CoreWeb
Useful
research, analysis, workflow
Not useful
publishing unedited output
Disclosed
we tell you where we use it

AI Digital Marketing

AI is genuinely useful for some marketing work and actively harmful for others. Most agency positioning on this is either breathless or defensive, so here is a plain account of where we use it and where we do not.

Digital Marketing

Channels that surround search, measured on one model instead of six.

Real gains
Mostly unglamorous.

Where it helps

Processing scale. Clustering tens of thousands of queries into topics, categorising a large content inventory, spotting patterns in log files, summarising competitor coverage across hundreds of pages. Work that was previously either sampled or skipped because it was uneconomic.

Drafting and structuring. First-pass outlines, restructuring existing material, generating variant copy for testing. Useful as a starting point that a person then does real work on.

Analysis support. Interrogating datasets conversationally, which surfaces questions you would not have thought to query.

Real limits
And why it matters commercially.

Where it does not

Publishing unedited output. It reads generically, it is increasingly detected and discounted, and it does not get cited by the very systems that produced it. Volume content produced this way is a liability on your domain rather than an asset.

Judgement calls. What to prioritise, what a client should not spend money on, whether a finding is significant or an artefact. These are the parts worth paying a human for and the parts a model will confidently get wrong.

Anything where being wrong is expensive. Regulated content, technical specifications, factual claims about your products. We use AI to draft and humans to verify, never the reverse.

7 areas
One body of work, not a menu.

All services

What’s included

Query and topic clustering at scale

Processing large search datasets into usable topic structures, which manual work can only sample.

Content inventory analysis

Categorising and assessing hundreds or thousands of existing pages to find decay, duplication and gaps.

Log and dataset analysis

Pattern detection across volumes of technical data that would otherwise be sampled.

Creative variant generation

Producing testable variations for paid campaigns and landing pages, where volume of options genuinely helps.

Workflow automation

Reporting assembly, alerting and routine analysis, freeing time for work that needs judgement.

AI visibility work

Which is a different service — see AI SEO and LLM Optimization.

Tooling guidance

Advising which AI tools are worth your budget, which mostly are not, and what to keep in-house.

Common questions
Asked on most first calls.

Questions

Will you use AI to produce our content?

For research, outlining and structure, yes. For published prose, no. We will always tell you which parts of a deliverable involved AI.

Does AI-written content rank?

Sometimes initially, poorly over time, and it does not get cited by assistants. Google’s position is about quality rather than authorship, and unedited output is rarely high quality.

Can AI replace our agency?

It can replace some tasks agencies used to bill for, which is why our pricing reflects judgement rather than hours. It cannot yet decide what matters, and it will assert wrong answers with complete confidence.

Should we build AI into our own marketing stack?

For specific repeatable tasks, often yes. As a general strategy, usually no. We are happy to advise on which of your workflows would genuinely benefit.

Is this just a rebranded service?

Partly, and we would rather say so. The genuinely new work is AI search visibility. The rest is using better tools for existing work, which does not really warrant its own product name.

Next step
We look at your site before the call.

Start with a free audit

An automated crawl plus a 30-minute walkthrough of what we can see from outside your site. If what you need isn’t something we do well, we’ll tell you.