Economics

The real economics of search arbitrage in 2026

The math of search arbitrage with current Tonic and System1 RSOC payouts, real Taboola and Outbrain CPV ranges, and what the margin actually looks like at $1K/day, $10K/day, and $100K/day in spend.

By Eyal RosenthalMay 6, 202617 min readAI-assisted research

Search arbitrage is the most-discussed and least-understood corner of affiliate marketing. The pitch is alluringly simple: buy traffic on a native network for $0.10 per click, send the click into a search feed where the user types or clicks a query, and earn $0.30 per click in revenue from the search-feed partner. Margin: 200%. Scale: as much volume as your bid will buy. Every couple of months an Iman-Gadzhi-adjacent influencer rediscovers it and pitches it as a "secret method." It is not secret. It is well over a decade old. And the actual margin, on a real campaign at real scale, is rarely what the pitch promises. This piece walks through the math.

What "search arbitrage" actually means

The technical term most insiders use is RSOC — Revenue Share on Click — or sometimes "AFD" (AdSense for Domains, the ancestor of the model). The structure has three parties:

The search partner — the company that pays you for the click. The big three in 2026 are System1 (the parent of Tonic.com), Yahoo (via the Yahoo Search Partner Network), and to a lesser extent Bing/Microsoft. They monetize the search query and pay you a revenue share, typically in the 60-80% range, on the gross ad revenue.

The traffic source — Outbrain, Taboola, Facebook, Google Discovery, MGID, RevContent, push networks. You pay them per click or per view to send users into your funnel.

The pre-lander — your domain, your page, your search-feed integration. The user lands on a "topic page" of yours, sees a list of ad-blended search results or a search box, types a query (or clicks a refined-search prompt), and gets routed to a results page where the search partner monetizes.

You earn the difference between what you paid the traffic source and what the search partner pays you. That difference, after every other cost, is your margin. The "every other cost" is where the pitch decks fall apart.

The actual revenue side: what RSOC pays in 2026

Tonic, the most-used RSOC platform among native-ad search arbitragers, is owned by System1. System1 is a public company (NYSE: SST) and discloses meaningful detail in their SEC filings. In their 2024 10-K they describe the RSOC product line under "Owned & Operated" and "Partner Network" segments and disclose blended CPC monetization in the $0.10 to $0.45 range depending on geo, vertical, and partner. The 2025 Q4 earnings call confirmed continued RPM compression in tier-1 verticals and modest RPM expansion in non-US English markets.

Yahoo's Search Partner Network publisher kit and the Bing/Microsoft Audience Solutions docs describe the same revenue structure. Public payout-rate disclosures from these networks are essentially nonexistent — the rates are negotiated per-partner and the partners are typically gagged by NDA. Tonic's public help center describes the categories ("auto," "insurance," "loans," "health," "home services") and confirms that high-CPC verticals like insurance and loans monetize 5-10x better per click than low-CPC verticals like recipes or DIY content.

I do not have a clean public number for the average Tonic RSOC payout per click in May 2026. The rough range based on public Tonic communications and System1 IR materials is $0.18 to $0.55 per click for tier-1 (US/UK/CA/AU) traffic in monetizable verticals, with most accounts settling somewhere in the $0.22-$0.32 range. Below that range, you're probably running a low-intent vertical or a feed that has been throttled. Above it, you're either in a premium vertical (insurance, mortgage, MAGA-adjacent political, certain legal categories) or you have a special-snowflake account with a custom rev-share.

The actual cost side: what Outbrain and Taboola actually charge

Outbrain and Taboola report blended CPC and CPM in their public filings. Outbrain (NASDAQ: OB) and Taboola (NASDAQ: TBLA) both file 10-Ks and 10-Qs with the SEC. From Outbrain's most recent 10-K and Taboola's investor materials, the gross-revenue-per-thousand-impressions ranges they disclose are not directly the advertiser cost — they're the publisher payout — but they bracket the advertiser side reasonably.

