Most portfolios already have “smart” point solutions: a BMS here, sub-meters there, access control elsewhere, and perhaps a few IoT sensors scattered throughout. Yet when energy costs spike, comfort calls pile up, or ESG reporting season hits, the data to answer simple questions—What’s driving that peak? Which sequence broke? Where’s the waste?—is scattered across silos.
This is the core reason many “smart” buildings underperform: no unified data layer. Without it, your investments in sensors, controls, analytics, and even AI never compound. They just coexist.
“When we make these buildings run better, the local community, the tenants, and the owner all benefit.” — Jan Huisman, Schneider Electric, Peak Property Performance Ep. 15
Below is a practical, operator-first playbook to unify your data, unlock automation, and make performance improvements stick.
The Problem Behind the Problems
Smart building failures don’t usually trace back to hardware. They trace back to an incomplete context:
- Siloed systems: Utility bills live in finance; BMS points sit with an integrator; access, occupancy, and lighting data sit in separate platforms.
- Inconsistent tagging: “Supply air temp” might be named five different ways across sites. If a human can’t recognize it instantly, an algorithm won’t either.
- No time alignment: Monthly bills, 15-minute meters, 1-minute BMS histories, and event logs aren’t aligned—so you can’t see cause and effect.
- Unclear ownership: If nobody “owns” the portfolio’s data layer, everyone is stuck with CSV exports and spreadsheets.
AI doesn’t fix any of that. In fact, it magnifies the gaps. As we’ve discussed on the Peak Property Performance podcast, AI requires three key elements to add value at scale: clean data, consistent context, and a clear control path.
The Unified Data Layer: What It Is (and Isn’t)
A unified data layer is not “one more platform.” It’s the connective tissue that aggregates, normalizes, tags, and time-aligns all building signals into a single, queryable stream—while preserving the provenance of each source.
Think of it as the independent foundation under your tech stack:
1. Ingest
- Utility interval + bills (elec, gas, water/steam)
- BMS points (HVAC, plant, AHUs, VAVs), lighting, access control, occupancy/people count, weather, tariffs, events
2. Normalize & Tag
- Adopt an ontology (Project Haystack/Brick)
- Map site-by-site variations to a common dictionary\
3. Time Align
- Put everything on a common clock so you can correlate cause → effect
4. Access & Control
- Open APIs so analytics, M&V, and agents can read/write
- Role-based permissions and cybersecurity guardrails
With this backbone in place, every downstream tool—fault detection, demand response, comfort analytics, autonomous setpoint tuning, ESG reporting—gets smarter and faster.
A Short, Real-World Arc: From Setpoints to Closed-Loop
Before:
- Demand charges rising, no clear “why” behind peaks
- BMS and lighting in separate silos; monthly bills stuck in finance
- Engineers buried in alarms and comfort calls
Unify:
- Pull utility interval + submeters + BMS/IoT into one layer
- Normalize tagging across sites; fix “rogue” points and schedules
- Align time series—so a 3:15 PM spike can be tied to a specific sequence
Automate:
- Agents surface projects: economizer logic, condenser ΔT, morning warm-up/pre-cool, schedule enforcement
- Pre-cool hot days; shed during events; verify savings in live data
- Keep changes that save; roll back those that don’t (closed-loop)
Outcomes:
- 15–20%+ reduction in electric spend in the first season (often higher where demand charges dominate)
- Peak kW shaved; fewer hot/cold calls
- Engineers refocus on high-value work instead of chasing alarms
As Mical Anselmo put it in Ep. 12, AI doesn’t “think” for the building—it observes patterns and enforces better decisions 24/7. But it only works because the owner controls the digital infrastructure and the data.
Start Here: A 90-Day Plan
You don’t need a forklift upgrade. You need momentum and compounding wins.
Days 0–30: Inventory & Intent
- Define outcomes: pick 2–3: (1) cut demand charges; (2) improve comfort; (3) automate M&V; (4) prep for ESG.
- Map sources: bills, interval meters, BMS, lighting, access/occupancy, weather. Identify gaps.
- Choose your ontology: Haystack/Brick (and stick to it).
- Quick wins audit: Are weekend schedules off? Is lighting actually shutting down? Is morning warm-up/pre-cool tuned?
Days 31–60: Build the Backbone
- Stand up the data layer: ingest, normalize, tag, align.
- Fix data hygiene: address bad sensors, stuck points, missing meters.
- Open APIs & permissions: decide who can read/write; document your governance.
Days 61–90: Operate & Prove
- Run agent-assisted initiatives: economizer enable, condenser ΔT, rogue schedules, pre-cool/shed logic.
- Close the loop: verify each change in live data; keep or revert.
- Publish M&V: tie savings to tagged points and timestamps; show comfort KPIs.
Governance: Make It Stick
Technology doesn’t fail—governance does. Put lightweight rules in place:
- Data ownership: the owner controls the data layer. Vendors integrate through open APIs.
- Change control: no “hero moves” in the BMS; all persistent changes flow through a documented path with rollback.
- Naming & tagging: every new project follows your ontology; no exceptions.
- Security: segment building networks, whitelist devices, rotate credentials, audit access.
This isn’t more bureaucracy—it’s how performance survives personnel changes, vendor swaps, and portfolio growth.
Budgeting: Treat It Like an Asset
Stop treating digital infrastructure as sunk OPEX. It’s a cash-flowing asset:
- Direct savings: energy reduction, demand charge management, fewer truck rolls
- Revenue protection: comfort = retention; downtime avoidance
- Valuation lift: NOI gains capitalized at your market cap rate
- Compliance & risk: faster ESG/CSRD reporting; fewer penalties
If a lighting system you already paid for can save $70k/year once it’s actually enabled (a real audit outcome we discussed on the show), imagine what a portfolio-wide data layer can unlock.
The Human Side: Expand the Team, Don’t Replace It
Engineers and property managers don’t become obsolete—they become more effective. Add two roles to the roster:
- Digital Architect: designs the data/controls backbone, ontologies, and integrations
- Digital Engineer (Ops): owns data quality, change control, and day-to-day automations
They don’t replace the people who turn wrenches and build tenant relationships; they amplify them.
The Payoff
A unified data layer turns fragmented “smart” projects into a performance engine. It shortens the path from signal → insight → action → verified savings. It enables real autonomy—not science fiction AI—by providing algorithms with consistent context and a safe control path.
And when the grid gets tight, regulations tighten, or rates jump (they will), the portfolios that already own their digital infrastructure won’t flinch. They’ll adapt.
If you’re ready to move from setpoints and spreadsheets to closed-loop performance, let’s architect the data layer once—and let every project benefit from it forever.
—
Inspired by conversations on Peak Property Performance with guests like Michael Anselmo (Ep. 12) and Jan Huisman of Schneider Electric (Ep. 15). OpticWise helps owners design, implement, and operate the digital infrastructure that makes buildings perform.