Every enterprise platform is a workflow
waiting to be traced.
Field notes, architecture teardowns, and honest comparisons on ServiceNow and the wider enterprise platform stack — written by people who have actually carried the pager.
This week's deep dive
Designing Flow Designer subflows that survive an upgrade
Most subflow failures aren't logic bugs — they're coupling bugs introduced two releases ago. Here's a versioning discipline that keeps your automation portable across Now Platform upgrades, plus the three anti‑patterns we found in eleven customer instances.
IT service management
CMDB health is a data problem, not a ServiceNow problem
A retailer's CMDB accuracy went from 41% to 93% without touching a single discovery pattern. The fix was organizational: ownership, not tooling.
Benchmarking ITSM platforms: ServiceNow vs Jira Service Management vs BMC Helix
A vendor‑neutral scorecard across workflow depth, total cost of ownership, and time‑to‑value for a 4,000‑seat enterprise rollout.
Domain separation: when it helps and when it hurts
A decision tree for MSPs and holding companies deciding between domain separation, scoped applications, and separate instances entirely.
Operations & observability
AIOps alert correlation: what actually reduces noise
We tuned correlation rules across 40,000 monthly events. Topology‑based grouping beat every ML model we threw at the problem — here's why.
Event Management's new topology view, six months on
An honest field report on adoption, the learning curve for ops teams, and where the feature still falls short of a true service map.
Cutting MTTR by 40% with predictive incident routing
Skills‑based assignment plus historical resolution data got tickets to the right engineer on the first hop, not the third.
Connecting the platform
Integration Hub vs custom REST: a decision framework
Spoke reuse, error handling, and licensing cost mapped against six real integration patterns so you stop deciding case‑by‑case.
Event‑driven ServiceNow: pairing MID Server with a message bus
Moving from polling integrations to a Kafka‑backed event model cut integration latency from minutes to seconds.
The API‑first enterprise is still mostly a slide deck
Why most "platform" strategies quietly route through three point‑to‑point integrations nobody wants to own.
Keeping the platform honest
Governance without bureaucracy: a lightweight CAB model
A three‑tier change approval structure that cut average change lead time from 6 days to 30 hours without raising risk exposure.
The real cost of technical debt in Now Platform instances
We audited 30 production instances. Customizations older than three releases correlated with a 2.3x higher upgrade cost — here's the data.
Virtual Agent NLU tuning, six months in
Intent containment climbed from 34% to 71% through a disciplined utterance‑review cadence, not a bigger model.
Enterprise platform snapshot
Get the next dispatch
One email a week. Architecture notes, platform comparisons, and the occasional post‑mortem — no vendor decks, no fluff.