SEO & AEO
This is not a question of choosing one but of setting the order and the weighting. We separate the stretches where one fix serves both from the stretches that have to go separately, and decide where resources go first according to the state the site is in.
In one paragraph
Which of SEO and AEO comes first is decided by the state the site is in. On a site that is not indexed, the document that would be cited never reaches the candidate pool; on a site with rankings but no citations, more SEO does not get you inside the answer. This page covers only the criteria for allocating resources by separating where the two overlap from where they diverge. Terminology and conceptual comparison are handled in a separate article.
The overlapping stretches are crawlability, indexing, information architecture, entity and freshness; the diverging ones are performance metrics, document structure and unit of measurement. The five overlapping lines serve both sides once fixed, so handling them before worrying about allocation costs less. The three diverging lines blur the judgement of performance when mixed, so they are recorded separately from the start.
Last verified
WHEN YOU NEED THIS
These are the situations we hear repeatedly in consultations. If any of them apply, start by measuring.
It is not a switch. Google AI Overviews selects its evidence documents with Google's ranking system and Naver AI Search runs on Naver's index, so a document missing from the index does not reach those engines' candidate pool. Deciding where the shared foundation ends and the divergence begins comes first.
Set the ratio before measuring the current state and the overlapping work gets paid for twice. Filter out the items that serve both once fixed — robots.txt, rendering, sitemap — and the share that actually needs dividing shrinks.
A conflict that arises because the metrics differ. One speaks in rankings and traffic, the other in mentions and citations. Rather than merging the two, put them side by side over the same period and the same question set and judgement becomes possible.
Ranking is not an AEO performance metric. If nobody put the questions to the answer engines and counted mentions and citations, that is an SEO report. Checking whether the denominator (questions × answer engines) is stated tells the two apart.
WHAT WE DO
Written as units of work rather than abstract proposals. The scope of an engagement is set from this list.
Crawlability and indexing are where SEO and AEO demand the same conditions. Blocked here, neither line of work produces a result, so we look at it before any allocation discussion.
Document structure and unit of measurement are where the two demand different things. A document tuned to a keyword and a paragraph that answers a question can coexist on one page, but not separating them at design time means rewriting later.
The two confirm performance differently. Merge them into one number and you can no longer tell which work produced the effect, so the measurement conditions get fixed in writing first.
The sequence is decided by the audit result. Which of the three below applies changes where to act first.
Factual information such as business name, address and service description is referenced by both search engines and answer engines. Differing notation per channel shakes ranking on one side and citation on the other, so it is managed in one place.
There is no reason to hold one weighting for a whole engagement. We re-measure under the same conditions, see which side's metrics moved, and shift the next period's share.
PROCESS
What you receive at each stage is stated alongside it. Durations are the working time Navirang controls; they are not a promise about when results appear.
For the same question set we record position in search results and mentions and citations in answer engines separately. The point is to put them side by side rather than merge them.
Comparison table
From crawlability, indexing, information architecture, entity and freshness we pick out the items that serve both once fixed. What is filtered here is not for allocation but precedes it.
Shared foundation checklist
We judge whether it is a no-indexing state, a rankings-without-citations state, or a new site with neither, and set the matching order. The reasoning is recorded so it can be verified later.
Sequencing judgement
Work is divided internal and external, and what each will be judged by — metric and denominator — gets fixed in writing. Where an existing SEO owner exists, edit rights on shared files are settled here.
Division table · metric definitions
We measure again under the same conditions, separate what moved from what did not, and shift the next period's weighting. Bad results are recorded as bad results.
Period-by-period comparison
DELIVERABLES
We do not do work that ends in conversation. The documents below remain, and become the baseline for the next measurement.
Search result position and answer engine mentions and citations for the same question set, side by side. Where the two metrics disagree is where the priority candidates are.
Crawlability, indexing, information architecture, entity and freshness marked pass or fail. Marked separately because one fix serves both sides.
