$17,988 is the number I would put ahead of any SEO proposal. That is twelve months at groas’s published $1,499 monthly fee, and the useful buying question is how many genuinely additional customers would repay it—not how many pages or AI mentions arrive along the way.

That is service fees alone, before internal time or any ad spend. I used to tell clients to weigh the fee against pages published and keywords moved. I was wrong. Pages can help win customers. They cannot repay an invoice by themselves.

The missing input is contribution per customer: the money left after the variable costs of delivering the sale. The break-even worksheet expresses the calculation plainly: revenue per win multiplied by contribution margin gives contribution per win; total investment divided by that contribution gives break-even wins.

Free tool: the ROAS Calculator works out your return on ad spend, your break-even ROAS and the budget a revenue goal takes.

I like that order. Start with what a customer leaves behind, then work out how many more customers the service must produce. The retainer stops being an abstract marketing expense and becomes a hurdle you can compare with your sales capacity.

SEO and AI search share work, not guaranteed outcomes

Traditional SEO helps you get found in Google. AI search optimization tries to get you cited when ChatGPT, Perplexity, or Google AI Overviews answers a buying question. Much of the underlying work overlaps:

  • Technical fixes that help search systems crawl your pages.
  • Buyer-focused pages that answer real questions in full.
  • Credible third-party mentions that give those systems a reason to trust you.

The fee is not unusual for managed work. Published small-business SEO pricing puts ongoing work around $1,500 to $3,000 a month, while basic small-business generative engine work starts around $1,500 to $5,000 a month. A broader published range runs from $390 to $10,000 a month, spanning monitoring-heavy packages through content and digital PR.

Those ranges describe different scopes, not equivalent products. They help place a proposal in the market; they do not establish whether it will pay for itself.

No tier buys a guaranteed AI mention. As the service overview explains, generated answers depend on retrieved sources. A provider promising a fixed position in ChatGPT is selling certainty it does not own. Judge the work it commits to doing, then test whether the economics justify buying it.

Contribution turns the annual fee into a customer target

Revenue is not the amount available to repay marketing costs. If a job brings in $2,000 but costs $1,500 to deliver, it leaves $500 in contribution. Labor, materials, shipping, payment fees: include the costs that rise with each sale.

For repeat business, use revenue and delivery costs from the same period. Keep currency and tax treatment consistent, as the worksheet requires. A single job and a year of subscription revenue are different measurement windows. Pick the window that matches how you actually collect the money, rather than whichever makes the proposal look easiest to approve.

Here is the fee-only model, using the published groas price and the worksheet’s break-even calculation:

Illustrative contribution per additional customerAnnual fee divided by contributionAdditional customers needed
$100$17,988 ÷ $100 = 179.9180 customers
$250$17,988 ÷ $250 = 71.972 customers
$500$17,988 ÷ $500 = 35.936 customers

These are illustrative inputs and calculated thresholds, not customer results reported by the worksheet or by groas. The divisions are shown to a decimal place; the customer requirements round up. You cannot collect a fraction of a customer.

This table tells you what must happen, not what will happen. The worksheet does not predict rankings, leads, or sales. It supplies an economic threshold, and this version covers the service fee only. Internal costs still sit outside it.

At the low-contribution end, 180 additional customers a year means 15 a month. Higher contribution reduces the volume required, but each missed customer leaves a larger gap. And customers are not enquiries: at an illustrative one-in-four qualified-enquiry close rate, 72 customers require 288 qualified enquiries.

That second division is where back-of-napkin math quietly dies. Check the sales requirement, not just the customer total.

A slow start compresses the repayment window

An annual target can look manageable until you ask when the customers arrive.

A published SEO timeline puts measurable results at roughly 3 to 6 months for an established site and 6 to 12 months for a new domain. Those are planning ranges, not a schedule for your site—and measurable results do not necessarily mean collected customer contribution. Crawling, indexing, visibility, and clicks can move before sales do.

Early diagnostics matter. They help show whether the work is moving. They pay no bills.

The following scenarios hold the annual fee constant and assume no contribution during the opening stretch. They are calculations, not forecasts:

Illustrative delayMonths left to collect contributionMonthly contribution needed to cover the annual fee
First 3 months contribute nothing9 months$17,988 ÷ 9 ≈ $1,999
First 6 months contribute nothing6 months$17,988 ÷ 6 = $2,998

At the illustrative $250 contribution per customer, those scenarios require about 8 or 12 additional paying customers per collection month, respectively.

The same annual fee creates a harder monthly target when returns arrive late. This is the trap in a neat spreadsheet with equal monthly columns: the spreadsheet allows an even spread. Search does not promise one. Make the delay visible before deciding the target looks comfortable.

Count additional customers, not every customer search touches

Only genuinely additional customer contribution belongs on the repayment side of the ledger. Qualified enquiries help estimate how you might get there; they are not repayment themselves.

