Design the prompt set
For your category, products, and buying situations, we build a set of questions users actually ask AI. That set is the measurement target for every later cycle. It is not changed casually, so results stay comparable.
SEO & GEO Solutions
Does your brand exist in AI answers?
SEO decides which rank you occupy in search results. GEO decides whether AI mentions you when it answers a question. We measure where you stand first, then optimize page by page.
What it means to be seen has changed
For the last decade, “being found” meant “ranking in the top ten links.” Users clicked through, compared, and decided.
Now users ask AI and get one answer. Who is mentioned and who is cited in that answer is the new number one. And there is only one slot.
This has three layers of impact for a business:
You do not know whether you are in the answer.
Rankings have tools. AI answers do not. Ask the same question three times and the answer may change. Without data, you cannot tell.
Being mentioned is not the same mechanism as ranking.
Rankings come from authority and links. Citations come from content that can be understood, sliced, and safely restated. Ranking first and never being cited is common.
A competitor may already be inside the answer.
Not because they spent more, but because their content structure happens to be citable.
GEO is not a new SEO. It is a new measurement object. SEO measures rank. GEO measures who is mentioned and who is cited when AI answers.
No optimization without an audit
We do not start by rewriting content. We start by measuring. Without a baseline, no later change can be shown to work.
Establish a baseline
Run the full prompt set on AI platforms and record three metrics—Mention (is the brand named), Citation (is your site cited as a source), and Share of Voice (your share versus competitors on the same question).
Gap analysis
Separate “not mentioned” from “mentioned but not cited.” Those are different problems with different fixes. Split technical issues (structured data, crawlability) from content issues (whether a paragraph can be cited on its own).
Optimize page by page
For named pages we deliver work that can be implemented, not a list of suggestions.
Re-measure on a cycle
Rerun the same prompt set and compare. Whether you can prove an effect depends on whether step 1 was done well.
Measure first, then decide what to change
Lite
NT$300,000/ quarter
One purchase per client
- 100 Prompts
- 4 hours consulting
- 3 measurement points
- GEO strategy report 1
- Monthly monitoring reports 2
Basic
NT$600,000/ year
- 500 Prompts
- 6 hours consulting
- 12 measurement points
- GEO strategy report 1
- Monthly monitoring reports 11
Focus Study
NT$180,000/ topic
Add-on for Basic customers only
- 50 Prompts
- 2 hours consulting
- 3 measurement points
- GEO strategy report 1
- Post-test closing report 1
Coverage
- Lite and Basic each cover one domain and one language
- Focus Study covers one domain, one language, and one topic
- Monthly prompt count is the number of prompts, not multiplied by how many platforms you measure
About consulting hours
Consulting hours are used for report walkthroughs and the matching monthly consulting meetings. After the hours are used up, additional consulting is NT$6,700 / hour.
The plan tells you what to change. This is where the work gets done.
Plans deliver reports and direction. Actually changing pages is a separate work item. You decide how much to do from the plan findings.
| Item | When it applies | Deliverables |
|---|---|---|
| Content Refinement | Existing pages need structured data, head settings, and paragraph wording adjusted | Technical Schema, Technical Other, copy-paragraph optimization sheet |
| Content Creation | There is too little citable content and new pages are needed | Technical Schema, Technical Other, copy-paragraph optimization sheet |
Implementation by us: deliverables are meant to be deployed by your team or site vendor. If you need us to implement, we bill actual hours worked.
Deliverables
From the plan
| Deliverable | What it contains |
|---|---|
| GEO strategy report | Prompt-set performance, Mention / Citation / Share of Voice baseline, competitor comparison, gap analysis, priorities, and execution cadence |
| Monthly monitoring report | Rerun the same prompt set and compare how each metric moved |
| Post-test closing report | Before/after comparison and conclusions after a topic cycle ends |
From implementation items
| Deliverable | What it contains |
|---|---|
| Technical Schema | Complete JSON-LD for that page, ready to deploy |
| Technical Other | Head-level and template-level code changes |
| Copy-paragraph optimization sheet | Seven-column spreadsheet: block / original / optimized copy / target prompt / CMS field length limit / current length / notes |
The copy-paragraph optimization sheet has a column for your CMS field length limit. That means every optimized paragraph we give you is already a length an editor can paste into the CMS without truncation. That is the difference between “consulting advice” and a deliverable you can implement.
A good fit
- Categories where buyers ask AI before they purchase (B2B technical products, professional services, high-ticket durables)
- Brands that already invest in SEO but cannot explain why traffic mix changed
- Multinational brands with multiple markets or languages that need to be measured separately
- Organizations with a large content library that want to know which pages are actually cited
Not a fit
- Very little site content → nothing citable; add content first
- Expecting a visible shift in a few weeks → change is measured in months
- Need content produced in languages other than Traditional Chinese → quoted separately
FAQ
Are GEO and SEO the same thing?
No. SEO puts you in search results. GEO puts you in AI-generated answers. The technical work overlaps (structured data, readable content), but the measurement object is completely different. SEO has not gone away. It is just no longer enough.
Which AI platforms are covered?
Google AI Overview is the primary measurement platform today, with a staged expansion to generative platforms such as ChatGPT, Claude, Perplexity, and Grok. Actual coverage for each cycle is confirmed when the engagement starts.
How do we know it is working?
Rerun the same prompt set in later periods and compare Mention, Citation, and Share of Voice. That is why we insist on a baseline first—without a baseline, no result can be verified.
How long until we see change?
About 1 to 3 months, depending on site size and how often you update. Speed tracks content volume and update cadence—sites that publish often are more likely to be re-indexed and cited.
Does the plan include page edits?
No. Plans deliver reports and direction. Actually changing pages is a separate work item (Content Refinement and Content Creation). Scope follows the plan findings.
Do you implement, or do we?
By default your team or site vendor implements. We deliver deployable JSON-LD, code-change notes, and a copy-paragraph optimization sheet. If you need us to implement, we bill actual hours worked.
What are consulting hours for? What if we run out?
They are used for report walkthroughs and the matching monthly consulting meetings. After the hours are used up, additional consulting is NT$6,700 / hour.
We have several markets. Can we run them separately?
Lite and Basic cover one domain and one language. For a specific issue in another market or language, add Focus Study: 50 prompts measured for one domain, one language, and one topic.
Can implementation items be purchased separately?
No. They cannot be purchased independently.
How is the monthly prompt count calculated?
It is the number of prompts. Measuring the same prompt on several platforms still counts as one.
First confirm where you stand
In one discovery call we learn your site size, primary markets, and current content structure, then tell you which plan to start with and how long change typically takes.
Book a discovery call