Marketix Digital, a Sydney search consultancy, today published a revenue model in which a hypothetical business that loses a fifth of its organic search visits is still A$43,400 a month short after gaining 5,000 AI referrals, each worth 83% more than an average organic visit. The analysis, written by managing director Shoaib Mughal and titled "Can AI search replace the revenue behind lost clicks", takes Adobe's US retail revenue-per-visit premium and applies it to an Australian dollar scenario. Marketix states plainly that the examples are not forecasts.

The point of the exercise is narrow, and it is arithmetic rather than measurement. Higher-value visitors, the model shows, only close a revenue gap when enough of them arrive. Under the 1.83 ratio Marketix borrows from Adobe, a business needs about 55 additional AI visits to replace the revenue from 100 lost visits of comparable value. Whether those 55 visits exist is the question the model cannot answer, and the evidence Marketix assembles on traffic volume suggests the answer varies a great deal by business.

In Short

Marketix Digital built a model showing that even if visitors from AI tools spend 83% more per visit, a business losing 20,000 organic visits a month and gaining 5,000 from AI ends up A$43,400 a month worse off. That matters to anyone whose revenue depends on search traffic, because a higher conversion rate from AI referrals is often cited as compensation for falling clicks. The practical point is that the size of the loss depends on how many visits disappear, how valuable they were, and how many new AI visits actually arrive, and none of these can be read from a conversion rate alone.

What Marketix published

The analysis sits in the Research category of the Marketix blog and carries today's date. A release issued from Sydney the same day carries the headline "Higher-value AI traffic leaves A$43,400 monthly revenue gap in Marketix model." Marketix describes itself as a Sydney-based "Search Intelligence & Advisory" consultancy, and its site lists SEO, Google Ads and research-led digital PR among its services for B2B, industrial, enterprise and e-commerce businesses across Australia.

That commercial interest is relevant to how the piece reads. The analysis closes with an invitation to request an "Investment Briefing" and a paragraph describing how the firm's advisory service "connects search demand, competitive visibility and customer economics." Its central conclusion, that search investment should be judged on revenue and contribution rather than rankings or conversion rates, is also the premise of the service Marketix sells.

On method, Marketix is unusually explicit. According to the method section, the firm reviewed the research today and "combines published findings with an illustrative revenue model." It "does not use client analytics or CRM records, and the examples are not forecasts for Australian businesses." The analysis refers to an accompanying workbook containing "the inputs, formulas and supporting calculations," and says it lets businesses test their own traffic volumes, visitor values and costs. The article itself does not link to that workbook.

Twelve numbered sources sit underneath the model: Adobe, IAB Australia and PwC, three Ahrefs studies, Similarweb, Seer Interactive, Pew Research Center, an August 2026 academic preprint, a 2025 statement from Google's head of Search, Google's Search Console documentation and Demandbase. The release describes the approach as combining "published research with revenue-replacement modelling" and adds that the Adobe figure is used "as an assumption; it does not measure actual Australian business losses or forecast future results."

The core arithmetic: 55 visits for every 100

The model rests on a single division. According to Marketix, "Additional AI visits required = displaced revenue / revenue per additional AI visit."

The input that drives the result is Adobe's. Adobe reported on October 29, 2026 that AI traffic to US retail sites generated 83% more revenue per visit than non-AI traffic in September, the widest gap it had recorded this year, and converted 90% better. Marketix turns that premium into a ratio of 1.83. If an AI visit is worth 1.83 times an ordinary one, replacing 100 lost visits takes 100 divided by 1.83, or about 55 additional AI visits. "It would still need to attract those extra visitors," the analysis notes.

The same ratio works in reverse. If every remaining visit generated 83% more revenue, Marketix calculates, a business could lose 45.4% of its visits and keep the same revenue. The firm is careful to frame that as an illustration of "the trade-off between traffic volume and visitor value; it is not a prediction of what AI will deliver."

A second calculation addresses visitors who still arrive. Ahrefs found in February 2026 that click-through rates at position one were 58% below the level its model estimated would have prevailed without AI Overviews. Marketix converts that into an order problem: if visits fall from 100 to 42, those 42 visitors "must convert 2.38 times as often to maintain the same number of orders." That is a 138.1% rise in the conversion rate, assuming unchanged order values and no replacement traffic.

Marketix extends the calculation across the first ten organic positions, using Ahrefs' relative click-through reductions for each. The required conversion increase falls with position: 138.1% at position one, 103.3% at position two, 86.6% at position three, 63.4% at position four and 48.4% at position five. Positions six to nine sit between 40.4% and 43.9%, and position ten needs 24.1%. Marketix labels the chart a derived calculation and adds that "this is not a sitewide forecast," since it assumes fixed impressions and order values.

