Bangkok studio · Behavioral Segmentation Analytics

The map is not the cohort.

Switch Routehub trains analysts and marketers to name segments from the paths people actually take — not from the personas a workshop once voted on.

Analytics dashboards on a desk used for cohort and funnel review
Studio still · cohort boards from a recent Atelier week Klong Goom Buengkhum, Bangkok
17taught cohorts since the studio opened
2,140practitioners through the rooms
11Thai enterprise teams on Field Residency
6.8 yrsaverage commercial-analytics tenure of faculty

Rooms now open

Courses that start from logs, not slogans

Each room is built around Behavioral Segmentation Analytics as a craft: taxonomy, boundary rules, and the awkward conversation with stakeholders who still want five neat personas.

Laptop showing charts used in cohort signal work

Flagship · 6 modules

Cohort Signal Architecture

Design event taxonomies, third-axis scores, and boundary rules that still hold after Songkran traffic and flash campaigns.

Course detail
Monitor with dense data visualizations

Studio short · 3 weeks

Recency–Frequency Recalibration

Repair RF models that punish high-consideration categories and inflate “loyal” buckets after one lucky month.

See listing
Notebook and laptop on a wooden desk during analysis work

Field methods

Micro-Segment Field Methods

When to split a cohort, when to refuse the split, and how to test a micro-segment without drowning the CRM.

See listing

From a recent Atelier

“The recency module assumed cleaner event logs than we had. We spent a week on hygiene before the rest of the room made sense — budget that week, or the architecture stays decorative.”
Mali, Chiang Mai · retail analytics lead · Cohort Signal Architecture

Why this studio

Segments fail quietly. We teach the failure modes first.

  • 01

    Hygiene before poetry

    Most “behavioral” dashboards are just recency painted gold. We start with event names, identity joins, and what must never be a conversion.

  • 02

    Boundaries that survive Thailand’s calendar

    Festival spikes, monsoon delivery lags, and 11.11 windows move cohorts. Drift detection is a module, not a footnote.

  • 03

    Language a brand lead can fund

    A segment that cannot be explained in one slide does not ship. Presentation is part of the architecture, not a polish pass.

Team reviewing notes around a table during a segmentation workshop
Workshop wall with notes from a live case discussion

How rooms run

Case first, theory only when it earns the hour

Each taught cohort works a live Thai retail or subscription brief. Faculty sit with the messy extract, not a toy CSV. You leave with a written boundary rule, a drift check, and a stakeholder memo — not a certificate wall.

If you need the full map of how we name routes, the Segment Atlas is the public version of our internal method.

Journal

Notes from the practice, not recaps of textbooks

Hands typing on a laptop while reviewing analytics

Retail · Thailand

How recency–frequency patterns mislead Thai retail teams

A lucky Songkran week can mint a “loyal” segment that evaporates in May. Here is the correction we use in Atelier.

Read the note

Method

Micro-segments versus personas: when to split a cohort

Splitting is not insight. It is a cost. We use three tests before we allow a new name.

Measurement

Reading session-depth as a loyalty proxy

Depth can mean confusion as easily as commitment. The proxy only works with an exit taxonomy.

All journal entries

Next step

Bring a messy extract, not a polished deck.

Pricing is listed without checkout. If a room looks right, write to the studio with the question you cannot answer from your current segments.

Write to the studio