Reading

What a dead landing page actually costs

Most “wasted ad spend” numbers are estimates stacked on assumptions. This one is not. A page that cannot convert converts nobody, so every dollar spent pointing at it is waste, and the figure is multiplication: minutes broken × spend rate.

How to calculate wasted ad spend

Rather run it on your own figures? The downtime cost calculator does exactly this sum, and the CAC calculator carries it through to what the waste did to your acquisition cost.

Assume the destination is dead for 3 hours and 33 minutes — the length of INC-2291, and a realistic gap between a midnight deploy and somebody in the next timezone clicking their own ad.

Spend spread evenly across the month. Real accounts are lumpier, and an outage during peak hours costs more than this table says.

How often is the page actually dead? One day's measurement

On 2026-10-01 we put 211 live paid-search destinations — five advertisers, one per vertical, taken from a paid index — through the same checker on this site, one at a time, as a phone, from a cloud address. This is what came back. It is one pass on one day, and it is not a rate: 80 of the 211 refused a datacenter request outright, one vertical refused every one, so any percentage over “whoever answered” would measure bot policy, not page health.

Of the 131 destinations that answered, 5 were hard 404s and 0 were soft 404s. Three of the five were in one vertical and each still had live ads pointing at it the same day. A status-code monitor would have reported those five; it would not have reported the 301 in front of three of them.

Method, column definitions and caveats are in destination-study.json, CC BY 4.0 — quote it with the caveats attached. The reference taxonomy it measures against is failure-modes.json.

What monitoring can and cannot get back

The arithmetic above is what the outage cost. It is not what monitoring saves, and a page that quietly treated the two as the same number would be doing the thing this one is about to criticise.

No monitor recovers the detection window. Probes run five minutes apart and one failure is a blip, so an incident is raised up to 10 minutes in, and a woken human needs about 5 more. Those 15 minutes bill at full rate whatever you buy. What monitoring removes is everything after them:

What this number leaves out

Stated because a figure is only as good as its edges, and all three of these push the same way — the real cost is higher than the one above:

• Flat-rate averaging. The table divides a month evenly by its minutes. Almost nobody spends evenly at four in the morning, and an outage in working hours bills faster than the average.
• Only the campaigns pointed at the broken thing. Use the spend behind the affected destinations, not the whole account. INC-2291 was 3 campaigns.
• Nothing downstream. No lost revenue, no damaged quality score from a spike in bounces, no cost of the engineer hours, no cost of the conversation with the client. Those are real and none of them are countable, so none of them are counted.

Why we will not tell you what you lost

The tempting number is revenue: conversions you would have made, times value. It is unknowable. It depends on what those visitors would have done, which nobody observed, because the page was broken. Any tool printing that figure is multiplying your historical conversion rate by traffic and calling the product a measurement.

So the interface says unknown where a number requires a guess, and prints only the one it can defend. That is a worse headline and a better product — the first time a customer checks a figure and finds it invented, every other figure becomes suspect.

When landing page monitoring is worth paying for

There is a number and it is computable, so here it is: the entry tier pays for itself at $31,680 a month of spend, and the tier with cost guardrails and an escalation rota at $159,680 — both derived from the same model as the calculator above, and both shown with their working on the pricing page. Below the first of those, a rare afternoon outage costs a few hundred dollars and monitoring it is not worth a subscription. We would rather say so than sell you something. It changes immediately for an agency, where the same failure is not an expense but a conversation with a client about their money.

The spend only runs up if nothing notices, which is the case more often than it sounds: most of these pages answer 200 OK the whole time they are broken, so an uptime monitor stays green through all of it. What catches them instead is landing page monitoring: the same fetch a customer makes, read the way a customer would read it.

A dead page is only one of the ways a campaign stops buying anything while the invoice keeps running. PPC monitoring covers the rest of them — the number that stops arriving, the cost that passes a line you set, the two systems that disagree about the same figure.