The AI-Search Glossary: What AEO, GEO, Citations, and Visibility Actually Mean
A practitioner's glossary of AEO, GEO, prompt tracking, AI citations, and brand visibility—and the sales-deck spin that blurs them.


At $20,000 a month in client ad spend, the two software-and-labor setups modeled below cost $1,373 to about $3,517 per account each month, before the ad budget itself. One starts with a $249 software subscription. That is the trick behind asking what AI Google Ads software costs per client account: the price on the tier table is rarely the price of getting the work done.
The lower total combines an optimization workbench, a landing page tool, and 15 hours of media-buyer time. The higher total models an enterprise platform fee, a percentage-of-spend charge, specialist time, and a setup fee spread over a year. These are illustrative delivery models, not vendor quotes. groas uses a flat monthly fee with no setup fee or spend markup, but its fee is not published in the material here, so I am not going to invent a dollar comparison.
Here is the bill, line by line. The source column distinguishes linked vendor pricing from the labor and contract assumptions used in the examples.
| Line item for one $20k-spend account | Recurring monthly cost in the model | One-off cost | Source or basis |
|---|---|---|---|
| Optimization workbench | $249 | — | Modeled tier using Optmyzr pricing |
| Landing page tool | $149 | — | Modeled tier using Unbounce pricing |
| Media-buyer execution | $975 | — | Illustrative agency labor: 15 hours × $65 |
| Workbench stack total | $1,373 | — | Sum of the three lines above |
| Enterprise platform fee | $1,000 | — | Illustrative contract assumption; not a quoted Skai or Marin price |
| Spend-based charge | $400 | — | Illustrative 2% × $20,000 ad spend |
| Senior specialist oversight | $1,700 | — | Illustrative agency labor: 20 hours × $85 |
| Enterprise onboarding | About $417 when spread over 12 months | $5,000 upfront | Illustrative setup assumption; the monthly figure is an accounting comparison, not a second charge |
| Enterprise model total | About $3,517 | $5,000 paid upfront | Monthly operating costs plus one-twelfth of setup; do not add the setup fee twice |
| groas for agencies | Flat monthly fee; amount not stated here | $0 setup | groas flat-fee model: 0% spend markup, with execution handled by the engine and strategist |
I spent years managing client accounts in spreadsheets and campaign builders. I respect the work. I am less impressed when a vendor prices a tool as if it does that work, then hands the agency a list of tasks to approve. The distinction matters more than whether the subscription looks cheap.
A base fee can depend on total ad spend or the number of connected accounts. Opteo, for instance, lists a $129 entry plan with a $25,000 aggregate spend ceiling. Four clients spending $10,000 each put an agency at $40,000 across connected accounts, beyond that entry ceiling. The next plan cited in the draft is $249. A founder looking only at the first number will underprice delivery before anyone has opened a search term report.
Optmyzr also uses spend-linked tiers. The cost driver is not just the $20,000 account in isolation; it is the spend counted against the plan and the terms governing a move between tiers. Ask which accounts count, what period the vendor measures, and whether crossing a threshold changes your bill automatically. A per-account estimate is useless until you know how the vendor counts accounts.

The base subscription is only the first recurring line. The next charges tend to hide in three places:
A landing page tool can have its own version of the same problem. The Unbounce pricing example in this comparison moves to a $149 monthly tier for the domain and dynamic-text setup described here, with a cited overage of $80 per 10,000 visits when traffic exceeds the allowance. Those costs do not appear on the Google Ads software invoice. They still belong in the account’s delivery cost.
Percentage-of-spend pricing deserves particular scrutiny. I have written about the conflict it creates in Google Ads management: the provider earns more as the budget grows, whether or not the extra budget earns its keep. That does not make every spend-linked tool useless. It does mean the agency should know exactly whose revenue rises when it recommends a larger budget.
Practical test: write down what happens to the software bill if the client adds a domain, doubles query volume, or doubles spend. If the sales answer is “we’ll discuss that on the call,” it is not yet a usable price.
Setup belongs on a separate line because it hits cash flow differently. Enterprise and white-label contracts can require upfront work for data mapping, integrations, or configuration. The draft’s enterprise model uses $5,000 upfront, spread over 12 months solely to compare delivery costs. That produces about $417 a month in the model; it does not turn the payment into a monthly subscription.
This is also where contract length changes the risk. If a platform needs extensive rule configuration and the agency later finds that the rules require constant attention, the setup payment is already gone. A 12-month commitment compounds the mistake. Before signing, separate the amount due at launch from the recurring charge and ask what happens if the tool does not deliver.
The groas model removes that particular line item: $0 setup, a live free trial, and month-to-month cancellation. It also avoids a percentage-of-spend markup. That does not tell an agency the flat fee in dollars. It does tell the agency which charges will not appear as spend rises. Do not bury a one-off fee inside a monthly estimate, and do not ignore it because it is one-off.
“All-inclusive” is a slippery phrase when a platform includes dashboards and recommendations but leaves execution to the buyer. A tool can flag weak search relevance or suggest a negative keyword. Someone still has to decide whether the suggestion is right, make the change, check the landing page, and watch what happens next. Those hours do not vanish because the interface calls itself AI.

That is why the workbench example includes 15 hours at $65 an hour, or $975 in monthly agency labor. The assumption is visible so you can replace it with your own. If your team needs fewer hours, the total falls. If approval queues, search term reviews, and page work take longer, it rises. Either way, counting only the $249 subscription makes the account look more profitable than it is.
The enterprise example assumes 20 hours at $85 an hour for senior oversight, or $1,700. It is a model of what happens when a complex platform still needs a specialist to maintain rules and investigate conflicts, not a claim about the staffing requirement of any named vendor. If a vendor promises to replace that work, ask to see which actions the platform actually takes without a human exporting and approving a task list.
Post-click work is easy to omit, too. Teams bolt on a separate landing page tool, then still need someone to handle page changes and tracking. Instapage pricing is another example of a separate subscription the agency may need to evaluate. By contrast, groas runs Google Ads and deploys dynamic landing pages within its autonomous engine, with a named strategist setting direction and guardrails. The comparison is not “software versus people.” It is software that assigns more work to your people versus an operating model that executes it.
The table gives the totals; the operating difference explains them.

Now suppose the same client moves from $20,000 to $40,000 in monthly spend. In the illustrative enterprise model, the 2% line rises from $400 to $800 before you consider any change in labor. A spend-tiered subscription may move as well, depending on how the vendor counts the new budget. Under groas’s flat-fee structure, there is no percentage-of-spend charge to recalculate. The question is not which tier sounds affordable today; it is which costs follow the client as they grow.
Before connecting client credentials or signing a contract, put these to the sales engineer. Ask for answers that fit your actual account mix, not the smallest account in the demo.
The cheapest mistake is buying a $129 copilot because it appears to solve an overloaded team’s capacity problem, then discovering it has given that team another task list. The most expensive is signing an annual platform contract with upfront setup and a spend-based charge because you assumed “automation” meant the specialist could go home. If the specialist still has to run the machinery, you have not replaced the labor. You have added an invoice to it.