Lesson 4 of 6 · 9 min read · beginner
How arbitrage businesses die
Six ways search arbitrage businesses fail: feed loss, account bans, clawbacks, cash crunches, policy changes and concentration, each with a documented example.
Most businesses fail slowly, through falling sales. Arbitrage businesses often fail in a day, with an email. Understanding how they end is more useful than any growth tactic, because survival in this industry is mostly the avoidance of six known deaths.
A market trader rents a stall by the week from a landlord who can change the rules of the market whenever he likes: what may be sold, where the stalls stand, whether there is a market at all. The trader can be excellent at trading and still be out of business on Monday. The skill that matters most is not being wholly dependent on one market.
1. Losing the feed
The feed is the only source of revenue. If the provider ends your account, or the search engine ends the provider's contract, revenue goes to zero at once, while campaigns already running keep spending until someone stops them. Causes range from poor traffic quality and policy breaches to a simple commercial decision by the search engine.
From violation to termination
A search feed is lent, not owned. The engine’s AFS and RSOC policies, plus the provider’s contract, say what the ads, pages and traffic must look like. Compliance is the price of keeping the tap open.
- A feed comes with a rulebook: A search feed is lent, not owned. The engine’s AFS and RSOC policies, plus the provider’s contract, say what the ads, pages and traffic must look like. Compliance is the price of keeping the tap open.
- Something breaks a rule: Typical causes: a misleading ad that promises what the page does not deliver, search terms unrelated to the article, an unapproved traffic source, or wording that pushes people to click ads. Each is a policy violation.
- It gets noticed: Automated checks, manual reviewers and advertiser complaints all feed a review. The provider is watching too, because the engine holds it responsible for its publishers.
- First rung: a warning: For a first or minor problem the usual result is a notice naming the issue and a deadline. Fixing the ad or page, and showing it, normally ends the matter. This is the cheap exit.
- Second rung: limits: If issues repeat or quality looks weak, the feed may be throttled: a feed cap on daily volume, fewer ads per page, or a ban on one traffic source. Revenue falls immediately while ad spend may still be running.
- Third rung: revenue taken back: Earnings tied to the violating traffic can be withheld or deducted, sometimes for weeks already reported. If $28,000 was estimated and $8,000 of it is judged non-compliant, that Clawback (revenue deduction) alone wipes out the month’s profit in our example.
- Last rung: termination: Serious or repeated breaches end in the feed being switched off, often without a second chance, and unpaid balances may be kept (feed suspension). Severe cases can skip every earlier rung. This is platform risk at its plainest.
- The damage travels: A terminated publisher rarely gets another feed quickly: providers share the same few engines and ask about history. And a provider with too many bad publishers risks its own contract, which is why approval is slow and monitoring constant.
This is not hypothetical, and it reaches the largest names. IAC, the US media group, reported that the Google services agreement supplying paid listings to its Ask Media Group expired on 30 April 2026, after Google gave notice that it would not renew. IAC said it ceased Search operations when the agreement expired and would report the segment as discontinued. Its Search revenue for the first quarter of 2026 was $17.1 million, down 76% on a year earlier.
2. Ad account bans
The other dependency is the place you buy traffic. An account ban on Meta, TikTok or a native network stops the supply of visitors. Bans follow policy violations in creatives or landing pages, payment problems, or automated risk systems that give little explanation. Appeals are slow. Operators who respond by opening fresh accounts under other identities breach the platform's terms again and tend to be banned across the board.
3. Clawbacks
A clawback removes revenue after the traffic has been paid for. On a thin margin it does not take much. If a business keeps 10% of revenue after traffic costs, a 10% deduction makes the month break even, and anything more is a loss on money already spent. Large clawbacks usually point to a quality problem in one source that was not caught in time, which is why the fraud track spends so long on detection.
4. The cash crunch
Arbitrage consumes cash as it grows. Traffic is charged daily and feeds pay on net payment terms, so growth widens the gap that working capital must fill. Companies fail while profitable on paper because a payment arrives late, a credit limit is cut, or a clawback lands in the same week as a large card bill.
An illustration: an operator spending $5,000 a day with payment 45 days after month end may have well over $300,000 outstanding at the peak of the cycle. If the provider delays one monthly payment by two weeks, the operator must find another $70,000 of spend from somewhere or switch campaigns off, and switching off resets the traffic platforms' learning phase, which hurts performance when campaigns restart.
