What is the default consensus policy?

ANY. If a notification rule doesn't specify a policy, a single region reporting down is enough to fire the alert. It's the most sensitive setting — it never misses a regional outage, but it also passes through single-region network blips. Change it per notification rule to Two-or-more or Majority to filter those out.

Which policy actually eliminates false positives?

Two-or-more or Majority. A probe-local network problem is confined to one region and can't reach a two-region threshold, so it never pages you. ANY, the default, is the most sensitive policy and will pass single-region blips through — if you want the classic 'false positives eliminated' behavior, you have to select Two-or-more.

What's the difference between 'Two or more' and 'Majority'?

At FourSight's four regions, nothing — both require two regions to agree, so they fire identically. Majority is defined as half-or-more of the running regions, so the two would only diverge if the region count changed. Today, pick either; they behave the same.

Does requiring consensus delay my alerts?

By seconds, not minutes. The four regions check in parallel, so consensus costs roughly one confirmation cycle rather than sequential retries. In exchange you get a pager that's right far more often, and the response-speed gain from trusting every alert dwarfs the confirmation delay.

Is multi-region quorum available on the free plan?

Yes. Every FourSight plan, including Free, checks from all four regions (US, Canada, Europe, APAC) and lets you set the consensus policy per notification rule. Multi-region checking and quorum are not paywalled.

Should I use quorum or a consecutive-failure threshold?

Both — they filter different noise. Quorum removes failures seen from only one region (spatial agreement). The consecutive-failure / confirm-down count removes failures seen in only one moment (temporal agreement). Setting both is what lets a monitor run for months without a false page.

Quorum Consensus Policies: ANY, 2+, Majority & All

What a Consensus Policy Decides

FourSight checks every monitor from four regions — US, Canada, Europe, and APAC — in parallel, on every plan including Free. Each region returns its own verdict: up, degraded, or down. The consensus policy is the rule that turns those four independent verdicts into a single decision: how many regions have to report 'down' in the same window before an alert actually fires. It's not a global switch buried in account settings — it lives on each notification rule, so you can page aggressively for your checkout endpoint and conservatively for a chatty internal service, from the same four regions, without running two monitoring products. Getting this one setting right is most of the difference between a pager you trust and one you've learned to snooze.

The Four Policies

FourSight offers four consensus policies. Because every check runs from exactly four regions today, each policy resolves to a concrete number of regions that must independently report down before the rule fires. The table below is the whole model — there is nothing hidden behind it.

💡 Yes, at four regions 'Majority' and 'Two or more' land on the same threshold — two. Majority is defined as half-or-more of however many regions are running, so the two policies would diverge only if the region count changed. We'd rather show you the real number than imply four meaningfully different thresholds when there are three.

Policy	Regions that must agree 'down' (of 4)	Character
ANY	1	Most sensitive — the default
Two or more	2	Filters single-region blips
Majority	2	A majority of four — same threshold as 'Two or more' at four regions
All	4	Least sensitive — global outages only

The Default Is ANY — Know What That Buys You

If a notification rule doesn't specify a policy, FourSight uses ANY: a single region reporting down is enough to alert. That default is deliberate — it maximizes detection, and it never lets a real outage go unpaged because three regions happened to be slow to agree. But be honest with yourself about the trade: ANY also passes through the exact failures multi-region monitoring is supposed to filter. A packet-loss blip at one probe's data center, one region's DNS resolver having a bad minute, a transient route flap on a single path — under ANY, each of those pages you, because each is one region reporting down. If your monitors have ever cried wolf at 3 AM and resolved themselves before you finished logging in, ANY is usually why. The fix isn't to distrust the pager; it's to raise the bar to two.

Monitoring a Commercial SaaS?

FourSight's free plan includes 10 commercial-safe monitors with multi-region validation — free forever, no card.

Start Monitoring Free

Choosing the Policy That Matches Your Tolerance

There is no universally correct policy — the right one depends on what a false alarm costs you versus what a missed regional outage costs you. Set it per notification rule, and let different monitors carry different policies.

Use Two or more (or Majority) for most production monitors

This is the setting that delivers the false-positive elimination multi-region monitoring is famous for. A probe-local network problem is, by its nature, confined to one region — it cannot reach a two-region threshold. What survives the filter is what a real outage looks like: failure visible from at least two independent networks at once. For the large majority of production endpoints, this is the policy you want, and moving your noisy monitors from ANY to Two-or-more is the single highest-leverage change you can make to alert trust.

Use ANY when a missed outage is unacceptable

For a revenue-critical endpoint where you would rather investigate ten blips than miss one regional outage, ANY is the right, honest choice — you're explicitly buying maximum sensitivity and paying for it in occasional noise. Pair it with a consecutive-failure threshold (below) to strip the most transient blips without giving up single-region detection.

Use All only for genuinely global signals

ALL fires only when every region agrees the target is down, so it will never page you for a regional outage — by design. That makes it appropriate for a narrow set of cases: a noisy background service where you only care if it's globally, totally gone, or a secondary signal you don't want waking anyone. Reach for it deliberately, understanding that a CDN edge failure taking out one geography's users will stay silent under ALL.

Quorum Is Spatial; Confirmation Is Temporal — Use Both

Consensus across regions and a consecutive-failure threshold solve related but different problems, and they compose. Quorum asks 'do enough places agree right now?' — it removes failures isolated to one vantage point. The confirm-down count (how many consecutive bad checks before an incident is treated as real) asks 'has this persisted across time?' — it removes failures isolated to one moment. A one-off timeout that clears on the next check dies to the confirmation threshold; a network blip local to one region dies to quorum. Set both and you've filtered noise in both dimensions, which is why a well-configured monitor can run for months without a single false page.

Regional Splits Are a Diagnosis, Not Just Noise

When some regions fail and others pass persistently, that split is information, and your policy choice decides whether you hear it. One region failing repeatedly while three pass points squarely at regional infrastructure — a CDN edge, a regional DNS problem, a routing fault affecting one geography's users — and the correct response (check your CDN's status for that region; do not roll back your deploy) is different from a global outage. Under ANY, that regional failure pages you, which is what you want if serving that geography matters. Under Two-or-more or Majority, it won't page — you'll see it recorded on the monitor's regional status, but no one gets woken. Neither choice is wrong; the point is to make it on purpose, monitor by monitor, rather than discover your policy mid-incident.

Setting It Up in FourSight

The consensus policy is a field on each notification rule. Create or edit a rule, choose your policy for down alerts, and it applies to every monitor in that rule's scope. The four-region checks themselves don't change — quorum only governs when a rule fires, not how or how often the regions probe. Multi-region checking and configurable quorum are available on every plan, including Free: even a free account gets all four regions and can choose ANY through ALL. That's unusual — most tools reserve multiple regions for paid tiers and offer no consensus control at all — and it's the most concrete reason a free FourSight monitor produces fewer phantom pages than a free single-region one.

Set a consensus policy on your monitors →