The A$43,400 scenario

The headline figure comes from a worked example. A hypothetical business receives 100,000 organic visits a month, each generating A$4, for A$400,000 in monthly revenue. It loses 20% of those visits, and Marketix assumes the lost visitors were worth the organic average. It then gains 5,000 additional AI referrals, each valued at A$7.32 - the A$4 average multiplied by Adobe's 1.83.

Measure Modelled monthly result
Revenue displaced by 20,000 lost visits A$80,000
Revenue per additional AI visit A$7.32
Revenue from 5,000 additional AI visits A$36,600
Displaced revenue replaced 45.75%
Direct-referral revenue shortfall A$43,400
Final revenue from the modelled cohort A$356,600
Additional AI visits needed to replace displaced revenue Approximately 10,929

The AI visits recover a little under half of what was lost. To close the gap entirely, the business would need about 10,929 additional AI visits in total, meaning about 5,929 more than the 5,000 already assumed.

"Businesses need to put a revenue target against their AI search investment. More valuable visitors only close the gap when enough of them arrive and buy," Mughal said in the release.

Costs push the threshold higher. At a 30% contribution margin and A$2,000 in extra acquisition costs, according to Marketix, monthly contribution falls by A$15,020. Restoring the original contribution after those costs would take approximately 11,840 additional AI visits. Marketix stresses that contribution here "is not company net profit."

When the lost visitors were worth less

The scenario flips if the visits that disappear were low-value ones. At A$1 per lost visit - a quarter of the organic average - the same 20% traffic decline removes only A$20,000. The A$36,600 from AI referrals then more than covers it, leaving revenue A$16,600 higher before acquisition costs. At half the organic average, the business ends A$3,400 short. At the full average, it is the A$43,400 gap.

The visit requirement scales the same way. According to Marketix, replacing lost visits worth 25% of the organic average takes 2,732 additional AI visits at A$7.32 each; at 50%, 5,464; at 100%, 10,929.

"The traffic decline is identical, but the financial outcome is different," the analysis states. "This is why a traffic-loss percentage alone cannot tell a business how much revenue is at risk."

That sensitivity is arguably the most useful finding in the piece. It means two companies reporting the same 20% fall in organic traffic can face opposite financial results depending on whether they lost early research visits or purchase-ready ones. The release makes the same point: "Losing low-value research traffic creates a smaller financial hurdle than losing purchase-ready customers."

The assumptions doing the work

Two inputs determine almost everything in the model, and both deserve scrutiny.

Adobe's 1.83 ratio

The first is the Adobe premium. Adobe's figures describe US retail websites in a single month, drawn from online transactions across more than one trillion visits, 100 million SKUs and 18 product categories, according to the company. AI traffic is counted only when users click a link. And the comparison is not with organic search: Adobe sets AI referrals against all non-AI traffic, a pool that includes channels such as email, paid search and social.

Marketix acknowledges the gap. Its method notes say "Adobe's AI-to-non-AI revenue ratio is used as a proxy for AI-to-organic revenue in the examples" and that the figures "do not establish that moving a visitor to AI search causes a revenue increase."

The ratio has also moved sharply. According to Adobe, AI traffic went from being worth less than half of a typical visit to nearly twice as much in the space of 12 months. By PPC Land's calculation, at a ratio of 0.5 rather than 1.83, replacing 100 lost visits would have taken 200 AI visits, not 55. The model's central number is therefore tied to one month of US retail data, at the widest gap Adobe had recorded this year.

The 5,000 extra visits

The second input is the volume of AI referrals. Marketix assumes 5,000 extra AI visits against 20,000 lost organic visits, which means AI replaces a quarter of the lost traffic by volume before any value premium is applied. The analysis does not explain how that figure was chosen.

Published data points to a lower ratio for at least some businesses. Brainlabs data that PPC Land reported in October showed organic sessions fell 10.5% across 54 clients while AI visits made up only about 5% of the loss; PPC Land's own arithmetic on those figures put the AI gain at about 821,000 sessions against a loss of about 14.69 million, or roughly 5.6%. Applied to the Marketix scenario, by PPC Land's calculation, a 5.6% replacement rate would bring about 1,120 AI visits rather than 5,000. At A$7.32 each, that is about A$8,200, or roughly 10% of the A$80,000 lost, leaving a gap near A$71,800 a month. The calculation mixes two unrelated datasets and is offered only to show how much the result depends on the volume assumption.

The traffic data Marketix cites itself points the same way. Across the 118,671 websites in Ahrefs' Web Analytics dataset, AI assistants accounted for 0.46% of visits in October 2026, while Google accounted for 24.74%. By PPC Land's calculation, that puts Google's share at roughly 54 times that of AI assistants in that dataset. Marketix cautions that the Ahrefs sites are "a selected vendor dataset, not a representative Australian sample," and that its channel categories overlap.

