Cohort 01 — Product Data

One-pager. This is the document that approves the cohort.

Date: 2026-04-14 Sponsor: Claire Chun, France Roy Core owners: Kate Hollywood, Mary Corrick (confirmed by Claire Apr 14)

The workflow

Balsam's product data workflow is how new SKUs, attributes, images, and localized content move from merchandising decisions into the systems that sell products — the website, retailer catalogs, localization pipelines, and planning tools.

Today this workflow runs on Excel spreadsheets (PIFs, product information files, with 200–300 columns), tribal knowledge held by a handful of people, manual transformations for each retailer that carries Balsam products, and a sequence of hand-offs that are mostly people filling gaps between systems that don't talk to each other.

Kate Hollywood mapped the workflow during the strategy sprint: 19 steps, 18 bottlenecks. She and Mary Corrick are the people doing this work today. They feel the friction every day.

Who feels it

Producers of product data:

  • Merchandising (e.g., the team around Brooks Bermingham) generates new SKUs but doesn't feel the downstream pain — they send it over and move on.
  • International teams need localized versions and hit delays.
  • Creative ops needs matching assets and struggles when they don't arrive on time.

Consumers:

  • eCommerce needs clean, complete data to list products.
  • Retailer operations (Bob McDonald's group) manually translates Balsam's product data into 4–5 retailer-specific formats per retailer. Target is 10 retailers by 2027.
  • Marketing needs product context to personalize and target.
  • Planning needs SKU attributes to forecast.

Customer:

  • A shopper in a new market sees a partial catalog at launch.
  • A retail partner waits for usable data.
  • A holiday shopper sees inconsistent content across channels.

Why it matters

Three reasons this is the first cohort:

  1. No PLM in 2026. Balsam decided not to move forward with PLM implementation this year. The existing workflow now carries weight it wasn't designed for.
  2. Retailer onboarding scales linearly today. Every new retailer means manual format translation. Ten retailers at that cost is a constraint on indirect sales growth.
  3. Upstream impact. Product data feeds content production, international launches, marketing, planning. Fixing the source helps four downstream workflows more than fixing any one of them.

Why upstream first

Claire named this explicitly: don't solve downstream pain (retailer-specific asks) in a way that just moves inaccurate data faster. Moving bad product data into ten retailer formats faster does not help. Fixing the source does.

This cohort's work is upstream — the PIF, the master data, the producer-to-consumer hand-off. Retailer-specific work is downstream of this and not in scope for Cohort 1.

What a better outcome looks like

By end of the cycle:

  • A clear, shared map of the full product data workflow that people across functions can point at and align on.
  • One or more manual hand-offs removed or automated, visible in Kate and Mary's day.
  • Skills (AI prompts, configured tools) that producers use to generate cleaner data at source.
  • A playbook for how new SKU attribute proposals get evaluated and added.
  • A handoff package for a follow-on engineering project (likely PIF versioning, or PIF-to-retailer translation) with requirements and evidence.

By end of the founding period (October 2026):

  • The retailer onboarding process is not gated on manual data translation.
  • International market launches don't wait weeks for localized data.
  • eCommerce sees clean, complete product data without heroic effort.
  • Kate and Mary's workflow looks different enough that they are the ones teaching the next cohort what changed.

Why now

The April 29–May 1 Monterey Bay onsite is the launch moment. Kate and Mary are in Balsam's strong AI users already. PLM-free 2026 is the trigger. Claire has confirmed Product Data is where she wants to start.

Scope

In scope:

  • Mapping the full product data lifecycle, producers and consumers.
  • Understanding the PIF structure and its constraints.
  • Prototypes that improve producer-side data quality.
  • Prototypes that improve hand-offs between producers and consumers.
  • Skills and playbooks for day-to-day use.

Out of scope:

  • Retailer-specific format translation (downstream).
  • PLM selection or implementation.
  • Broader merchandising strategy.
  • International localization pipeline rewrite (could be a follow-on cohort).

Open questions to resolve by kickoff

  • Who is the right merchandising producer to have in the cohort? (Not Brooks — too senior for 20%. A hands-on merchandiser who actually creates SKU records.) To raise with Claire / merchandising leadership.
  • Clint Borrill has been proposed internally as the hacker/consumer. To raise with Claire first, then ask Clint directly if she agrees.
  • What data access does the cohort need and who grants it? (France's area.)
  • Who's the sponsor from Bob's team if retailer onboarding enters later cohort work?