Writing
Notes from the build log.
Essays and teardowns on data + AI go-to-market engineering.
How we built this site: the same system thinking we sell
A behind-the-scenes look at rebuilding antifragilegtm.ai — content as data, design checks in CI, screenshot-driven reviews, and why a marketing site should be engineered like a GTM system.
engineering · meta · design
Speed-to-lead is a routing problem
Why your inbound leads sit for hours before a rep acts.
routing · inbound · speed-to-lead
Pilot then scale: the enrichment discipline
Test on 50 rows, measure, then commit budget. Every time.
enrichment · clay · cost
Volume problems are usually data problems wearing a costume
Why we rebuilt a cold-call motion by fixing the data layer first, and the general rule hiding inside that project.
philosophy · data · outbound
GTM engineering is what RevOps looks like with a data layer
The emerging discipline that sits between RevOps and software engineering.
gtm-engineering · revops · philosophy
Every closed deal is training data
Why win/loss analysis should be a pipeline, not a workshop.
win-loss · scoring · data
Deliverability is a system, not a setting
Domain warmup, validation, and bounce monitoring as infrastructure.
deliverability · outbound · infrastructure
Laplace smoothing: the scoring hack for early-stage teams
How to build usable scoring models with fewer than 50 closed-won deals.
scoring · statistics · data
Territory assignment is the most underrated routing input
Why orphaned accounts and rep collisions kill pipeline quietly.
routing · territory · revops
AI agents for GTM need guardrails, not blind trust
Why agents writing directly to CRM is how you create duplicates at machine speed.
ai · agents · engineering
Warehouse-first GTM, explained
Why your CRM should be an interface, not a database.
warehouse · architecture · data
Niche signals beat generic intent data
Why the sharpest outbound teams build their own monitors.
signals · niche · outbound
CRM decay is measurable and preventable
How to quantify data rot and build guardrails that stop it.
crm-decay · hygiene · data
Parallel dialing is a data problem
The story of 100 dials becoming 1,500 without changing the dialer.
outbound · dialing · data
Freemail inbound is not an edge case
Why 30% of your best inbound leads are being mis-routed.
routing · inbound · identity-resolution
The practical guide to waterfall enrichment
How to chain providers for coverage without blowing your budget.
enrichment · waterfall · data
Your ICP is lying to you
Why aspirational ICPs produce aspirational pipeline (i.e., none).
icp · scoring · data
Enrichment is not a project, it is a pipeline
Why one-time CSV uploads guarantee CRM decay within six months.
enrichment · crm-decay · data
Signal engines are not Slack channels
Why your signal program died in a shared channel and what to build instead.
signals · adoption · engineering
The merge-map method: dedupe without data loss
How we deduplicated 320k accounts while preserving every activity record.
dedupe · crm · data
Why reps ignore your CRM (and it is not their fault)
CRM adoption fails because the data is wrong, not because reps are lazy.
crm · adoption · data