The advertiser-side reality, from running campaigns and from talking to operators who run more spend than I do, is:

Taboola US, content-arbitrage vertical: $0.08 to $0.45 CPC, with the vast majority of scaled campaigns landing in the $0.10-$0.18 range. Premium publishers (NBC, USA Today properties, Newsweek-tier) cost more; long-tail publishers cost less but convert less.
Outbrain US, content-arbitrage vertical: $0.10 to $0.55 CPC, slightly more expensive than Taboola on average for similar audience targeting, with the highest-end placements (CNN, BBC, Washington Post) sometimes cresting $1.50.
MGID and RevContent: typically $0.04 to $0.20 CPC for tier-1 traffic, but with materially lower click quality and higher fraud risk on RevContent's lowest-tier publishers.
Push notification networks (PropellerAds, RichAds, Adsterra): $0.001 to $0.02 per click, but click intent is a fraction of native and search-feed monetization is correspondingly weaker.

Translating this into the search-arbitrage P&L: you are typically buying traffic at $0.12 CPC blended across Taboola/Outbrain, sending it into a Tonic feed that monetizes at $0.25 RPCick blended, and pocketing the difference minus everything else.

The 23 line items the pitch deck hides

Here is what the YouTube guru never tells you. The "$0.12 in, $0.25 out, 100% margin" pitch ignores:

Search-feed click-through rate (CTR) on the lander. The Tonic feed shows you 4-12 ad-blended search results. Not every user clicks one. CTRs in the search-arbitrage industry typically run 50-85%. So your "$0.25 RPCick" becomes "$0.25 × 0.65 = $0.16 effective revenue per inbound click."

Revenue-share with Tonic. Tonic typically takes 20-30%. If the search partner monetizes at $0.40 gross, you see $0.28-$0.32 net. Tonic's terms describe the rev-share structure but the exact percentage is per-account.

Compliance throttling. New accounts on Tonic, Yahoo, and System1 partners are throttled for the first 30-90 days, often at 60-80% of standard payout, while compliance teams confirm traffic quality. Reading the Yahoo Publisher policies makes this explicit.

Domain warm-up. A brand-new domain on Tonic typically gets payout-throttled for two to six weeks. New domains also get banned faster from native networks because the trust-graph signals are weak.

Native-network click loss. Outbrain and Taboola charge per click delivered, but a non-trivial fraction of clicks (5-15% in my experience, lower for cleaner buyers) never make it to the lander — bot filtering, double-clicks, slow-load drop-off. The networks' click measurement docs acknowledge this band.

Pre-lander bounce. Even if the user reaches the lander, a meaningful fraction bounces before clicking the search feed. Below-the-fold layouts, slow page loads, or boring topic pages can drop pre-lander CTR from 75% to 40% — and the difference is your margin.

Geo and device mix drift. A campaign that bids at $0.12 blended CPC will, if not actively managed, drift toward whichever publisher and device combination is cheapest, which is usually also lowest-quality. Within 72 hours your blended monetization can drop 20%.

Creative fatigue. A winning headline-image-pre-lander combo decays. Outbrain's own creative-fatigue documentation confirms this. You need to launch new creatives constantly. Each new creative is dead spend until it proves out.

Approval-cycle dead time. New campaigns spend a day or three in compliance review on Outbrain and Taboola before they go live. New creatives can sit in pending. New advertiser accounts can sit in pending for weeks.

Tooling cost. RedTrack, Voluum, Bemob, FunnelFlux, or whichever tracker you use — typically $50-$500/month at small scale, and scales linearly with click volume.

Hosting cost. Pre-landers serve at scale. CDN bills (Cloudflare, BunnyCDN) run $50-$2,000/month at the volumes search-arb operators push.

Domain costs. Most search-arbitragers run multiple domains for compliance and brand-safety reasons. $10/yr per domain, but you need 5-50 of them.

VPS and proxy costs. If you're running scrapers, A/B testing infrastructure, or any kind of evasion logic, that's $200-$2,000/month easily.

Payment-processing delays. Tonic and System1 typically pay net-30 or net-45 from month-end. Outbrain and Taboola charge daily or weekly via the credit card on file. Your working-capital cycle is inverted: you front the money for traffic, you wait six to eight weeks for revenue.

Currency and FX. International buyers pay traffic in USD typically but get paid in USD by Tonic — currency moves matter at scale.

Compliance penalties. A single Outbrain account suspension can vaporize $10K of working capital tied up in pending creative and balance.

