Upwind

1. One-liner

Every cheap dust sensor on a UK site now needs a justified action level. This derives and defends it.

2. Trend signal — why now?

In September 2025 the IAQM published a Position Statement that quietly stranded a large installed base of construction dust sensors. It is worth being precise about what it says, because the whole idea rests on it.

UK construction dust monitoring works to an action level: PM₁₀ of 190 µg/m³ averaged over one hour, above which an alert goes to the site manager and a documented response follows. The Position Statement’s finding is that the cheap sensors widely deployed to detect this cannot reliably see it:

“Work by IAQM members has shown that some LCS systems now commonly used for ambient air dust monitoring at or just outside a construction site under-report PM₁₀ concentrations, particularly at concentrations approaching or above the IAQM recommended site action level for PM10 concentrations of 190 µg/m³. As a result, local pollution events may go undetected and, therefore, investigation of the cause and implementation of dust management actions may not be triggered as intended.”

The certification that everyone relied on no longer covers it either:

“MCERTS certification alone is not sufficient to demonstrate that LCS systems are suitable for this application.” (Recommendation 3)

Behind that sits MCERTS Product Certification Bulletin PC4, which withdrew the previously certified ranges for indicative sensors; per IAQM’s account the typical indicative measurement range now tops out around 150 µg/m³ — below the 190 µg/m³ level the monitoring exists to catch.

Then comes the recommendation that creates a product. It addresses sensors already deployed:

“(8) Where such LCS systems are currently in use or its use has been agreed by stakeholders, if the LCS system is not replaced, lower site-specific action levels should be considered. Use of existing background data and how the site-specific action level has been derived and agreed should be documented in any reporting. The site-specific action level(s) is dependent on the equipment and ambient concentration range but should be well within the measurement range against which the equipment has been validated… Comparing upwind and downwind monitoring location data can help determine whether the observed increases are attributable to site activities or to regional or background sources.”

And “should” here is not advisory. The document defines it:

“Throughout this document ‘should’ is intended to indicate that compliance with a statement is NOT optional unless there are reasonable and justified circumstances to deviate, and this must be documented in any reporting.”

So as of September 2025, anyone running a low-cost PM₁₀ sensor on a UK construction site has a new, non-optional obligation: either replace the hardware, or derive a lowered site-specific action level, justify it against the equipment’s validated range and local background data, agree it with stakeholders, and document the derivation in every report. Recommendation 5 adds that they must confirm which sensor, configuration and correction algorithms are in use — noting pointedly that “algorithms can intentionally remove peaks as part of cleaning the ‘noise’ in the data.”

The enabling change on the other side: a paired upwind/downwind deployment used to mean two instruments and two telemetry contracts. An ESP32-S3 with an SX1262 LoRa transceiver is now about $10 in parts and runs for months on a battery with no cellular (Seeed Studio, 2026-05-26), which is what makes a second node — the one that measures what blew in from elsewhere — economically trivial.

Provenance:

• Signal 1 (demand): IAQM Position Statement, September 2025 — MCERTS certification “alone is not sufficient”, deployed LCS under-report at the 190 µg/m³ action level, and where systems stay in use a lowered site-specific action level must be derived, agreed and documented, with “should” defined as “NOT optional” — https://turnkey-instruments.com/wp-content/uploads/2025/09/IAQM-PS-Construction-Monitoring-FINAL-2025.pdf — 2025-09
• Signal 2 (feasibility): IAQM’s own named method is comparing upwind and downwind data to separate site contribution from background; paired nodes are now cheap because ESP32-S3 + SX1262 LoRa is ~$10 BOM with months of battery and no cellular — https://www.seeedstudio.com/blog/2026/05/26/esp32-lora-guide/ — 2026-05-26
• Signal 3 (economic): An established, densely competitive UK consultancy market already charges £6,000–9,000+VAT per year for a single-location real-time PM10 scheme, and this obligation adds recurring evidentiary work to every deployment rather than removing any — https://alkaliconsultants.com/air-quality-services/construction-dust-monitoring-pm10 — accessed 2026-09-17 Category: Regulatory arbitrage

Explicitly not signals. Two tempting hooks were checked and discarded. The EU Ambient Air Quality Directive 2024/2881 “December 2026 deadline” is marketed by sensor vendors as a construction driver; it binds Member States to run ambient networks, imposes nothing on contractors, and post-Brexit does not apply in Great Britain at all. And Section 61 of the Control of Pollution Act 1974, routinely cited in dust-monitoring marketing, contains the word “dust” zero times — it is a noise consent regime.