Genuinely additional means you would not have won the customer without this work. The break-even worksheet warns against crediting every organic enquiry to new SEO activity, because existing customers, referrals, and paid campaigns also influence sales.

A customer who searches your business name after a friend’s recommendation is not fully earned by SEO. Neither is a ChatGPT user who also clicked your ad last week. Attribution has limits. If you cannot state those limits in a sentence, I would not treat the total as settled evidence.

Ledger comparing the service fee with contribution from additional customers

Rankings, AI mentions, and published pages belong upstream. The distinction between revenue evidence and visibility metrics matters here: traffic, impressions, rankings, and authority scores can improve without producing the business return you need. Bookings, revenue, and channel mix bring the analysis closer to the buying decision.

Use visibility to diagnose progress, not to repay the fee on paper. Keep the work log and the contribution model connected, but do not pretend they are interchangeable. And never count the same customer again because Google and ChatGPT both appeared somewhere in the journey. Multiple surfaces do not create multiple customers.

During the wait, demand evidence of execution

While the economics remain uncertain, execution is what you can inspect. That means pages written and shipped, technical issues fixed, and credible third-party mentions earned—not merely a dashboard showing that those things would be useful.

Published descriptions of managed generative engine services distinguish specific work and transparent tracking on fixed prompts from guaranteed positions. That is the distinction I would carry into a proposal review.

My walk-away signs are straightforward:

  • A guaranteed position in ChatGPT.
  • A score-only report with no pages or fixes attached.
  • No baseline recorded before work starts.

A report should read like a work log, not a trophy case. What was published? Which technical issue was fixed? Which mentions were earned? How did visibility change on the same set of buyer prompts? A changing prompt set makes a reassuring chart easier to produce and a useful comparison harder to make.

Stopwatch on SEO reports illustrating the wait before revenue arrives

groas is built around continuous execution across paid and organic search: specialized models do the ongoing work, while a named human strategist sets direction and stays accountable for the outcome. Actions are logged with reasoning, so there is something concrete to inspect while the contribution model is still waiting for evidence.

That is the operating model I would choose over a stack of recommendations left for the owner to implement. It does not make a citation or ranking guaranteed. It makes responsibility visible. If a provider cannot show what changed, another monthly visibility score is not much comfort.

Buy only when margin, timing, and capacity agree

I would buy when the following conditions line up—not when the proposal has the longest deliverables list.

  • The contribution math clears with room to spare. At $500 contribution per job, the annual fee needs 36 additional customers, or 3 a month across a full year. That is a target a sales operation can assess. Whether it is plausible depends on your business, not how modest it looks in a table.
  • The business can carry the delay. The $1,499 monthly fee still leaves the account while crawling, fixes, and drafts are underway. If the delayed-return scenario would strain payroll, the problem is timing. An attractive annual total does not solve it.
  • The scope buys implementation. For a busy owner, done for you must mean pages shipped, site issues fixed rather than listed in a PDF, and earned mentions tracked against a consistent baseline. Monitoring with a content garnish is still mostly monitoring.

I would wait when the arithmetic demands customers the business cannot serve. The $100 contribution example needs 180 extra customers a year, averaging 15 a month. If crews, calendars, or the service area cannot support that volume, the test fails before it starts. A marketing forecast does not create delivery capacity.

I would also wait if measurement cannot distinguish new search-driven business from repeat purchases, referrals, and paid clicks. Otherwise, the report can claim credit for work the business had already won. More activity does not repair that baseline.

The question is not whether SEO and AI search can produce useful visibility. It is whether enough of that visibility can become additional, profitable business on a calendar you can afford.

Put the decision in a sheet, not a sales deck

To run the break-even test, start with these inputs:

  • Revenue per typical win.
  • Variable cost per win.
  • Close rate on qualified enquiries.
  • Months before additional customer contribution starts arriving.

Then follow the worksheet’s calculation:

  1. Subtract variable cost from revenue to get contribution per win.
  2. Divide the annual service fee of $17,988 by that contribution and round up to whole customers.
  3. Divide the customer requirement by your qualified-enquiry close rate.
  4. Spread those requirements across the months in which you expect to collect contribution, not automatically across the full year.

Compare the monthly requirement with jobs you actually completed last quarter, not names sitting in the CRM. Keep the limits visible: this is a fee-recovery model, not a prediction of search performance or a complete accounting of every internal cost. If the alternative is hiring, put the full cost of an in-house hire beside the service fee before deciding which operating model to test.

Buy the test only if the required customers fit your margins, capacity, cash timing, and measurement. If they do, groas offers the execution model I would want behind that investment: continuous work with a named strategist who owns the result, rather than suggestions waiting for someone to implement them at midnight.

If the model works only by treating every enquiry as new, every mention as a sale, or the opening month as productive as the last, keep the money. Fix measurement first. Then buy the test you can prove.