Search-partner clawbacks. All RSOC contracts include "low-quality traffic" clawback provisions. A cohort that doesn't post-click monetize well can result in retroactive payout reductions. The Yahoo publisher policy describes this category explicitly.

Tax and corporate overhead. Once you're over $50K/month in revenue you need an LLC, an accountant, and probably quarterly estimated taxes.

The Iman-Gadzhi-style course you bought to learn this. Sunk cost.

Affiliate manager / account manager turnover. Every 6-12 months your AM at the search partner changes and you re-negotiate rates from scratch.

Vertical drift. The verticals that pay best in 2026 (insurance, mortgage, certain MAGA-adjacent political, AARP-targeted senior content) are the verticals with the most aggressive compliance review. The verticals with easy compliance (recipes, DIY) pay terribly.

Click-quality post-mortems. When traffic doesn't perform, you spend hours sifting through campaign logs, not running new campaigns. Opportunity cost is real.

The fact that whoever taught you this is now your competitor. Anyone running a $1K/day course is funded by the course revenue, not the arb. They are also creating their own competition every cohort.

The actual math at three scales

Below is a synthetic but representative search-arbitrage P&L based on what operators I trust report. I do not have permission to publish their actual numbers, so these are best-guess representative bands.

$1K/day in spend (small operator)

Spend: $1,000/day on Outbrain at $0.12 blended CPC = ~8,300 inbound clicks
Pre-lander CTR: 65% → ~5,400 search-feed clicks
Search-feed RPCick (net to you, after Tonic rev share, after throttling on a small account): $0.20 → $1,080 gross revenue
Tooling/hosting: ~$25/day amortized
Net daily margin: ~$55/day, or 5.5% on spend

This is the level where most people quit. The margin is real but trivial; one bad week of creative fatigue or compliance throttling and you're at zero.

$10K/day in spend (mid-tier operator)

Spend: $10,000/day at $0.13 blended CPC across Outbrain + Taboola + MGID = ~77,000 inbound clicks
Pre-lander CTR (tighter optimization): 72% → ~55,000 search-feed clicks
Search-feed RPCick (better account, less throttling): $0.24 → $13,200 gross revenue
Tooling/hosting: $200/day
Net daily margin: ~$3,000/day, or 30% on spend — but this includes a ~10% net buffer for compliance hits, clawbacks, and creative-test losses, so realized is closer to 20%

This is the level where the model genuinely works. Most search-arb operators who tell you they "make $X/day" are at this scale or just above.

$100K/day in spend (top-tier operator)

Spend: $100K/day at $0.14 blended CPC = ~715,000 inbound clicks
Pre-lander CTR: 70% → ~500,000 search-feed clicks
Search-feed RPCick (premium account, custom rev-share): $0.28 → $140,000 gross revenue
Tooling/hosting/staff: $5,000/day (3-5 person team, full ETL, compliance staff)
Net daily margin: ~$35,000/day before clawbacks, or 35% on spend gross — but at this volume, clawbacks and compliance penalties are non-trivial and realized margin is typically in the 18-25% range

At this scale, the constraint is no longer "can I find the offer" — it is "can I keep accounts active across multiple ad-network entities, multiple search-partner relationships, and multiple compliance regimes." The work is operations, not marketing.

Why the pitch decks lie

The simple version is that the pitch deck is sold to the small-operator audience. At $1K/day, you can produce a single screenshot of a single good day on Tonic and claim a 100% margin. You are not lying about that day. You are lying about the average across 30 days, which includes the launch week, the compliance pause, the creative-fatigue cliff, and the clawback. Anyone who has run this for more than a year knows the average is the only number that matters and that the average is much smaller than the highlight-reel single-day numbers.

The other reason the pitch decks lie is that the people selling courses make their money from the courses, not from the arb. If their arb actually worked at the rate they claim, they would not have time to film the course.

