← Back to the check

How the income at risk check works

This page shows every formula, every published input with its link and date, the three scenarios, the cap, the spend shares and a full worked example. You can read it without running the check, and we encourage you to check the maths yourself.

Run the check

What the check estimates

  1. Your AEO, GEO and SEO readiness, as a simple count of checks passed.
  2. An estimate range of the income at risk when AI search doesn't cite or name you.
  3. Where your current marketing spend has no published AEO/GEO benchmark.
  4. How much of what you already spend you could re-scope to AEO and GEO each month.

All amounts are in AUD, ex GST. Every figure is an estimate built from published research, and every benchmark is shown below with its source. Where no published figure exists, the check says "no published benchmark" and shows no number. Apart from those, the numbers come from your own answers, arithmetic on them, and AISearch Global's prices from our price list of 14 Nov 2026.

The formulas, step by step

  1. Your new-customer income each month. Multiply new customers per month by your average first sale. If your figures include GST, divide by 1.1 (GST is 10%, ATO (current, S37)).
    New-customer income = customers × average sale ÷ GST divisor
  2. The part that came from Google. Multiply by the share of new customers who found you on Google. You enter this yourself: there's no published figure for it.
    Google income = new-customer income × % found on Google
  3. SEO leak. Of your Google income, the share where an AI Overview shows, times the clicks lost when you're not the answer, times the share of your test questions where you're not cited.
    SEO leak = Google income × P_aio × L_click × (1 − share cited)
  4. AI assistant leak. Of your new-customer income, the share of buyers who use an AI assistant to research, times the share whose choice the AI answer shapes, times the share of your test questions where you're not named.
    AI assistant leak = new-customer income × U_ai × I_influence × (1 − share named)
  5. The cap. The same buyer can use both Google and an AI assistant, so the two leaks can overlap. The estimate is never allowed to be more than the published share of buyers who use AI search.
    Cap = new-customer income × Cap share
  6. Income at risk. The smaller of the two leaks added together and the cap. Multiply by 12 for a year.
    Income at risk = MIN(SEO leak + AI assistant leak, Cap)
  7. Spend you could re-scope. For each marketing channel, your monthly spend times the published share of that kind of budget that marketers point at AI visibility. It's the same dollars pointed at AEO and GEO work, not extra spend. Channels with no published share show no number.
    Re-scope = Σ (channel spend × published share)
  8. Payback ratio. Income at risk divided by the re-scoped budget. A number above 1 means the estimated income at risk is larger than the re-scoped budget. It is not a forecast of return.
    Payback ratio = income at risk ÷ re-scope
  9. Potential gain. If you become cited and named: the Google clicks won back by being cited, plus the AI assistant leak recovered, never more than the income at risk. G_cite = (20,743 − 9,445) ÷ 33,500 = 33.7%, from Seer Interactive (24 Apr 2026, S11). This is a conditional estimate, not a guarantee.
    Potential gain = MIN(Google income × P_aio × G_cite × (1 − share cited) + AI assistant leak, income at risk)

If you don't check your test questions, the check assumes you're not currently cited or named (share cited = 0, share named = 0). That's a stated condition about your situation, not a benchmark.

Every published input, in three scenarios

The check always shows all three scenarios as a range. Each figure below is taken from a published study, rounded as in our model.

