The in-house versus agency question is usually framed as cost. The variable that predicts success is whether the model can get changes deployed.
SEO recommendations die in the implementation queue
Most programmes fail because the work is never built, not because the analysis was wrong. An agency with no route into the engineering backlog produces excellent documents that nobody actions. An in-house hire with no developer time has exactly the same problem for more money.
Ask where the developer capacity comes from before choosing a model
If the answer is "we'll raise tickets", the programme will move at the speed of whatever else that team is doing. Securing a standing allocation of engineering time matters more than any decision about who does the SEO thinking.
The hybrid usually wins
In-house ownership for prioritisation, relationships and institutional knowledge; external specialists for technical audits, migrations and capability the team does not have year-round. The in-house person's main job is getting work shipped, not producing analysis.
Scale decides the answer
Below a certain size, a fractional consultant beats a junior hire because seniority matters more than hours. Above it, the coordination load alone justifies someone internal. The awkward middle is where most bad hires happen.