Stack wiring

Q5 asked for a brief answer. This is the longer version of the brief answer.

The four tools stay

Tool	Job
HubSpot	Holds the account record and the score. Routes. Owns the handoff.
Clay	Enrichment, job changes, and the GitHub identity build.
Customer.io	Product-triggered lifecycle for self-serve. The activation journey.
Lemlist	Outbound on high-intent accounts. The sequence.

Two additions, and why

Segment. The four tools all sit outside the product. Every trigger in the signals reference is a product, docs, or repo event. Something has to carry those into HubSpot. Segment does that, with one event spec.

A warehouse (BigQuery). The score gets computed somewhere. Not in HubSpot, which can't do the joins. Not on the product's ClickHouse cluster, for two reasons: GTM data is small and join-heavy, which is the workload ClickHouse is worst at, and marketing queries don't belong on the production cluster. Segment also has no first-class ClickHouse destination, so BigQuery is the path of least engineering.

That's it. Two lines on the bill, each justified by a gap the current stack can't close.

What moves where

product / docs / repo events
        │
        ▼
     Segment ──────────────► BigQuery (score computed here)
        │                        │
        ▼                        ▼
     HubSpot ◄──────────── score written back
        │
        ├── self-serve  ──► Customer.io
        └── sales-side  ──► Lemlist ──► Slack (queued by signal type)
        ▲
     Clay (enrichment, job changes, GitHub identity)

Self-hosted users

None of this sees them. No product events, so Segment is blind. The only path in is external: GitHub identity through Clay, docs traffic, community. That's a build, not a purchase. Commit email to domain to account, with fallbacks for the noreply case.

I've built this before: SignalX scans developer-tool repos, maps contributors, resolves public emails, and scores them.

What I'd say no to

• A CDP beyond Segment. Segment is the CDP.
• Redis or any real-time serving layer. There's no real-time requirement yet. Adding one now is architecture ahead of the workflow.
• A separate identity-resolution tool like Common Room, unless the goal expands from GitHub to Slack and Discord as one graph. It overlaps Clay too much to justify on the GitHub gap alone.

One thing to check before committing: Lemlist now ships its own Signal agents and Enrichment. Some of what's assigned to Clay above may already be covered. Worth a look before paying twice.

---

Source: signoz_assignment_progress.md, Q5 section, 5 Sep 2026.