What has to come first now and the reasoning behind it. The reasoning has to be on record for the next measurement to show whether the judgement was right.
What each SEO and AEO metric is counted by, what the denominator of the AEO citation rate is, and on what cadence it is recorded.
Ownership per task, split internal and external. For items where several owners touch one file — robots.txt among them — edit rights are consolidated in one place.
The five overlapping lines are not for allocation but precede it. The three diverging lines make it impossible to tell which work produced the effect if reported together, so they are recorded separately from the start.
HOW IT CONNECTS
Our work moves as one piece. SEO builds the foundation for being found by search engines, AEO raises the odds of that information being cited in an answer, structured data helps machines understand the facts, and content supplies the evidence there is to cite.
| Area | Relationship to this work |
|---|---|
| SEO | The parent area covering all the foundation work in the overlapping stretch |
| Indexing | The premise of the judgement — with no indexing, neither ranking nor citation holds |
| Technical SEO | Actually fixing the shared foundation — rendering, URL format, sitemap |
| AEO audit | Measuring the current value on the citation side — half of the comparison table comes from here |
| Citation monitoring | Confirming by re-measurement under the same conditions whether the split may change |
FAQ
A The state of the site decides. If indexing is not happening, indexing and crawlability come first — in that state more content reaches neither the search results nor the candidate pool for an answer. If rankings appear but AI answers do not cite you, the foundation already exists, so resources move toward document structure and entity consistency. On a new site with neither, building the shared foundation in one pass and writing by question from the first document reduces the cost of rewriting later.
A There is no set ratio. What helps instead is filtering the overlapping items out before dividing, which shrinks the share that actually needs allocating. Crawlability, indexing, information architecture, entity and freshness serve both sides once fixed, so there is no reason for them to be billed twice whichever budget they sit in. What remains after that is SEO-only and AEO-only work, and their weighting is set from which side's metrics were emptier at the first measurement, then adjusted at each re-measurement.
A It does not carry over, because engines gather documents by different routes. Google AI Overviews selects evidence documents with Google's ranking system, but per OpenAI's own documentation ChatGPT Search uses third-party search providers (Bing, Shopify) alongside its own crawler (OAI-SearchBot). So a rise in Google ranking can be the basis on the Google AI Overviews side without explaining citation in another engine. For allocation this is the point: do not infer one engine's performance from another's metric — record engines separately and you can see which moved.
A We do not recommend it. Google AI Overviews selects evidence with Google's ranking system and Naver AI Search with Naver's index, so a document missing from the index is not even a citation candidate in those engines. And the crawlability, indexing, information architecture, entity and freshness that AEO needs are the same items as SEO's foundation work — skipping does not make that work disappear. What is genuinely backwards is doing AEO with ranking as the goal: the foundation is shared while performance is confirmed with each side's own metrics.
A SEO by ranking, click-through rate, organic traffic and indexed page count; AEO by mentioned or not, cited or not, citation share and factual accuracy of the description. When reading an AEO figure the denominator has to be read with it — a citation rate becomes an entirely different value depending on what was taken as questions × answer engines. We measure our own site against 40 questions × 7 answer engines = 280 cells, and that baseline is currently in progress.
A It can. In the overlapping stretch, though, two owners editing the same file overwrite each other's work. robots.txt, the sitemap, structured data and canonical URLs get their edit rights consolidated in one place, with change records shared. We mark the items an internal team or the existing agency can handle separately in the division table, keeping them out of our scope.
A We cannot promise a date. Search engine re-indexing cycles and answer engine document refresh cycles are both external conditions, and answer engine responses differ for the same question depending on the moment and the session. What can be guaranteed is the order and the measurement conditions, not the result. What we promise is that repeated measurement under the same conditions makes it possible to confirm from the record whether anything changed.
We measure search result position and answer engine citation for the same question set and show them in a comparison table. The order gets decided together from that table.
We reply within one business day.