5. Policy change
Sometimes nobody did anything wrong and the rules simply moved. Three recent examples show the scale.
| Change | What happened | Reported effect |
|---|---|---|
| Microsoft Bing distribution changes, 2024 | Perion Network said Microsoft Bing changed pricing and mechanisms in its search distribution marketplace in early 2024 and later excluded a number of publishers | Perion's Search Advertising revenue fell 53% in 2024 to $162.7 million, and its total revenue fell from $743.2 million to $498.3 million |
| Google parked-domain opt-out, 2025 to 2026 | Advertisers were opted out of parked domains by default from March 2025, and parked domains were removed from the search partner network on 10 February 2026 | Team Internet Group's Search segment revenue fell 52% in the first half of 2025, and trade press reported about 200 job cuts |
| Google RSOC tightening, 2025 | Restricted features for many RSOC publishers from August 2025, and mandatory reporting of upstream ad text from 1 November 2025 | Fewer related-search units per page and closer checking of ads against pages, as reported by trade press and Google's help pages |
None of these was a punishment of one bad actor. Each was a platform changing its product, and each removed a large share of revenue from companies that had complied with the previous rules.
6. Concentration
Concentration risk is the condition that makes the other five fatal. One feed, one traffic source, one vertical, one winning creative, one key employee: each is a single point of failure. The Perion case is the textbook example. A company with hundreds of millions in revenue had, by its own account, a large part of its search business tied to one partner, and the change at that partner reshaped the whole company.
Fragile
- One feed provider, one search engine
- One traffic platform, one ad account
- One vertical carrying most of the profit
- Scaling on estimated revenue
- No cash reserve, credit fully used
- Policy news read after the fact
More resilient
- Two or more feeds where approvals allow
- Several traffic sources, each declared
- Profit spread across verticals and geographies
- Decisions on finalised revenue
- Reserve sized to a bad month's clawback
- A named person tracking policy changes
How it usually happens in practice
- A good runA campaign works. Spend is raised quickly, funded by credit.
- Quality slipsTo find more volume, the operator widens targeting or adds a cheaper source. RPC drifts down.
- The warning is missedEstimated revenue still looks fine. The provider's quality feedback worsens but nobody reviews it.
- The deduction arrivesFinalised revenue is well below estimate. The month is a loss, and the cash was already spent.
- The cap or termination followsThe provider limits or ends the feed. Remaining balances are held pending review.
- Credit is dueCard and credit-line payments fall due with no incoming revenue to meet them.
Key takeaways
- The six common deaths are feed loss, ad account bans, clawbacks, cash crunches, policy change and concentration.
- Feed loss reaches even large companies: IAC ceased its Search operations when a Google agreement expired in April 2026.
- Policy changes at Microsoft in 2024 and Google in 2025 and 2026 removed large shares of revenue from compliant companies.
- Profitable operators still fail when payment timing and clawbacks collide with daily traffic bills.
- Concentration is what turns any single shock into the end of the business.
Questions people ask
Why do search arbitrage businesses fail?
Usually for one of six reasons: the search feed is withdrawn, an advertising account is banned, revenue is clawed back after traffic was paid for, cash runs out between paying for traffic and being paid, a platform changes its rules, or the business depended on a single feed, source or vertical. These often arrive together and quickly.
What happens when you lose a search feed?
Revenue stops immediately, while any campaigns still running keep costing money until paused. Unpaid balances may be withheld during a review, and deductions can still be applied to earlier periods. Getting a new feed takes weeks and may be refused if the previous one was lost for quality or policy reasons. Staff and credit costs continue meanwhile.
What happened to Perion's search business?
Perion Network said that in 2024 Microsoft Bing changed pricing and mechanisms in its search distribution marketplace and then excluded a number of publishers. Perion reported that its Search Advertising revenue fell 53% in 2024 to $162.7 million and total revenue fell to $498.3 million from $743.2 million. It has since shifted its focus toward other advertising lines.
How can an arbitrage business reduce platform risk?
It cannot remove it, only limit it. Useful steps are holding more than one feed where approvals allow, using several declared traffic sources, spreading profit across verticals, keeping a cash reserve, making decisions on finalised revenue, and assigning someone to follow policy updates. Many larger companies also build income outside search arbitrage altogether.