SEO automation: what to automate and what to leave
SEO automation is worth using for work that is mechanical and repetitive: crawling, technical auditing, keyword research, checking content against a fixed standard, publishing, and reporting. It is worth avoiding for work that needs judgement or first-hand knowledge: deciding what your business actually does, supplying facts only you know, and setting the standard the checks apply.
The common mistake is not automating too much. It is automating the wrong half.
The parts that automate well
Crawling and technical auditing
A crawl is the same job every time. Fetch each page, follow the links, record what broke. It is tedious, it needs doing constantly rather than quarterly, and nothing is lost by having a machine do it.
This is the least controversial automation in the field and still the most under-used, mostly because a quarterly audit is easier to sell than a nightly one.
Research
Search volume and difficulty are already numbers. Sorting opportunities by what they could bring against what they would cost is arithmetic. A person doing this by hand is doing arithmetic slowly.
Checking against a fixed standard
This is the one that matters, and the one most tools skip.
A machine applies the same threshold at three in the morning on a Sunday that it applied on Monday. It does not get tired near a deadline, it does not decide that a piece is close enough, and it does not have a stake in the work passing.
That consistency is the actual argument for automation. Not speed.
Publishing and reporting
Once the decision to publish is made, everything after it is clerical.
The parts that do not
Anything needing first-hand knowledge
Your prices. Your process. Why a customer chose you over the firm down the road. A model that does not know these will produce something plausible in their place, which is worse than leaving a gap.
The correct behaviour is for a draft to flag what it does not know and stop. Ours does; you get a note saying which two paragraphs need you.
Judgement with commercial consequences
Whether to compete for a term that would bring traffic you cannot serve. Whether a claim about your work is defensible. These are decisions with a person’s name on them.
The standard itself
The checks run automatically. Deciding what belongs in them does not. Ours is reviewed by a person continuously, because the list of clichés worth banning changes and the right sentence length is a judgement, not a constant.
The volume trap
Most SEO automation is sold on output. More articles, more pages, more posts, per month, per dollar.
Volume is the easy part to automate and the least valuable. Publishing forty weak pages gives a search engine forty reasons to think less of the site, and gives a language model forty unremarkable sources.
The useful measures are different, and duller:
- Does the work run when nobody chases it.
- Is there a log you can read afterwards.
- Does anything below the standard get stopped before it publishes.
A tool that publishes everything it drafts has automated the wrong half.
A reasonable setup
- Crawl nightly, fix what is safe automatically, flag the rest in plain language.
- Research continuously, rank by realistic opportunity rather than by volume.
- Draft against a written standard, not against a word count.
- Check every draft against that standard mechanically, with a pass or reject verdict and numbered reasons.
- Publish only what passed, and keep the log.
- Have a person own the standard and review it.
Five of those six are automatable. The sixth is the one that decides whether the other five are worth anything.