An AI marketing strategy that survives contact with reality

Find what actually binds first

Map how demand reaches revenue today: where enquiries come from, how fast they are answered, what proportion convert, and where they stall. Then find the narrowest point.

It is very often not content. In our experience the most common real constraint is speed to lead, and the second is qualification. A business answering enquiries in two days does not have a content problem, and adding content to it produces more unanswered enquiries.

More content applied to a follow-up problem produces more content and the same revenue.

Rank everything against three things

There are only three ways to increase profit: get more customers, make each customer worth more, or cut what it costs to deliver. Score every candidate against those, and park anything that does not land clearly in one.

This test is also how you should judge proposals from us. Ask which of the three we are moving and roughly by how much. A good answer is specific enough to be checked later.

Build the voice layer before the production layer

This is the step almost everyone skips, and skipping it is why so much AI content gets rewritten before publishing. A rewrite puts the cost straight back where it was removed from, and the team correctly concludes the tool did not help.

The voice layer is training on your actual published material, defined personas, product guardrails, a banned-phrase list, and a test where output is compared against your existing site. It takes a week or two and it determines whether everything after it is usable.

Ship one channel end to end

One channel, running properly, with a human approval gate, before widening. It proves the quality bar holds under real conditions and it gives you something to point at internally, which matters more than it should.

Widening early is the most common way these programmes fail. Four half-working channels produce less than one working one and are much harder to diagnose.

Measure revenue, not output

More content is not a result. The numbers that count are qualified enquiries, cost per acquisition, time from enquiry to first response, and hours returned to the team.

Set them before you start, because measuring after the fact invites the metric that makes the project look successful. If they have not moved within a quarter, the strategy was wrong. Change it rather than defending it.