3. The opportunity

The obvious version of this idea — a cheap sensor undercutting £700/month consultancy packages — is dead, and the Position Statement is what kills it. A £10 optical sensor cannot see 190 µg/m³ reliably, so selling one as a compliance instrument would be selling a device that fails silently at exactly the moment it matters. Worse, the small sites that would want a cheap option mostly have no duty at all: the GLA’s SPG says plainly that for smaller sites monitoring “can be simply visual”.

What the Position Statement creates instead is paperwork attached to hardware that already exists. There is an installed base of low-cost sensors on UK sites, deployed under agreements made before September 2025, that are now non-compliant unless someone derives and documents a lowered action level for each one. That derivation is genuinely non-trivial: it depends on the specific sensor model, its configuration, its correction algorithm, the validated measurement range, and local background concentrations — and it has to be agreed with the LPA and restated in every monitoring report.

Today that is consultant work, done by hand, per site, repeatedly. It is also work that the consultancies are not especially keen to advertise, because the honest version of the conversation is “the kit we specified for you last year now needs a justification file, or replacing.”

The product is the derivation engine and the evidence file: take the sensor you already have, pair it with an upwind node, derive a defensible site-specific action level from its actual validated range and the site’s real background, and produce the document that gets it agreed and keeps it agreed. The hardware is a means to the evidence, not the thing being sold.

The second node matters more than it looks. Recommendation 8 names upwind/downwind comparison as the method for attributing increases to the site rather than to background, and established UK practice is a transect across the site aligned to the prevailing south-westerly. Every consultant knows this; almost nobody does it on a mid-sized site, because historically it meant doubling the instrument spend.

4. Target market

• Primary customer, and this is a deliberate change from the obvious framing: the air quality consultant or environmental manager who already has low-cost sensors deployed across multiple UK sites — small-to-mid consultancies (3–30 staff) running monitoring contracts, and in-house environmental teams at mid-sized contractors. They carry the professional obligation, they read IAQM Position Statements, and they bill the work.
• Why they buy: IAQM defines “should” as not optional, and these are IAQM members whose professional practice is judged against it. Every deployed LCS is now an exposure until its action level is derived, documented and agreed. Doing that by hand across fifteen sites is a week of senior time that is hard to bill in full.
• Secondary customer: the contractor with an in-house sensor and a planning condition, sold through the consultant.
• Rough TAM reasoning: honestly soft, and the sprint exists to fix it. The population is “UK sites with a deployed low-cost PM₁₀ sensor” — not published anywhere. Directionally, the UK consultancy layer is dense (Alkali, Greenavon, RED, Air Dust Odour, Anderson Acoustics, Campbell Associates and many regional firms), and each runs tens of deployments. At £250/month per consultancy seat plus £40/month per monitored site, 60 consultancies averaging 12 sites is roughly £525K ARR. Reaching $1M needs either deeper site counts or the contractor tier.
• Why now for them: the Position Statement is dated September 2025 and PC4 July 2025. The obligation is live, the installed base is large, and nobody has shipped a tool.

