In short: These are not two budgets competing for the same money. Sort the actual tasks and the shared column is the longest one, while neither single-purpose column is long enough to justify a separate strategy.
Fund the shared foundation first, because GEO work sitting on broken technical SEO produces nothing. Then decide the split at the margins using your own data rather than an industry ratio, and set separate success measures for each, because the same number cannot judge both.
Most Of The Work Is Shared
The framing of SEO against GEO is mostly a marketing artefact. Sort the actual tasks by who they serve and the overlap is immediate.
| Serves both | SEO only | GEO only |
|---|---|---|
| Crawlability and indexing | Title and meta description tuning | Prompt-based visibility testing |
| Server-rendered HTML | SERP feature targeting | AI crawler access policy |
| Information architecture | Click-through rate optimisation | Answer-first page restructuring |
| Genuine expertise and authorship | Pagination and faceted navigation | Entity consistency across third parties |
| Earned links and mentions | Core Web Vitals tuning for ranking | Per-platform citation logging |
| Structured data | Sitemap management | Original data publication |
| Original, useful content |
Read the middle and right columns honestly and something becomes clear: neither is large enough to justify a separate department, a separate strategy document, or the argument about which one wins.
The sequencing rule: the shared column is not one item among three, it is the prerequisite for the other two. GEO work built on a site that crawlers cannot read produces nothing at all, not a reduced result. Fund the middle column first and completely.
Where The Genuine Tension Sits
Three conflicts are real, and pretending otherwise makes the advice useless: page length, conversion copy and crawler access.
Length
Comprehensive pages have historically performed well in search. Under retrieval, length that dilutes the answer costs you at the selection step.
The resolution is structural rather than a compromise on word count: keep the depth, and put a self-contained answer near the top so the extractable passage exists regardless of how long the page is. That is what the summary block at the top of every lesson on this site does.
Conversion copy
Persuasive writing is deliberately vague where specifics would invite objections. "Trusted by leading brands" is a conversion sentence and an uncitable one. Systems cannot repeat it because it asserts nothing checkable.
The resolution is not to strip persuasion out. It is to make sure that somewhere on the page, the specific version also exists: which brands, how many, since when. Both readers get served, and the vague version stops being the only version.
Crawler access
Blocking AI crawlers protects content from being used without compensation and removes you from the answers those systems generate. There is no configuration that gets both. This is a business decision about whether visibility or control matters more for your specific content, and it should be made explicitly rather than inherited from a template.
Splitting The Margin
Once the shared foundation is funded, some effort remains to allocate. Industry ratios are worthless here because the right answer depends on facts only you have. Four of them:
- How much AI-referred traffic you already see, and its trend. Filter your analytics for the assistant domains. Rising from a small base with a high conversion rate is the strongest signal to lean in.
- Whether assistants describe you correctly. Ask five of them what you are and who you are for. A wrong description is urgent regardless of traffic volume, because it is being served to buyers today.
- How your buyers actually research. Technical and professional audiences moved earlier and further than general consumer audiences. Ask ten customers rather than reading a survey about a population that is not yours.
- The state of your technical foundation. If the audit in What Stays the Same found real problems, that is where the marginal hour goes, and it counts as both.
A reasonable default for a business with a working technical foundation and early AI-referred traffic: keep conventional search work at its current level, and fund GEO from new budget or from whatever you are currently spending on tactics you cannot defend.
Cutting proven search work to fund an unproven channel is the trade that most often goes wrong, and it goes wrong slowly enough that nobody attributes it correctly.
Measure Them Separately Or Not At All
The most common failure in combined programmes is not the split. It is judging both with one number, which guarantees that whichever is working looks like it is not.
- Search work is judged the way it always was: impressions, positions, organic sessions, conversions from them.
- GEO work is judged on mention frequency across repeated fresh sessions, per platform, plus the accuracy of how you are described, plus referral traffic from assistant domains.
- Shared work shows up in both, and should be reported in both rather than assigned to one.
Never blend the platforms into one score. Given how little the platforms overlap in what they cite, an average across them removes the only information the number contained.
Build The Plan In An Afternoon
- List everything currently in your search and content backlog. Sort each item into the three columns above. The shared column is the longest of the three, and that is visible in the table rather than a matter of opinion.
- Move every shared item you have blocked or deferred to the top. That is your first sprint, and it needs no argument about GEO to justify it.
- Run the four marginal questions above and write down the answers with dates. Your split follows from those, and you can revisit it in a quarter against the same four.
- Set up two reports, not one. Search metrics in the first, mention frequency and description accuracy in the second, per platform.
- Pick one deliberately single-purpose GEO project for the quarter. Original data is usually the highest-leverage choice, because it is the only work that cannot be summarised away from you.
- Diarise a review at the end of the quarter, against what you wrote down rather than against what you remember expecting.
Key Takeaways
- The shared column is the longest of the three, so the argument about which discipline wins is mostly a false choice.
- The shared foundation is a prerequisite, not a third of the effort. GEO on a broken technical base returns nothing rather than less.
- Real tensions exist in three places: page length, conversion copy and crawler access. Each has a structural resolution rather than a compromise.
- Split the margin using your own traffic, description accuracy, buyer behaviour and technical state, not an industry ratio.
- Report the two separately, per platform, or the one that is working will look like the one that is not.
Check yourself
Before you move on
Not scored, not recorded, and not part of the certificate. Both answers are settled by a sentence in this lesson, and the reasoning appears whichever option you pick.
- 01
You have a working technical foundation and early AI-referred traffic. What is the defensible budget move?
- 02
Comprehensive pages have historically performed well. What does retrieval change about that?