Marketing — Charter
What marketing owns in an AI-operated company — positioning, demand generation, attribution, and why generating more content is usually the wrong response to having AI.
Marketing owns how the company is understood and how demand arrives. AI makes producing content almost free, which sounds like an advantage and is mostly a trap: when everyone can produce volume, volume stops working.
The functions that got more valuable are the ones AI does not do — a position worth holding, original evidence, and a reason to be believed.
What this role owns#
Positioning. What we are, who for, and what we are not. Everything else follows.
Demand generation. Making the right people aware, and giving them a reason to act.
Content that is worth reading. Which increasingly means content with something in it nobody else has.
Attribution. Where demand actually comes from, measured rather than assumed.
Brand consistency. How the company looks and sounds, everywhere.
KPIs#
| Measure | Why this one |
|---|---|
| Qualified leads, by source | Not traffic. Leads a salesperson would accept |
| Cost per qualified lead, by channel | Where the money works |
| Conversion, visit → lead → qualified | Where the funnel actually leaks |
| Organic search growth | The compounding channel, and the slowest |
| Content performance concentration | Usually a small number of pages produce most results. Find them, make more like them |
| Brand search volume | Whether people are looking for you — the honest brand measure |
| Attribution coverage | Share of leads whose source is actually known |
Why more content is the wrong instinct#
The economics changed in a specific way:
Generation cost fell to near zero, so generated content has near-zero scarcity value. Search engines and readers both adjusted. Thin content now actively harms a domain rather than merely underperforming.
What became scarcer: original data, genuine expertise, an opinion that could be wrong, and anything the author had to do work to know.
The practical rule: publish something only if it contains something nobody else could have written. Original measurement, a position taken, a mistake described honestly, a process that actually runs. Everything else is noise you are paying to host.
A useful test before publishing: what does this contain that a reader could not get from a general AI assistant in ten seconds? If the answer is nothing, do not publish it.
AI agents in this function#
Research agent — market, competitor and topic research with sources.
Draft agent — first drafts from an outline and a stated position. The position comes from a person.
Repurposing agent — turns one substantial piece into the formats each channel needs. This is the highest-return use: the expensive part was the thinking, and it should be reused.
SEO analysis agent — keyword and structure analysis, internal linking gaps.
Performance analysis agent — what worked, what did not, and what the pattern suggests.
What stays human: positioning, the opinion in any piece, original research, anything published under a named byline, and the decision that something is good enough to publish. The last one is where quality is actually controlled.
SOPs#
- Publishing standard — every piece has a stated audience, a position, and something original. Nothing publishes without passing that test.
- AI-assisted content rule — drafts may be generated; facts are verified, positions are human, and a person is accountable for what goes out.
- Content structure — consistent format so readers and search engines can both parse it.
- Attribution setup — every campaign and channel tagged before launch, never retrofitted.
- Quarterly performance review — what to stop, what to do more of.
Templates#
Content brief · campaign plan · positioning statement · publishing checklist · SOP format for recurring campaigns.
Workflows#
In: product changes · customer insight from Sales and Support · market signals · performance data.
Out: qualified leads to Sales · published content · positioning · campaign results · market feedback to Product.
Handoffs: Sales for lead quality — a loop, not a one-way pass · Legal for claims and disclaimers · CEO for positioning changes.
FAQ#
Should we publish AI-generated content?#
Drafts, yes. Published unreviewed and unenriched, no — not mainly for policy reasons but because it contains nothing a reader could not get elsewhere in seconds, and thin content damages the domain that hosts it.
How much content should we publish?#
Less than you think, better than you currently do. One substantial piece that gets cited outperforms twenty that get skimmed, and it keeps performing.
How do we compete when everyone has AI?#
With what AI cannot supply: original measurement, a stated position that could turn out wrong, real failures described honestly, and distribution. Those were always the differentiators — the change is that everything else stopped being one.
What is the fastest improvement available to most marketing functions?#
Attribution. Most cannot say where their qualified leads come from, which means budget is allocated on belief. Tagging everything before launch costs almost nothing and changes what you stop doing.
What else is coming for Marketing
Charter Ready
What this department owns and is accountable for.
KPIs Not yet
The numbers it is judged on.
AI Agents Not yet
What is automated, and what stays human.
SOPs Not yet
How the recurring work is done.
Templates Not yet
The documents it produces.
Workflows Not yet
How work enters, moves and leaves.