5. Product sketch (MVP)

• Sensor-model knowledge base. Pick your deployed unit; the product holds its validated measurement range, inlet type (heated or not), and known correction-algorithm behaviour, because Recommendation 5 requires the operator to confirm exactly this.
• Action-level derivation. Computes a defensible lowered site-specific action level from the sensor’s validated range and the site’s measured background — with the working shown, because the derivation itself must be documented.
• Upwind node, paired. A cheap battery-and-LoRa second unit placed upwind, so every downwind exceedance can be differenced against background and attributed to site activity or not, using IAQM’s own named method.
• The justification file. One dated document per site: sensor and configuration, validated range, background data used, derivation method, the agreed level, and who agreed it — in the form that goes to the LPA and then into every monitoring report.
• Re-justification alerts. When the sensor’s firmware or correction algorithm changes, when background shifts seasonally, or when guidance moves again, the affected sites are flagged rather than silently drifting out of compliance.
• Exceedance closeout with attribution. Reading, upwind comparison, wind context, mitigation applied, sign-off — appended to the same file.
• Fleet view for the consultant. Which of my sites have a current, agreed, documented action level, and which are exposed.

6. AI angle — what’s load-bearing

Weaker here than in most ideas in this catalogue, and it should be said plainly rather than dressed up.

The derivation is statistics and lookup, not language modelling: validated range, background distribution, an appropriate lowered threshold. The genuinely AI-shaped work is twofold. First, the sensor knowledge base — extracting validated ranges, inlet types and algorithm behaviour from manufacturer documentation, MCERTS certificates and independent performance literature that exists as scattered PDFs, and keeping it current as certifications and bulletins change. Second, the justification document — turning a derivation into the written argument a planning officer and an IAQM-member consultant will both accept, against varied condition wording.

If a founder cannot get comfortable with that being the AI content, the honest reading is that this is a domain-heavy data-and-documents product with a hardware accessory, and it should be built and priced as one. That is not a criticism of the opportunity; it is a criticism of forcing an AI narrative onto it.

7. Localization angle

UK-only, structurally. IAQM Position Statements, MCERTS PC4, PAS 4023:2023, the 190 µg/m³ action level and the planning-condition discharge route are all UK constructs with no direct overseas equivalent. The Position Statement itself notes its logic extends to mineral extraction and waste sites — that is the adjacent expansion, not a new country.

The narrowness cuts both ways: no US or EU entrant will build this, and the product does not port without redoing the regulatory layer wholesale.

8. Business model — path to $1M–$5M ARR

• Pricing: £250/month per consultancy for the platform and knowledge base, plus £40/month per monitored site. Upwind hardware node at £180 one-off or bundled above three sites. £400 one-off “retrofit justification” for a single site’s existing deployment — the wedge, sized against a day of senior consultant time.
• ACV: ~£8,800/year for a consultancy running twelve sites.
• Rough math to $1M ARR: ~90 consultancies at twelve sites each, or fewer consultancies with deeper deployments. This is a small, identifiable, reachable population — a strength for sales and a ceiling on scale.
• Rough math to $5M ARR: requires the contractor tier and the adjacent regimes (mineral extraction, waste, industrial sites) that the Position Statement explicitly covers, plus likely the noise and vibration evidence the same customers produce for the same sites.
• Expansion path: retrofit justification → per-site subscription → upwind hardware → fleet compliance view → the full monitoring evidence file across dust, noise and vibration.

9. Go-to-market wedge — first 100 customers

1. The Position Statement is the cold-email. It is dated, quotable, from the customer’s own professional body, and defines “should” as not optional. “Your deployed sensors now need a documented site-specific action level — here is one of your sites, derived, in a file you can send to the LPA” is a pitch that requires no education, only attention. Lead with a real derivation for a real site, not a demo.
2. IAQM membership is a published, bounded list. The consultants who must comply are the ones who belong to the body that issued the statement. This is a few hundred firms, not a mass market — an entirely tractable direct-sales motion, and the reason the customer is the consultant rather than the contractor.
3. Planning portals identify live deployments. Dust Management Plans and CEMPs are published planning documents naming the monitoring scheme and often the instrument. That finds sites with sensors already installed under pre-September-2025 agreements — precisely the non-compliant installed base.
4. Go to the EHOs for the format. Environmental health officers receive the reports. Getting three boroughs to confirm what a derivation file must contain converts the product from plausible to specified, and their answer becomes the sales asset.
5. Partner with the sensor vendors who now have a problem. Recommendation 6 puts the burden on manufacturers to “collate and publish independent evidence” of performance at elevated PM₁₀. Vendors whose kit under-reports need their customers to have a compliant path that is not “rip it out”. A justification layer is the thing they can point at.

