When a channel stops working, diagnose before you replace it

Something that worked for two years stops working. Cost per lead doubles, or the leads stop converting, or the volume just falls away.

The instinct is to declare the channel dead and move the budget somewhere else. Sometimes that is right. Usually it is expensive and premature.

Find out which number moved

A channel is a chain: impressions, clicks, enquiries, qualified enquiries, sales. Something broke at one link, and the fix depends entirely on which.

Impressions fell. Supply or eligibility. Budget capped, bids uncompetitive, an algorithm change, a policy issue, a disapproved asset, or seasonality.

Clicks fell but impressions held. The creative or the messaging stopped landing, or you are now showing to a worse audience.

Enquiries fell but clicks held. The problem is on the site, not in the channel. A form broke, a page changed, load time degraded, or the offer lost relevance.

Enquiries held but quality collapsed. Targeting drifted, or the campaign is now attracting a different intent. Very common after a platform automation change.

Everything held but sales fell. It is not a marketing problem. Look at pricing, at the sales process, or at the market.

That single diagnosis eliminates most of the wrong answers, and it takes an hour.

The causes worth checking first

Something changed on your side. A website release, a tracking change, a form edit, a page URL change. Check your own deployment history against the date the number moved. This is the answer far more often than anyone expects.

Tracking broke. Performance did not fall, measurement did. Check whether sales actually fell, in the books, not in the dashboard. Chasing a fix for a reporting fault is a specific and common waste of a quarter.

Auction pressure increased. More competitors, or one competitor spending heavily. Visible in impression share and in cost per click.

Creative fatigue. The same asset shown to the same audience for long enough stops working. Predictable, fixable, and it tends to be gradual rather than sudden.

A platform change. Automation defaults, attribution model changes, privacy changes. These move things without warning.

The market moved. Seasonal, economic, or a genuine change in how buyers behave.

Sudden versus gradual tells you a lot

A sharp drop on a specific date is almost always something that changed: a release, a policy action, a tracking break, a platform update. Look for an event.

A slow decline over months is fatigue, auction pressure, or drift. Look for a trend.

Treating one like the other is how the wrong fix gets applied for six weeks.

Before you abandon it

Ask whether the channel ever worked for structural reasons, or whether it worked because of a specific campaign, offer or moment that has passed.

A channel that worked because of one exceptional creative is not a channel that works. It is a creative that worked, and the fix is another creative, not another channel.

And ask what it would cost to rebuild elsewhere. Starting a new channel costs a learning period, new creative, new tracking and usually a quarter before it is informative. That is the real comparison, and it is frequently more expensive than fixing what exists.

When it genuinely is dead

Some channels do end. The audience left, the economics stopped working, the platform changed in a way that cannot be worked around, or the cost of acquisition permanently exceeded what a customer is worth.

The test is whether you can articulate why, mechanically. If the explanation is "it just stopped working", the diagnosis is not finished.

The protection against it

No business should be dependent on a single channel, and most are.

The point is not to spread thinly across five. It is to have a second channel that is genuinely working at some scale, so a failure in the first is a problem rather than a crisis, and so you have somewhere to move the money that does not need a quarter of learning first.