Explore the projects
/ Applied Research

Where frameworks get stress-tested

The Lab runs experiments on live multilingual data. What survives becomes a recommendation. What doesn't is documented just as carefully.

Over-the-shoulder view of a practitioner editing multilingual search content on a laptop, natural daylight from a wide window to the left, multiple language strings visible on screen, a notebook with handwritten test notes open beside the keyboard
Over-the-shoulder view of a practitioner editing multilingual search content on a laptop, natural daylight from a wide window to the left, multiple language strings visible on screen, a notebook with handwritten test notes open beside the keyboard
Close-up of hands arranging printed workflow diagrams on a whiteboard surface in a real office, natural overhead light, sticky notes with short text labels visible at the edges, a coffee mug in the far corner
Close-up of hands arranging printed workflow diagrams on a whiteboard surface in a real office, natural overhead light, sticky notes with short text labels visible at the edges, a coffee mug in the far corner
— Current Projects

Two active lines of inquiry

GEO Experiment — 01

How retrieval systems rank multilingual content

We ran identical content structures across six language variants and tracked citation frequency in generative engine outputs. Three structural patterns consistently outperformed; two that conventional SEO doctrine favors did not.

Working notes include the test corpus, the ranking deltas, and the two hypotheses that failed. The methodology is open.

Workflow Prototype — 02

A localization handoff that doesn't break at scale

We prototyped a content handoff sequence for a twelve-locale operation using off-the-shelf tooling. The goal was to cut review cycles without adding headcount. Current iteration reduced coordinator touchpoints by roughly a third.

The prototype log shows every version, the constraints we hit, and what we'd reconfigure given a different stack.

+ How the Lab works

Open notes, not polished case studies

Every Lab project publishes its reasoning as it develops — including the dead ends. If a framework can't survive scrutiny in the notes, it doesn't reach a client engagement.

Read the articles

Finished thinking becomes an article

When a Lab experiment resolves into a stable framework, it moves to the Articles archive — written up fully, with context for teams deciding whether to adopt it.