Three capabilities, usually in three different teams
The barrier to doing GEO in-house is not budget. It is that the required skills rarely coexist in one person or team, and a project missing any one of them tends to become SEO copy with new headings.
- Engineering: robots, schema, rendering, site structure, crawl log analysis
- Content: turning business knowledge into structure an answer can be assembled from
- Data: fixed question sets, before-and-after comparison, anomaly spotting
The cost that is easy to leave out
Count tools, learning time — model and retrieval behaviour change quickly and have to be tracked — and opportunity cost. Engineers on this are engineers not building the product.
If it needs one to two people continuously on a small team, that is usually already beyond a reasonable in-house scope. Converting that fixed cost into a variable one is often the better trade.
What to require from a supplier
Projects fail on unclear deliverables more often than on wrong method. Writing down what will be delivered, before signing, is worth more than arguing afterwards.
- Before-and-after results on the same question set, with full response text
- Each conclusion tied to a platform, a timestamp and a specific prompt
- Clear separation between verified findings and open items
- A fix list that names pages and positions, not 'improve content quality'
Warning signs you need no expertise to spot
These show up during ordinary conversations and get talked past.
- Promising rankings or guaranteed citations — model answers cannot be promised
- Screenshots without the raw responses, so nothing can be reproduced
- Cannot state the measurement: how many questions, which platforms, when
- Conflates being mentioned with being recommended
- Treats domestic and international engines as interchangeable, or claims one approach covers every platform
The usual answer is a hybrid
Most teams do not need to choose between fully in-house and fully outsourced. Buy the diagnosis and the baseline, then execute internally.
The test is simple: diagnosis happens once, execution happens continuously. Capability has to be built, not bought.
In-house and outsourced differ on whether execution is sustainable. The diagnosis can be bought; the capability has to be built. Sources reviewed on 2026-10-17. This article records a method with verifiable boundaries; it does not claim that any search or AI platform will index, cite or recommend anything.