CO2 Sensors in Open-Plan Offices: Ventilation Assumptions vs What Monitors Actually See

Consumer carbon-dioxide monitors crept from niche air-quality Twitter into mainstream desk setups after years of headlines about filtration, wildfire smoke, and return-to-office anxiety. In an open plan, a tidy number on a small screen promises objectivity: parts per million, updated every few seconds, sometimes graphed into reassuring waves. Facilities teams and architects, meanwhile, size outdoor air using codes, occupant loads, and diffuser layouts that assume statistical humans in statistical chairs. When those two stories disagree, arguments start—and they usually hinge on assumptions nobody wrote down.

This article is about what desk CO₂ sensors actually measure, where they lie by omission, and how to interpret them next to the ventilation design intent of a modern office. You will not get a single “healthy” threshold to tattoo on your policy PDF; you will get a clearer picture of how tightly coupled human comfort is to controls nobody on the sales tour mentioned.

CO₂ as a proxy, not a toxin scorecard

Indoor CO₂ tracks exhaled human breath against outdoor air dilution. High readings generally mean low per-person ventilation rate relative to occupancy, not that the room is “full of poison.” Codes use CO₂ differential as one way to infer ventilation adequacy in commissioning because it correlates with body odor perception and cognitive performance studies—not because 1,200 ppm is a cliff edge for toxicity. Treating the display like a smoke alarm breeds panic; treating it like a tracer for dilution breeds better questions. Context is the entire game.

That framing matters when leadership asks for “green” numbers. A packed all-hands can spike CO₂ briefly while still being safe; an under-occupied floor with stuck dampers can read “fine” near a leaky window wall while a core zone suffocates. Averages hide sins.

Open plans mix air whether you want them to or not

Thermal plumes, ceiling jet inlets, and internal heat loads create mixing patterns that CFD posters simplify into pastel arrows. Reality includes printer alcoves that run hot, kitchenettes that trip exhaust interlocks, and corners where two VAV boxes fight. A desk sensor captures a local pocket. Ten meters away, the experience differs. “The floor reads 900 ppm” is almost always “this desk’s microclimate trends toward 900 ppm under these weather and occupancy conditions.”

What cheap sensors smooth over

NDIR modules drift with temperature, age, and calibration assumptions. Auto-baseline routines that silently subtract “background” can erase slow leaks of reality if windows stay closed for weeks. Cross-sensitivity to humidity and barometric pressure shows up as wobble on rainy days. Firmware that averages aggressively makes graphs pretty and hides fifteen-minute plateaus that matter for cognitive fatigue studies. None of this makes consumer devices useless; it means two devices side by side disagreeing by a hundred ppm is ordinary, not evidence of sabotage.

Design airflow versus experienced airflow

ASHRAE 62.1-style thinking allocates outdoor air by occupant density and floor area, then trusts delivery paths. Retrofits that densify desks without revisiting OA reset schedules leave controls optimized for a 2015 headcount. Meeting rooms with occupancy sensors may minimum-flow to save energy—sensible—until back-to-back video calls stack bodies faster than the VAV ramp. CO₂ traces expose those mismatches faster than complaint tickets because humans adapt and stop noticing staleness.

Post-pandemic resets many buildings never finished

Early 2020s guidance pushed higher ventilation and better filtration in bursts; some sites implemented temporary overrides, left them, then partially rolled back when energy bills spiked. Others installed MERV-13 media but skipped fan-speed verification, trading pressure drop for paper compliance. A desk sensor cannot see filter rating, but it can show whether increased outdoor air actually reaches the bay where people sit—or only the corridor outside the landlord’s spec suite. Use CO₂ to validate that narrative with timestamps, not vibes.

Triangulate with humidity, VOC proxies, and particulates

CO₂ alone cannot distinguish “stale” from “irritating.” Dry winter air with adequate OA still feels harsh; high VOC off-gassing from new furniture can feel oppressive while CO₂ looks innocent. Many teams pair CO₂ with PM2.5 during wildfire weeks and with relative humidity targets in winter. The combination reduces false certainty: if CO₂ climbs but PM stays low and RH is stable, you are probably looking at occupancy-driven dilution issues, not outdoor smoke intrusion.

