When Your Best Channel Stops Working
One channel produces most of your enquiries and has done for three years. That is not a strength. Ask anybody who was running social ads in 2021.
Most of your work comes from one place. Referrals, or search, or one advertising platform, or a single partner. It has worked for years and nobody has felt any need to change it.
That is a concentration risk, and it is the kind that produces no warning before it arrives.
The example everybody lived through
Apple introduced App Tracking Transparency with iOS 14.5 on 26 April 2021, requiring apps to ask permission before tracking users across other companies’ apps and websites.
The change was announced in advance and was not aimed at any individual advertiser. Its effect on businesses relying on precise social targeting was immediate: the signal that made that channel work degraded, costs rose, and measurement stopped agreeing with reality.
Nobody did anything wrong. The ground moved, and the businesses that suffered most were the ones for whom that channel was not a channel but the entire pipeline.
Channels do not die, they get worse
A clean failure would be easier to respond to. What actually happens is a slow erosion that is difficult to distinguish from a bad quarter.
Costs drift up. Volume drifts down. Quality gets slightly worse. Each month is explicable on its own, so the response is to try harder within the channel rather than to conclude that something structural has changed.
By the time it is undeniable, you have lost a year and you are starting a new channel from zero under revenue pressure, which is the worst possible condition for learning one.
Three ways it happens
A platform changes. A privacy rule, an algorithm change, a policy that reclassifies your category. Outside your control and usually not announced in terms you would notice.
A person leaves. The most underestimated version. If most of your referrals come through one contact at one firm, that is not a channel, it is a relationship, and it ends when they move.
Competition arrives. Somebody with more budget enters your search terms or your platform, and the economics stop working for you specifically rather than in general.
Test the second channel before you need it
The right defence costs about ten percent of your marketing budget and is almost never spent, because the money always has a better immediate use in the channel that is working.
Run something small and permanent in a second channel. It does not need to be profitable. It needs to exist, to have been learned, and to be capable of being scaled inside a month when it has to be.
The reason this matters is that starting a channel takes six to nine months to understand, and the emergency version is always more expensive and worse. Ten percent spent now buys that learning in advance.
The one that cannot be taken from you
Everything that depends on a platform is rented. Two things are not.
A list of people who have chosen to hear from you, with their contact details in a system you own. And a body of material that ranks and gets sent, on a domain you control.
Neither can be reclassified, deprioritised or repriced by somebody else’s policy change. Businesses that came through 2021 in reasonable shape were disproportionately the ones with both.
Where AI helps you see it early
The problem is detection. Slow erosion looks like ordinary variance, and nobody is watching a channel closely enough to distinguish them.
AI separates decline from noise. Comparing against the same period last year, controlling for seasonality and for what else changed, is exactly the analysis nobody runs monthly because it takes a day. Run continuously, it flags a real trend while it is still a trend.
AI reads quality, not just volume. A channel frequently degrades in the fit of the enquiries before it degrades in the count. Reading what people actually send tells you months earlier than any cost figure.
AI makes the second channel affordable. The reason businesses run one channel is that each one needs attention nobody has. When the responding, following up and reporting are handled, running three becomes possible for a team that could only manage one.
That last point is the real answer to concentration risk. Diversification failed for most small businesses on capacity grounds rather than strategy, and capacity is the constraint that changed.
Do not abandon it too fast either
The opposite error is worth naming, because it is expensive in a different way.
A channel having a poor quarter is not a channel dying, and businesses that switch at the first bad run spend years starting things and finishing none. The test is whether the decline holds against the same period last year, and whether the mechanism has changed or only the numbers.
Build the second channel because concentration is a risk, not because the first one had a bad March.
Know your concentration number
Most owners overestimate how diversified they are, because they count channels that exist rather than channels that produce.
Take last year’s new clients and attribute each to where it genuinely originated, not to the last thing they clicked. Then work out what share came from the largest single source.
Above about sixty percent, a change to that source is a serious event for the business rather than a bad quarter. Most firms that run this calculation for the first time find they are well above it and had not realised, because referrals feel like many sources rather than one channel.
Do the same for the referral base specifically. If three people account for most of it, that is three relationships, and two of them will move jobs within five years.
The question to answer this quarter
If your largest source of enquiries produced half as much next month, what would you do.
If the answer is a plan you could execute, you are fine. If the answer is that you would try harder in the same channel, you have a concentration problem, and the time to fix it is while the channel is still working.
Sources
Apple, App Tracking Transparency. Introduced with iOS 14.5, 26 April 2021.
AI Optimize handles the responding, following up and reporting that made running more than one channel unaffordable. That work sits under Paid Ads Management.
Related reading
How Much Should You Actually Spend on Ads
Not a percentage of revenue, and not what you can afford. Work back from what a client is worth, then check whether the number is large enough to buy an answer.
What Apple's Privacy Change Did to Your Ad Targeting
The single largest shift in paid acquisition this decade was not a platform update. It was a permission prompt, and most advertisers still run accounts built for the world before it.
WHAT WE BUILD