Citely
中文 Start free
Case studies

Different industries, one method:
found, cited, chosen.

Below are clients GKM Research has worked with, across different industries. To protect them, anyone who has not authorised publication appears under an industry code name; where a result has not finished being measured, we leave it blank rather than fill in a flattering number.

Attractive case studies are common. Checkable ones are rare.
We would rather leave a number blank than print growth we cannot stand behind.

〔TBC〕 means there is not yet a checkable number — not that a result is being hidden.

Four industries, four starting points

Pick the one closest to your situation

Each one states the challenge, what we did, and what was measured.

CASE 01 · B2B precision manufacturer

Turning a spec sheet into answers an AI will cite

The challenge

Nearly all site traffic came from brand terms — only people who already knew the company could find it. Buyers searching "who can supply this specification" or "how do I choose this material" before purchasing were being intercepted almost entirely by a handful of competitors. Every unfamiliar buyer was being handed to the other side.

What we did
  • Diagnosed the brand versus non-brand traffic split and confirmed the ceiling was a non-brand vacuum
  • Worked back from procurement decisions to the educational keywords behind them
  • Turned the facts on the spec sheet — certifications, specifications, lead times — into content that can be cited directly
  • Strengthened the technical foundations and structured data in parallel
Non-brand organic traffic
〔TBC〕
Target terms appearing in AI citations
〔TBC〕
Technical health improvement
〔TBC〕
CASE 02 · An aesthetic clinic in northern Taiwan

A highly regulated field still has to be chosen by an AI

The challenge

Prospective patients always ask an AI "who is good for this treatment" and "what is this clinic like" before deciding — and this clinic barely appeared in the answers. The expertise and the cases existed; they had simply never been organised into trustworthy content an AI would cite. It is also a heavily regulated field, so content cannot overstate outcomes.

What we did
  • Built a hard compliance gate so every piece meets medical advertising rules
  • Turned clinical expertise, aftercare guidance and real cases into answers an AI can put forward
  • Produced variants for what different AI engines favour, with a citation forecast before publishing
  • Kept measuring share of mentions across the major AI answers and calibrated month by month
AI citations on decision-stage keywords
〔TBC〕
Non-brand organic traffic growth
〔TBC〕
Where consultation bookings come from
〔TBC〕
CASE 03 · A cross-border e-commerce brand

Several language markets — where should the effort go

The challenge

Shoppers ask an AI "how do I choose in this category, which brand is better" before buying, and this brand was not in the answer. They wanted to enter several language markets at once but had no read on how search intent and competitive intensity differed between them.

What we did
  • Ran search intent analysis per language market to find the high-intent, low-competition junctions
  • Turned product advantages into answers an AI recommending products will cite
  • Ordered the markets by commercial value so effort was not spread thin
  • Built a data flywheel so citation data from each market strengthened the next
Non-brand traffic in target markets
〔TBC〕
Cross-language AI citation coverage
〔TBC〕
Precision keyword positions held
〔TBC〕
CASE 04 · A B2B niche service provider (ESG / professional services)

Demand is growing, but competitors own the answers

The challenge

Demand was growing quickly, yet when prospective clients searched "what is this service", "how is it done" and "who does it", the answers were owned almost entirely by a few competitors. Brand terms made up an unusually high share of traffic — meaning the company could only catch clients who already knew it.

What we did
  • Established that demand was being intercepted exclusively by competitors, and set educational non-brand keywords as the main growth source
  • Estimated commercial value conservatively from the transferable share of competitors' non-brand traffic, and ordered the work by it
  • Deployed teaching-led knowledge clusters aimed at the question stage of the search
  • Ran AI citation work and technical foundations together, so the content stays an asset long term
Traffic on educational non-brand terms
〔TBC〕
Target terms appearing in AI citations
〔TBC〕
Change in where enquiries come from
〔TBC〕
The shared delivery cycle

Whatever the industry, the same cycle runs

01

A conversation first

Look at where the opportunity is before deciding whether to work together.

02

Intent research → strategy design

Find the high-value, low-competition positions worth holding.

03

The system executes

Content and technical work in parallel, with a citation forecast and a review before anything is published.

04

Continuous tuning and results review

Measure real citations, reconcile them against the forecast, and run an after-action review each cycle.

How we present case studies

Four principles we hold to

  • Unauthorised clients are always anonymous or coded— protecting the client is the floor. We will not disclose an identity we should not, just to have a name to point at.
  • Results stay blank until they have been measured— the 〔TBC〕 you see means we do not yet have a checkable number, not that a flattering result is being withheld.
  • Revenue treated conservatively; uncertain financial figures not disclosed— anything touching a client's revenue is handled conservatively, and withheld where it needs to be.
  • No invented citation rates, no invented case studies— every number here can be checked, and every case is real.

We think an honest case study with a blank in it earns more trust than a flattering one that cannot be stood behind.

Rather than reading someone else's case study,
look at your own opportunity.

Thirty minutes, and real measurement tells you whether AI cites you when someone asks about your industry.

Book an AI visibility check