Phone-booth pods and “not really open” open plans

Acoustic pods and micro-rooms complicate the open-plan map. A pod with local recirculation and weak tie to central OA can spike CO₂ during two-person interviews while the outer floor reads pristine. Facilities may exclude pods from global sequences if they were procured as furniture, not HVAC scope. If your spike graphs correlate with pod bookings, the fix is pod-specific exhaust or schedule limits—not cranking the entire floor’s OA blindly.

Portable air cleaners: friend, foe, or confounder

Desk-side HEPA units change particle counts without changing CO₂ much. They can also short-circuit mixing: a strong local fan pulls clean supply toward one corner, leaving another starved. Before attributing CO₂ drops to “the building improved,” log whether cleaners ran continuously and where. Operations loves simple stories; physics prefers annotated timelines.

Placement rules of thumb

• Breathing zone height — chest height beats floor or ceiling unless you are diagnosing supply air directly.
• Distance from people — too close to one talker tracks their exhalation plume; too far blends slowly.
• Away from direct supply jets — cold supply can stratify and fool sensors about mixing.
• Log location metadata — if you move the puck, your history changes meaning.
• Note orientation of ceiling diffusers — sitting directly under a cold jet can change both thermal comfort and how quickly breath mixes into the sampled stream.

What to ask facilities instead of waving a screenshot

Bring curiosity, not indictment. Useful questions: What is the OA reset strategy? When were balances last verified? Are any zones on fixed minimums for pressurization? Is economizer free-cooling locked out under certain humidity? Do kitchen exhaust interlocks drop make-up air in a way that affects your bank? CO₂ data becomes persuasive when paired with timestamps correlating to HVAC mode changes—not when brandished as a moral score.

Policy and comfort collide at the thermostat wars

Energy targets push tighter control bands; comfort complaints push overrides. Sensors give occupants language beyond “it feels stuffy.” That can accelerate fixes or accelerate conflict if leadership treats consumer devices as calibrated legal instruments. Align on which BMS trends are authoritative and whether spot measurements trigger work orders or just conversations.

Certifications, leasing, and the paperwork gap

WELL points and LEED credits encourage monitoring and responsiveness, but the path from a desk puck to a landlord work order crosses lease clauses, base-building versus tenant improvement boundaries, and sometimes unmetered common shafts. A tenant’s CO₂ log may prove discomfort without proving liability. Document who owns OA resets, who pays for rebalancing, and whether after-hours HVAC cuts include your zone. Sensors make the discomfort legible; contracts decide who pays to fix it.

Commissioning is a snapshot; operations are a weather forecast

Balancing reports taken on a mild spring day age poorly when winter stack effect reverses pressure relationships or when a neighboring suite vacates and dampers idle oddly. Revisit sequences seasonally. Trend CO₂ against outdoor enthalpy and wind speed if your BMS exposes those points—consumer devices cannot, but facilities can export overlays. The goal is to stop arguing about a single 1,400 ppm screenshot and start looking at conditional distributions: hot calls versus cold mornings, windy facades versus calm ones.

Accommodations, migraines, and the human variance problem

Some people are exquisitely sensitive to stale air; others notice nothing until CO₂ is cartoonishly high. Disability accommodation processes sometimes request environmental logs. Be careful about over-promising precision. Offer repeatable measurement protocol: same device placement, logging interval, and parallel BMS trend if available. HR and facilities should coordinate language so individuals are supported without implying consumer hardware is medical-grade instrumentation.

Economizers, free cooling, and the CO₂ curve you misread as “people”

When outdoor air enthalpy drops, economizers can slam additional OA dampers open. CO₂ falls not because fewer humans arrived, but because dilution jumped. Conversely, enthalpy lockouts can clamp OA during muggy afternoons while occupancy rises—CO₂ climbs even though “the fans are loud.” Overlaying economizer stage on your CSV prevents blaming coworkers for physics your controls already chose. Night-flush sequences and morning warm-ups create similar mirages if you eyeball graphs without HVAC mode annotations.

A pragmatic takeaway

Desk CO₂ monitors in open offices are best used as relative tools: compare Tuesdays to Thursdays, window-open experiments to closed, pre- and post-filter changes. They challenge ventilation assumptions productively when everyone agrees on limits of precision. They become toys when treated as absolute truth painted on a morale slide. The building already had an air story; the sensor just made it legible—local, noisy, and wonderfully specific.

If you take one habit away, make it this: screenshot the spike, then append weather, schedule, and HVAC mode before you send the email. That single discipline converts arguments into investigations—and investigations usually find a damper someone forgot to tell you about.