Benchmarks used in the income-at-risk and potential-gain formulas
InputConservativeModerateAggressive
P_aio
Share of your Google searches that show an AI Overview
8%
Seer: AI Overviews show on 8% of commercial searches
Seer Interactive (24 Apr 2026, S11)
25%
Conductor: AI Overviews appear on 25% of searches on average
Conductor via Business Wire (13 Nov 2025, S17)
36%
Seer: AI Overviews show on 36% of informational searches
Seer Interactive (24 Apr 2026, S11)
L_click
Clicks lost when an AI Overview shows and you're not the answer
47%
Pew: 8% clicked with a summary vs 15% without, a 47% drop (1 − 8 ÷ 15, rounded)
Pew Research Center (22 Jul 2025, S09)
58%
Ahrefs: number 1 result gets about 58% fewer clicks
Ahrefs (4 Feb 2026, S10)
72%
Seer: 9,445 vs 33,500 clicks per million when not cited, a 72% drop (1 − 9,445 ÷ 33,500, rounded)
Seer Interactive (24 Apr 2026, S11)
U_ai
Share of Australian buyers who use an AI assistant to research a purchase
27%
Ideally: 27% regularly use AI to research before buying
Ideally (1 Sep 2026, S19)
31%
Ideally: 31% used ChatGPT the last time they researched
Ideally (1 Sep 2026, S19)
40%
WPP: 40% go straight to ChatGPT, Gemini or Claude
WPP Media via AdNews (17 Sep 2026, S20)
I_influence
Share of those AI users whose choice is shaped by the AI answer
3%
Ideally: 3% would buy on an AI recommendation alone
Ideally (1 Sep 2026, S19)
26%
WPP: 26% completely changed what they planned to buy
WPP Media via B&T (17 Sep 2026, S20b)
61%
WPP: 61% discovered brands they hadn't considered
WPP Media via AdNews (17 Sep 2026, S20)
Cap
Most the estimate can be, as a share of your new-customer income
40%
WPP: 40% use AI tools directly
WPP Media via AdNews (17 Sep 2026, S20)
50%
Ideally: 50% have used AI to research before buying
Ideally (1 Sep 2026, S19)
75%
WPP: three in four (75%) use AI search for purchase decisions
WPP Media via AdNews (17 Sep 2026, S20)
G_cite
Clicks won back by being cited in an AI Overview, as a share of clicks with no AI Overview
33.7%
Seer: (20,743 − 9,445) ÷ 33,500 = 33.7%
Seer Interactive (24 Apr 2026, S11)
33.7%
Seer: (20,743 − 9,445) ÷ 33,500 = 33.7%
Seer Interactive (24 Apr 2026, S11)
33.7%
Seer: (20,743 − 9,445) ÷ 33,500 = 33.7%
Seer Interactive (24 Apr 2026, S11)

Spend optimiser shares

Published share of each channel that could be re-scoped to AEO and GEO
ChannelConservativeModerateAggressive
SEO (search engine optimisation)12%
Conductor (updated 14 Apr 2026, S28)
24%
Fractl (2026, S26)
50%
Digiday (25 Mar 2026, S27)
Content (blog, articles, video, copywriting)12%
Conductor (updated 14 Apr 2026, S28)
24%
Fractl (2026, S26)
39%
Fractl (2026, S26)
Agency retainer or management feesUses the SEO row only if you say the retainer is mainly SEO or content work. Otherwise: no published benchmark.
Google Ads (search ads)No published benchmark, so no number
Meta and other social ads (Facebook, Instagram, TikTok, LinkedIn)No published benchmark, so no number
Other online ads (display, YouTube, Microsoft Ads)No published benchmark, so no number
Organic social media (posting, management)No published benchmark, so no number
Email and SMS marketingNo published benchmark, so no number
Directories and listings (incl. lead sites)No published benchmark, so no number
PR and mediaNo published benchmark, so no number
Print, radio, TV and outdoorNo published benchmark, so no number
Events, sponsorships and trade showsNo published benchmark, so no number
Website build, hosting and upkeepNo published benchmark, so no number
Any other marketing spendNo published benchmark, so no number

The full worked example

A business with 20 new customers a month, an average first sale of A$2,000 ex GST, 40% found on Google, not cited and not named, and an agency retainer that is mainly SEO and content work. Monthly spend: SEO A$1,500, Content A$500, Agency retainer or management fees A$1,000, Google Ads A$2,000, Meta and other social ads A$800, Directories and listings A$100.

New-customer income = 20 × A$2,000 = A$40,000. Google income = A$40,000 × 40% = A$16,000.

