Values & Principles¶
How we decide when a decision is hard.
1. Safety before marketing¶
When claims and safety pull in opposite directions, safety wins. Conservative compliance posture is a moat, not a handicap — it's what unblocks procurement, which is what unblocks revenue. Better to under-claim and deliver than over-claim and explain.
How to apply: If marketing copy leans on a fact I can't defend in a DPA review, rewrite the copy. If a feature expands scope into a regulatory gray area, price it as custom (with legal review) rather than ship it to all.
2. APAC-native, not APAC-washed¶
We know APAC because we built the signal network here, market by market. Don't position as "global with APAC expansion" — that's the category we're replacing. Every segment, every number, every methodology claim is rooted in APAC-native collection.
How to apply: When describing coverage, lead with markets. When describing methodology, lead with local signal sources. Don't borrow generic Western data-marketing language (e.g., "rich psychographic profiles") if it's not what we actually do.
3. Own the infrastructure that owns us¶
Customer-facing lead flow, identity, DSR portal, and domain routing must be on infrastructure we control. Parent company (NB Media) conveniences are acceptable for internal tooling but not for anything a procurement reviewer or regulator would audit.
How to apply: If a shared service creates cross-org coupling on our customer experience, migrate. See ADR-0002.
4. Document decisions, not just state¶
Current state is readable from code and configs. What's missing when a new person joins is why. ADRs, playbooks, and git log capture the why — if it's not in one of those places, it doesn't exist.
How to apply: Material decision → ADR before execution. Recurring process → playbook. One-off? Commit message that explains the why, not the what.
5. Distribution over volume¶
A segment listed where buyers already shop — today Eyeota (part of Dun & Bradstreet), with AWS Data Exchange and Snowflake Marketplace as expansion targets — beats a segment we shout about from our own channels. Distribution reaches procurement where procurement already is.
How to apply: When allocating effort between "more content for us" vs. "more presence on a marketplace," favor marketplace. See ADR-0001.
6. Transparency buys trust¶
The IAB Transparency Standard exists because buyers don't trust opaque providers. Lifetime days per segment, methodology disclosure, opt-out portal — these are not compliance tax. They are sales enablement.
How to apply: When a buyer asks "how did you build this segment?" — answer specifically, with numbers, without reading from a canned script. If we can't answer specifically, that's a signal we need better documentation internally.
7. Small team, sharp tools¶
We're small. Tools and automation should give each person 3-5x leverage, not lock us into rituals. Prefer markdown in Git over Notion. Prefer LLMs doing the first draft over hiring to fill gaps prematurely.
How to apply: When evaluating a new tool, ask: does this scale with team size, or does it require team size to scale? Prefer the former.
8. Ship the first version small¶
Launches are sequenced. Trust page, Partners page, Methodology page — each ships useful on day one, then grows. Don't hold a page for the "complete" version; ship something defensible, improve continuously.
How to apply: A lean page with 3 load-bearing sections beats a perfect page planned for next quarter. If the ship-date slips twice, cut scope.
9. If you can measure it, measure it¶
Launch retrospectives, DSR metrics, marketplace-sourced pipeline — anything repeatable gets a metric. Not a dashboard for its own sake; a metric so we can tell if the thesis is working.
How to apply: Every playbook has a metrics section. Every ADR has review triggers. No open loops.
10. The AI-loop is a feature, not scaffolding¶
We use LLMs — specifically Claude Code — as a core operational loop. Strategy, decisions, code, docs all flow through LLM-readable markdown and Git. This isn't a phase; it's how we scale without headcount.
How to apply: Knowledge must be LLM-maintainable. Decisions go in ADRs. Context goes in playbooks. State goes in Git. If a human-only workflow emerges, ask whether it should be.
These are not slogans. They are how we settle tie-breaker decisions. When two options look equally good, pick the one that aligns with more of these.