The click research behind the model

Much of the analysis is a survey of studies on how AI summaries in Google results affect clicks. Marketix presents them together but warns that "their percentages should not be combined into a single estimate of Australian traffic or revenue losses," because they cover different countries, periods and groups of websites or users.

The Ahrefs study from February 2026 compared December 2023 and December 2025 desktop data across 300,000 keywords against a trend-adjusted estimate of performance without AI Overviews. PPC Land reported at the time that Google's AI summaries were associated with a 58% lower click-through rate for top-ranking pages, up from 34.5% in April 2025. Marketix notes that the 58% concerns position-one click-through rates, "not a universal fall in website visits."

Seer Interactive's April 2026 update showed a partial recovery. Organic click-through rates for searches with AI Overviews fell from 3.19% in January 2025 to 1.31% in December, then recovered to 2.36% in February 2026. Marketix calculates that as an 80.2% rebound from December but still 26.0% below the January 2025 starting point. It also notes that "the widely quoted 85% uses more rounded inputs," and that Seer classified AI Overview queries retrospectively without a seasonality adjustment. The levels differ from the Seer release PPC Land covered in November 2025, which put organic click-through on 3,119 informational queries with AI Overviews at 0.61% in September 2025, down from 1.76% in June 2024. The two sets of figures are not directly comparable.

Pew Research Center's US browsing study, published in July 2025, found that users clicked a traditional search result on 8% of visits where an AI summary appeared, against 15% where it did not. Marketix notes that Pew measured search visits rather than ranking-position click-through rates, and matched March 2025 browsing with results collected in April.

France offers a before-and-after comparison. AI Overviews launched there on July 22, 2026, and according to Marketix, Ahrefs later found a median click-through-rate decline of 23.1% among highly exposed websites, against 5.7% across all 963 websites studied. Marketix flags the study's short post-launch period and the different exposure levels of its groups.

Academic experiments take a different route, changing what users see rather than observing it. A preprint by Stephanie T. Wang and colleagues, submitted to arXiv on August 18, 2026, reports a preregistered field experiment with 1,100 participants that tested Google results with AI Overviews and AI Mode removed. According to Marketix, removing AI Overviews in that experiment, which ran in March, increased outbound click-through rates by an estimated 8.8 percentage points, a figure adjusted for incomplete compliance because only 51.1% of AI Overviews were successfully hidden. It is a separate study from the randomized experiment by researchers at the Indian School of Business and Carnegie Mellon, which found that AI Overviews cut outbound organic clicks by 39.8% among 1,065 US desktop Chrome users.

Google's position is the counterweight. In August 2025, Liz Reid, head of Google Search, said total organic click volume to websites was relatively stable year over year. Marketix points out that "a stable overall total can still hide losses for individual businesses and types of search," and that Google "did not publish the underlying aggregate dataset with its traffic statement." In July 2026, Google's Nick Fox said AI features in Search were sending billions of clicks to websites every week, again without a denominator.

None of these studies measured revenue. "These studies measured browsing or clicks, not business revenue," Marketix writes, which is precisely the gap its model tries to fill with assumptions.

Referrals and influence are different measures

The analysis separates two things that are often blurred: the visits AI sends and the purchases AI influences. A shopper might discover a supplier in a chatbot answer, search for the brand days later, and submit an enquiry through Google. Under last-click attribution, Marketix notes, those journeys can be credited to different channels "without revealing which touchpoint changed the decision."

Similarweb's State of Ecommerce 2026 release, published in September, found that consumers who use AI for shopping research also use search 89% of the time. Marketix cautions that the release "does not provide enough detail to translate its shopping-research finding into Australian sales."

For that reason the model deliberately excludes assisted sales. According to the method notes, it "does not assign unmeasured assisted sales to AI or assume that new AI sales are entirely additional to other channels." The A$43,400 figure is therefore labelled a direct-referral revenue shortfall. Some of that gap could, in principle, be recovered through branded searches or direct visits prompted by AI answers. The model neither counts those nor rules them out.

Measurement tools have not closed the gap either. Google's Search Console report for generative AI features provides visibility data for AI Overviews and AI Mode, but Marketix notes that it "does not provide a dedicated AI revenue field," that its dates use Pacific Time, and that queries "cannot be directly matched to individual analytics sessions or sales." When Google added the generative AI reports to Search Console in June 2026, the five data categories it listed did not include clicks.

Marketix proposes separating four questions in reporting: whether demand has changed, whether visibility has changed, what business the traffic generates, and how customers are finding the company, with each sale counted once.

The B2B and Australian angle

For business-to-business companies, the time between a visit and a sale complicates the picture further. In August, Demandbase reported that monthly ChatGPT referrals to the B2B websites it measures grew from roughly 645,000 in June 2025 to 2.6 million in June 2026. Marketix says that growth "makes the channel worth monitoring, while the commercial test remains the quality of the opportunities it brings," and notes that Demandbase's figures "do not establish a conversion premium." It illustrates the point with Australian industrial examples: "A dust-monitoring enquiry, an equipment-hire lead and a low-value consumer question can have very different values."

