Marketing in the Age of AI

Marketing teams adopted AI for drafting quickly, but few trust it with decisions about what to build or who to sell to.

AI has made it cheap to generate marketing content. Anyone can produce a competent post or a plausible positioning line in seconds. The thinking stayed hard: understanding demand, choosing where to play, and telling a good answer from a plausible wrong one. The tools cover the easy, visible parts of the job and leave the valuable work almost untouched.

This series covers that argument in four pieces. The short version of the argument comes first, and each section notes where the full piece takes it further.

Part one: the marketing value chain

Advertising, social and content are the visible part of a much bigger discipline. Marketing is the craft of understanding demand and organising a business to meet it profitably. The whole job rests on three questions: who is the customer and what do they need, what proposition do you build for them, and how do you deliver it and get paid.

That job has five pieces: understanding demand, strategising where to play and how to win, creating the product and proposition, communicating through content and channels, and measuring what drove revenue. The familiar AI marketing tools work in the communicate box: they generate the post or the variant. Those tools are useful, and most of them do the same job. The hard, high-stakes work is in understand, strategise and measure.

Part one lays out the full value chain, the three growth levers every plan reduces to, and why automating only the communicate box scales activity without improving the decisions that make the output worth anything. Read the full piece: The Marketing Value Chain.

Part two: how AI changed each box

AI compressed cost and time in every box of that chain. That saving is worth having. However, the failure mode is the same in every box: a plausible output with nothing to check it against.

Each box has its own version. In understand, the model produces a clean-looking ideal customer profile that is confidently wrong: a buyer who looks real and is not yours. In strategise the failure is a positioning draft that reads well and differentiates nothing, and in create it is polished, interchangeable copy. In communicate the failure is sending the wrong accounts more messages than ever, and in measure it is optimising lead counts while revenue drifts.

AI applied to a bad strategy scales the bad strategy. Part two maps the work before and after AI and gives you the questions to ask of any AI workflow you build: which box it is in, and which check catches its failure mode. Read the full piece: Before and After AI.

Part three: judging B2B positioning

Positioning is the box where a plausible wrong answer is hardest to detect. A model asked to position a B2B product returns a fluent line, and nothing on the page tells you whether it is any good.

In B2B, positioning's job is to move a buyer closer to preferring you as a partner. Being top of mind with buyers is the most important factor, and the strength of the client relationship is the second. A useful line says something about the relationship and the experience of working with you, and makes a claim that sets you apart from rivals. A clever tagline on its own does neither. The test is whether a competitor could truthfully claim the same line.

Part three covers what positioning is for in B2B, why B2B buying differs from B2C (multi-party decisions, long lags, a short, gated list of vendors) and how to use the test as a constraint every time you ask a model for a positioning line. Read the full piece: B2B Preferred-Partner Positioning.

Part four: the evaluation function

One idea underlies the first three parts: the evaluation function. An evaluation function is your judgement of what a good output looks like, written down as a test the model can apply, and a model does not have yours unless you supply it.

A prompt is judgement you retype and trust again every time; it is invisible and unversioned. Our working guide for marketing teams covers writing that prompt well and building a reusable skill from it, step by step. A skill is that judgement written down once, version-controlled, with a gate: a point where the tool stops and names what it is missing, such as the ideal customer profile, the competitor it beats or the demand it captures. A tool without a gate produces a confident answer at that point. The gate's refusal is the evaluation function at work. A weighted driver model is the pre-AI version of an evaluation function, with explicit drivers, weights and rules, and a machine can apply it directly.

Part four sets out the five steps: name the failure mode, write the test that catches it, add a gate, keep the reasoning, version it. When generation is cheap, that discipline is the advantage. Read the full piece: The Evaluation Function.

Practical steps for marketing leaders

The four parts lead to three practical steps. Point AI at understand, strategise and measure, and treat the communicate box as a commodity. Before you ship any AI output, name its failure mode and write the test that would catch it. Write your judgement down once, as a skill with a gate, so each use adds to it and it no longer depends on memory.

Judgement about what is good stays with you, and it is worth more now that generation is cheap. For the wider backdrop, see the intelligence curve.

The marketing skills pack includes a positioning skill with the gate described above.