10. Build complexity — justification

Medium. The software is a derivation engine, a document generator and a curated knowledge base — no heavy infrastructure. The upwind node is deliberately simple: it measures background, not the action level, so it is not being asked to do the thing cheap sensors cannot do, which sidesteps the trap that killed the obvious version of this idea. It still needs an outdoor enclosure and months of battery life, which is real but bounded work.

The genuine difficulty is domain: assembling accurate validated ranges and algorithm behaviour per sensor model from fragmented sources, and getting the derivation methodology right enough that a chartered practitioner will put their name on its output. That needs an air quality professional involved from day one, not as a reviewer at the end.

Two people, 3–4 months to a credible v1 for a handful of common sensor models, then widen coverage.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Software and documentation support, plus an ancillary background sensor. No certification gate: the upwind node measures background and is not sold as the compliance instrument.
Ethical — no harm / dark patterns	⚠️ Pass with a condition	A tool that derives lower action levels could be misread as helping sites dodge triggers. It must do the opposite — the entire point of Recommendation 8 is that a lower threshold is what makes a weak sensor honest, because the alternative is events going undetected. If the product ever optimises for fewer alerts rather than for detection, it has become the harm it exists to prevent.
Market exists (evidence above)	✅	Dated professional-body obligation, established consultancy market, identifiable installed base.
1–5 person team can build this	✅	Two people plus a domain expert, 3–4 months.
Launchable with <$50K / ₹40L	✅	No MCERTS certification needed for a background node; main cost is domain expertise.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	15/20	A live, dated, non-optional professional obligation with an installed base already out of compliance, and real detection failure behind it. Docked because it is a paperwork exposure rather than a bleeding wound — nobody is being fined weekly, and it can be deferred until a report is challenged.
Demand evidence	15	11/15	The Position Statement is unusually strong primary evidence, verified in the source document. But demand is inferred from an obligation, not observed: I found no consultant or contractor saying “this is costing me”. That gap is the single biggest weakness and the reason this is VALIDATE.
Build feasibility	15	10/15	Software is straightforward; the background node is simple by design. Docked for the sensor knowledge base, which must be accurate across models to be trusted, and for the outdoor hardware.
Distribution clarity	15	11/15	IAQM membership is a bounded, named list and planning portals identify live deployments — genuinely concrete. Docked because it is a small population that will move at professional-services speed.
Revenue mechanics	15	10/15	£250 + £40/site against a day of senior consultant time is defensible. Docked because the customer population is small enough that $1M needs deep penetration of a niche, and the $5M path depends on adjacent regimes that are not yet validated.
Time to first revenue	10	6/10	The £400 retrofit justification can sell before the hardware exists, which helps. But the derivation must be credible to a chartered practitioner first, so this is months not weeks.
Defensibility	10	6/10	The sensor knowledge base and the LPA-accepted derivation formats compound, and regulatory depth deters casual entrants. Docked because a determined consultancy could build this internally, and because a future IAQM clarification or a new generation of heated-inlet sensors at low cost could dissolve the problem entirely.
Total	100	69/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required — with the domain half non-negotiable. This product’s output is a professional justification that an IAQM member signs. Without a credible air quality practitioner in the founding team, nobody will use it and they would be right not to.