The Australian market context comes from paid search. IAB Australia and PwC put search advertising revenue at A$8.6 billion in FY26, up 13.2%, within a A$19.8 billion internet advertising market. Marketix uses the figure to show "how much businesses invest in reaching customers through search," while noting that it "measures paid-search advertising revenue, not organic traffic value." No Australian organic traffic or AI referral data appears anywhere in the model. The A$4 visit, the 20% decline and the 5,000 AI visits are all hypothetical inputs, and the only empirical figure in the scenario is a US retail ratio.

What the model does and does not show

Marketix's method section lists the model's limits in detail. The studies cover different markets and periods. The Ahrefs dataset is selected. Seer did not adjust for seasonality. The France study used a short window. Wang's estimate depends on a compliance adjustment. Google released no data with its traffic claim. Lost visitors are assumed to carry the organic average value, and retained visitors keep their value unless stated.

What the model does establish is structural. A conversion-rate premium, however large, is a ratio; revenue depends on that ratio multiplied by volume. "A better conversion rate alone cannot settle the investment decision," the analysis states. "Higher revenue per AI visit does not establish that enough additional visits are available to replace lost organic revenue."

What it does not establish is the size of any real gap. The A$43,400 depends on an unexplained 5,000-visit assumption and a ratio drawn from a different country, sector and comparison group. Changing either moves the result by tens of thousands of dollars in either direction, as Marketix's own sensitivity table shows.

Why this matters for marketers

The debate over AI search has repeatedly set two kinds of numbers against each other: usage figures for AI tools, and referral figures for the websites those tools draw on. Last week, OpenAI's Adam Fry said ChatGPT's volume of search-like queries had reached roughly one-fifth of Google's, while Taboola tied a 6.9% fall in third-quarter revenue to fewer ad impressions and cited AI summaries. Marketix's analysis sits in the space between those two facts, distinguishing activity inside an AI interface from revenue delivered outside it.

For organic search teams, the model reframes a common argument. Higher conversion rates from AI referrals have been cited as evidence that falling clicks matter less than they appear. Ahrefs reported in June 2025 that AI search visitors to its own site converted 23 times better than organic visitors, while making up 0.5% of visits. Marketix's arithmetic shows why such premiums, on their own, cannot answer whether revenue is recovering. In April, HubSpot reported a 27% year-over-year fall in organic traffic for its customers. As an illustration only, by PPC Land's calculation, a 27% fall in Google visits within Ahrefs' October dataset would remove about 6.7% of all visits, roughly 14.5 times the entire AI assistant share; even at 1.83 times the value, and even if every AI assistant visit in that dataset were new, they would match about 13% of the loss.

For paid search buyers, the same logic applies to the value of a click. The Brainlabs data that PPC Land covered in October also found that Search campaigns had fallen to 41% of Google Ads clicks across 31 accounts, part of an argument that more buyers arrive having already decided. If AI tools influence decisions without sending the final visit, both organic and paid channels may be credited with sales they did not originate, and AI referral traffic will look smaller than its influence. Marketix's model counts only the direct visit, which makes its A$43,400 a measure of what direct AI referrals recover on their own, not of what AI search is worth in total.

The open question is volume. Every study in Marketix's survey measured clicks, and Adobe measured value per visit, but none measured how many additional AI visits a specific business can expect as its organic traffic declines. That is the number the model needs most, and the one it has to assume.

Timeline

Summary

Who: Marketix Digital, a Sydney search consultancy, and its managing director Shoaib Mughal, drawing on published research from Adobe, Ahrefs, Seer Interactive, Pew Research Center, Similarweb, Demandbase, IAB Australia and PwC, Google and academic researchers.

What: A revenue-replacement model showing that a hypothetical business losing 20,000 of 100,000 monthly organic visits, and gaining 5,000 AI referrals worth 1.83 times an average visit, remains A$43,400 a month short. Under Adobe's ratio, about 55 additional AI visits are needed to replace the revenue from 100 lost visits of comparable value, and the result reverses if the lost visits were low-value.

When: Published today, with research reviewed the same day. The key input is Adobe's September 2026 data, released on October 29, 2026.

Where: Published on the Marketix blog and in a release issued from Sydney. The model uses an Australian dollar scenario but relies on US retail data, and no Australian traffic or revenue figures appear in it.

Why: Higher conversion rates from AI referrals have been cited as compensation for falling organic clicks. Marketix's arithmetic shows that revenue depends on visit volume as much as visit value, and that the share of lost revenue AI recovers cannot be read from a conversion premium alone, a distinction that matters for anyone budgeting organic or paid search as AI tools absorb more queries.