Worked example, a month, ex GST
LineConservativeModerateAggressive
SEO leakA$16,000 × 8% × 47% = A$602A$16,000 × 25% × 58% = A$2,320A$16,000 × 36% × 72% = A$4,147
AI assistant leakA$40,000 × 27% × 3% = A$324A$40,000 × 31% × 26% = A$3,224A$40,000 × 40% × 61% = A$9,760
CapA$40,000 × 40% = A$16,000A$40,000 × 50% = A$20,000A$40,000 × 75% = A$30,000
Income at riskA$926 (A$11,107 a year)A$5,544 (A$66,528 a year)A$13,907 (A$166,886 a year)
Re-scopeA$1,500 × 12% + A$500 × 12% + A$1,000 × 12% = A$360A$1,500 × 24% + A$500 × 24% + A$1,000 × 24% = A$720A$1,500 × 50% + A$500 × 39% + A$1,000 × 50% = A$1,445
Payback ratio2.6×7.7×9.6×
SEO recapture if citedA$16,000 × 8% × 33.7% = A$431A$16,000 × 25% × 33.7% = A$1,348A$16,000 × 36% × 33.7% = A$1,941
Potential gainA$755 (A$9,064 a year)A$4,572 (A$54,864 a year)A$11,701 (A$140,413 a year)
Gain ÷ re-scope2.1×6.4×8.1×

Spend with no published AEO/GEO benchmark in this example: A$2,900 a month (Google Ads, Meta ads and directories). It's shown, not re-scoped.

Try it yourself: take any line above, put the numbers into a calculator and you should get the same answer, give or take a dollar from rounding.

The readiness checks and why each matters

Each pillar is shown as "X of Y checks passed": a simple count with no weighting. "Top 3 fixes" are the first three checks you mark "No", in this order.

Readiness checks
PillarCheckWhy it matters (published source)
AEODoes your site let AI crawlers in (such as GPTBot, ClaudeBot and PerplexityBot)?AI visibility checks test whether AI crawlers can access your site, OtterlyAI (2026, S32)
AEOIs your main content in the page itself, not loaded only by JavaScript?None of the major AI crawlers render JavaScript, Vercel (with MERJ) (17 Dec 2024, S39)
AEODo your pages answer customer questions directly, for example with an FAQ?Question searches trigger AI Overviews 85.9% of the time; comparisons 95.4%, Seer Interactive (24 Apr 2026, S11)
AEODoes your site have business structured data (ABN, address, area served)?Google uses structured data as clues to what a page means, Google Search Central (current, S40)
AEOIs your site cited in Google's AI Overview for at least one of your 5 test questions?Cited brands get 20,743 clicks per million against 9,445 when not cited (+120%), Seer Interactive (24 Apr 2026, S11)
GEODo ChatGPT, Gemini or Perplexity name your business for at least one of your 5 test questions?31% of Australians used ChatGPT the last time they researched a purchase, Ideally (1 Sep 2026, S19)
GEOAre you listed on the other websites AI answers cite in your industry (directories, reviews, local media)?A brand's own site is only 5% to 10% of the sources AI cites, McKinsey & Company (16 Oct 2025, S13)
GEODo your name, address, phone and prices match on Google Business Profile and review sites?21% would check review sites when AI recommends a brand they don't know, Ideally (1 Sep 2026, S19)
GEOWhen you ask AI about your business, does it state your prices and claims correctly?66% of Australians have used AI to check an advertising claim, WPP Media via AdNews (17 Sep 2026, S20)
SEODoes your site meet Google's basics (indexable, page titles, HTTPS, works on phones)?Google Search Essentials lists these as requirements for Search, Google Search Central (current, S38)
SEODo you show on page 1 of Google for your main local search?The number 1 result loses about 58% of clicks when an AI Overview shows, Ahrefs (4 Feb 2026, S10)

We don't check or score llms.txt, because there's no published proof that it has an effect.

Studies we used

Please check these figures yourself using the links above. They're estimates from published studies, not guarantees.

What this check can't tell you

Our values: trust, transparency and accessibility.

Run the check

Made by AISearch Global. This check, this page and their code: © 2026 Viveka Mohan Das. All rights reserved.