Key assumptions to validate (3–5)

1. Assumption: Consultants and contractors with deployed LCS are actually acting on the Position Statement rather than ignoring it. How to test: Interview 20 IAQM-member consultants. Ask what they did about their deployed sensors after September 2025. If the answer is mostly “nothing yet”, that is informative in both directions — it may mean no urgency, or it may mean a backlog waiting for a tool.
2. Assumption: An LPA will accept a derived, documented lowered action level from a third-party tool. How to test: Take a worked derivation to environmental health at three boroughs and ask what would make them reject it.
3. Assumption: The installed base is large enough to matter. How to test: Sample 100 recent Dust Management Plans from planning portals and count how many specify a low-cost sensor versus a reference-grade or heated-inlet instrument. This also produces the TAM number the proposal currently lacks.
4. Assumption: Consultants will pay rather than build a spreadsheet. How to test: Sell the £400 retrofit justification to five firms before building the platform.

Risk flags

1. Demand is inferred, not observed. No customer voice was found. An obligation on paper is not the same as a queue of buyers, and professional bodies issue guidance that practitioners quietly absorb all the time. This is the main risk.
2. The problem may be engineered away. The Position Statement notes that OPCs with heated inlets can measure above the action level. If heated-inlet units become cheap, the compliant path becomes “buy better hardware” and the justification layer shrinks to a transitional market. Watch sensor pricing closely.
3. Guidance drift. This entire opportunity was created by one Position Statement and one certification bulletin within three months of each other. The same machinery can revise it. Track IAQM and MCERTS bulletins as a standing task.
4. Small, slow market. A few hundred consultancies buying professional tooling is a real business but a bounded one, and sales cycles will be measured in months.
5. Single-source dependency on the key document. The Position Statement was read in full and quoted from the PDF, but via a vendor-hosted mirror rather than an iaqm.co.uk canonical URL, and MCERTS PC4 itself could not be retrieved (csagroup.org returns 403). Before building, obtain both from canonical sources and confirm PC4’s actual wording on withdrawn ranges — the 150 µg/m³ figure is IAQM’s characterisation, not read first-hand.

14. Structured verdict

Score:                  69/100
Verdict:                VALIDATE
Confidence:             Medium
Best-fit builder:       Air quality practitioner (IAQM member or equivalent) paired
                        with a technical founder comfortable with light embedded work;
                        the domain credibility is the scarce half
Time to revenue:        2–3 months for retrofit justifications sold manually;
                        4–6 months for the platform and paired node
Capital to launch:      £20–35K
Top 3 assumptions to validate first:
  1. 20 IAQM-member consultants say what they actually did about deployed LCS after
     September 2025 — urgency, backlog, or indifference
  2. Three boroughs' EHOs say what a derivation file must contain to be accepted
  3. 100 sampled Dust Management Plans show how much of the installed base is
     low-cost sensors — this is the missing TAM number
Kill criteria:
  - Abandon if fewer than 5 of 20 consultants have taken or planned any action on the
    Position Statement — an obligation nobody acts on is not a market
  - Abandon if 2 of 3 EHOs say they would not accept a third-party-derived action level
  - Abandon if the sampled DMPs show low-cost sensors are a small minority of deployments
  - Abandon if heated-inlet OPCs reach low-cost price points, which converts the
    compliant path from justification to replacement

15. Next step — 1-week validation sprint

• Day 1: Retrieve the Position Statement and MCERTS PC4 from canonical sources and confirm the withdrawn-range wording first-hand. If PC4 does not say what IAQM says it says, stop here — the entire thesis rests on it.
• Day 2: Sample 100 Dust Management Plans from planning portals; count deployments by instrument class. That number is the TAM.
• Day 3–4: Interview 20 IAQM-member consultants. One question that matters: what did you do about your deployed low-cost sensors after September 2025, and who paid for it?
• Day 5: Take one fully worked derivation — real site, real sensor model, real background data — to three boroughs’ environmental health teams and ask what they would reject.

Falsifiable outcome: proceed only if (a) PC4 confirms the withdrawn ranges, (b) at least 5 of 20 consultants have acted or plan to act, and (c) at least 2 of 3 EHOs accept the derivation format in principle. Fail (a) and the idea is void. Fail (b) and it is a real obligation that nobody is paying to discharge — which is a consultancy afterthought, not a product.