What a realistic 2026 search-arb operator looks like

A real operator running this profitably in 2026 typically has:

Two to four traffic-source accounts (Outbrain, Taboola, MGID, RevContent, sometimes Bing/Microsoft Audience), each warmed-up and compliance-clean
Two to four search-partner relationships (Tonic, Sedo, AdMarketplace, Bodis, Domain Holdings)
Five to thirty domains across niches
A custom tracker (Voluum, RedTrack, or homebrew) with full click-level attribution
A creative engine — either in-house designers/copywriters or, increasingly, a structured prompting workflow with image generation and a human compliance reviewer
A compliance person whose job is "don't get banned"
A working-capital line of $50K-$500K depending on scale
A daily P&L review and a weekly account-health review

This is a real business. It is also not the kind of business that anyone running a $997 course will ever describe accurately, because describing it accurately makes it sound like work.

Why the simple model breaks down at scale

The numbers above describe an idealized linear scale. Reality is non-linear in three specific ways that compound and that the pitch deck never addresses.

Diminishing-returns inventory. The CPC you pay rises as you scale spend on a given offer. Outbrain's Conversion Bid Strategy documentation describes the bid-floor mechanics implicitly — the more inventory you want, the higher you have to bid to get it. A campaign that sustains $0.12 blended CPC at $1K/day will often need $0.16-$0.18 blended CPC to sustain $10K/day on the same offer. That 30% bid lift is half your margin if your monetization is $0.24 net. Operators who don't budget for this discover their per-dollar margin compressing as they scale, exactly when they're celebrating revenue growth.

Compliance-team attention scales sub-linearly. A small account doing $1K/day flies under the radar. A growing account doing $10K/day starts getting personal compliance attention. A scaled account doing $100K/day is on the compliance team's daily review queue. The probability that a borderline creative gets paused increases at every tier. Many operators who scale fast hit a "scale wall" at the moment they cross some compliance-team-attention threshold and everything they thought was working becomes contingent on review.

Search-partner clawback risk scales super-linearly. Search partners (Tonic, Yahoo, Bing) reverse low-quality clicks at increasing rates as your aggregate volume grows. The reasoning is partly statistical (more volume = better detection of quality drift) and partly defensive (a $100K/day operator's bad cohort is a much bigger search-partner risk than a $1K/day operator's). Reversal rates that look healthy at small scale routinely double at large scale on the same offer.

The compound effect is that scaling search arbitrage is much harder than the linear math suggests. The operators who get from $1K/day to $100K/day are typically running multiple parallel campaigns and offers, not scaling a single offer 100x.

What "exit" looks like in this business

A topic almost nobody covers honestly: what happens when an operator wants to stop running search arbitrage. The answer is uncomfortable.

The business has almost no asset value beyond cash. There is no recurring-revenue contract. There is no IP that's portable to another business. The traffic-source accounts are non-transferable (terms of service prohibit account sales). The search-partner relationships are non-transferable. The domains have some value but only as URL inventory for a similar operation. The creative library is somewhat reusable but rapidly depreciating.

For comparison, a comparable-revenue SaaS or content business has 5-15x revenue valuations on the multiples public-comparable space. A search-arb business at $5M/year in revenue typically has an enterprise value of approximately $5M/year in revenue (i.e., one year's free cash flow) — and only if a buyer can be found who can run the operation, which is rare. The business is a job, not an asset.

This is why successful search-arb operators almost universally pivot. The most common pivot is into product — building a SaaS or content business that addresses the same audience, where the recurring revenue creates real enterprise value. A few pivot into agency. Almost none "exit" the way a SaaS founder exits.

The implication for operators thinking about whether to enter the business: the right framing is not "will this make me wealthy" but "will this fund the thing I actually want to build next." For operators who think about it that way, the math can work. For operators who imagine the search-arb itself as the long-term business, the math rarely does.

A note on regulatory tail risk

One more honest framing: the search-arbitrage business sits in a regulatory grey zone that regulators have, so far, mostly ignored. The combination of native-ad creative claims, search-feed monetization, and aggressive lander designs has been the target of several FTC enforcement actions adjacent to the model but no broad-spectrum enforcement against the business model itself. A future where the FTC, state AGs, or the search partners themselves decide to be more aggressive is plausible. Operators with single-business-line dependence on search-arb are exposed to that future. Operators with diversified revenue are less so.

This is a low-probability, high-severity tail risk, and the right way to manage it is the same way you manage any tail risk: don't put all your capital and attention in the single basket.