changed. Optimizing your homepage instead of answer pages, assistants cite pages that answer one question well, not brochures. Writing for keywords instead of moments, covered at length in what you are being taught about AI visibility is wrong. And declaring victory or defeat at week two, before a single model refresh has had time to read your work; the honest timelines are in why it is so hard to show up in AI.
One more that costs teams quietly: doing all of this in a spreadsheet nobody owns. Assign the loop to a person with a monthly calendar slot, or hand it to software. The programs that die do not die of difficulty, they die of nobody's-job.
The first month ends with a decision, so make it deliberately. Put one number in front of whoever owns budget: moments where you appeared at day zero versus day thirty, next to the hours the month actually consumed. Then pick one of three paths and write it down. Sustain: the scripted monthly hour keeps the basket measured and ships two or three fixes a cycle, right when early movement was visible and capacity exists. Scale: the software path from the build-or-buy guide, right when the manual loop is winning but starving for hours. Pause: legitimate if the baseline showed your category barely produces brand recommendations yet, in which case calendar a re-baseline for next quarter, because that answer has a short shelf life. The one indefensible outcome is drift, a strong first month that nobody formally continued, which is how most AEO efforts actually end.
The first month is very doable manually: baseline, on-page answers, schema, and consistent facts. The hard part is doing it continuously across four assistants and dozens of buyer moments, which is where software earns its keep.
Expect weeks, not days. Assistants refresh their reading of the web on their own cycles. Measure monthly against the same baseline questions.
Optimizing only for best-X keywords. People describe situations to AI, not keywords, and assistants often decide who to recommend upstream of any search phrase.
Done manually: roughly a day of setup and a half day per month of maintenance, plus whatever content fixes you ship. Done with software: Aethon starts at $199 per month with the loop automated. Either way the baseline itself is free with the GEO Grader.
Mostly yes. Question baskets, direct answers, and fact consistency need no engineering. Schema is the one piece where a plugin or a developer hour helps, and it is a one-time setup on most sites.
Baseline all four from day one, because the gaps between them are diagnostic. Fix systemically, the shared input layer first, then handle assistant-specific gaps once the shared work lands.
Smaller and steadier than month one: re-measure the basket, ship two or three fixes aimed at the nearest-to-flipping moments, and log everything. Two hours of maintenance beats another heroic sprint.
Whoever owns organic growth, with the calendar authority to protect the monthly slot. Ownership by committee is the most common way the loop dies.
In a hurry, or on a tight budget? Two focused guides split this out: how to improve your AEO fast covers the quick wins that move AI answers within days, and how to fix AEO and GEO without spending a lot is the free-first playbook.