Who Funds This, and On What Terms

Most people write the market section first. For a connected device analytics company that is the wrong order: the funding route you can realistically reach determines how much product you are allowed to build in year one. A debt-funded plan and an equity-funded plan for the same business look nothing alike, and submitting the wrong one is the commonest reason a technically strong founder gets a polite no.

The US debt route: SBA 7(a) against contracted revenue

In fiscal year 2025 the SBA approved $37.3 billion across 78,078 7(a) loans, at an average loan size of $497,789 SBA FY2025 lending summary. That average matters: it tells you the size of ask a 7(a) desk sees every day and processes without friction. A $180,000 request to fund twelve months of data engineering and a penetration test is a normal file. A $2.4 million request to build a platform before a single paying design partner is not.

Two NAICS codes cover this business. 518210 is Computing Infrastructure Providers, Data Processing, Web Hosting and Related Services, which fits a subscription analytics product. 541512 is Computer Systems Design Services, which fits an implementation and managed-service model. Pick the one that matches where the majority of revenue comes from in year one, and say so explicitly in the plan, because the code drives which size standard and which lender appetite you are measured against. The SBA does not publish approval rates per NAICS code, so treat any claim you read about an approval percentage for a specific code as directional rather than factual.

What a 7(a) underwriter tests on a software file is unglamorous: debt-service coverage from contracted recurring revenue rather than pipeline, a personal guarantee, and some collateral position even though your main asset is code. So the forecast needs a signed-contract line separate from a bookings line. If those are the same number in your spreadsheet, you will be asked to redo the file.

The UK route: small debt plus SEIS and EIS equity

British Business Bank Start Up Loans cap at £25,000 per founder at 6% fixed, up to £100,000 across four founders. That is useful runway for a two-person team on a managed platform and nowhere near enough for an in-house ingest pipeline. Which is why almost every UK connected device analytics launch we see pairs it with SEIS and EIS equity from an angel syndicate.

Grant funding here is less dependable than founders assume. Innovate UK Smart Grants offered £25,000 to £500,000 for six to eighteen month projects and £25,000 to £2 million for nineteen to thirty-six month projects, funding up to 70% of eligible costs for a micro or small company, and had backed more than 800 projects worth over £350 million before the round was suspended in January 2025 TBAT Innovation. Build the base case without grant income and put it in an upside column with a named competition and a date, or leave it out.

The equity comparable investors will reach for

When an angel or seed fund prices this category, they anchor on two companies. The first is Memfault, which raised a $24 million Series B led by Stripes with 5G Open Innovation Lab, Partech Partners and Uncork Capital participating, for an embedded device observability platform, and was later acquired by Nordic Semiconductor TechCrunch, 2023. The second is Samsara, which reported $1.25 billion of FY2025 revenue, up 33%, roughly $1.5 billion of ARR growing 32%, and 2,506 customers above $100,000 of ARR, up 36% Samsara Q4 FY2025 results, SEC.

Use both deliberately. Memfault shows that a focused device-side product reaches an institutional round and a strategic exit; Samsara shows the ceiling. Neither justifies a valuation, so keep revenue multiples out of the deck and let your own contribution margin carry the pricing argument.

Market Size, Device Counts & Growth

Three different numbers get quoted for this category, and knowing which one you are using is part of looking credible. The Business Research Company sizes connected device analytics at $44.28 billion in 2025, reaching $103.15 billion by 2029 at a 23.5% CAGR The Business Research Company, 2025. 360iResearch puts the same category at $43.96 billion in 2025 rising to $54.22 billion in 2026 and $125.6 billion by 2030 360iResearch, 2025. The narrower IoT analytics segment is sized at $47.59 billion in 2025 growing to $143.62 billion by 2030 at 24.72% Mordor Intelligence, 2025.

Pick one source, state it, and never blend two. A reader who checks will find the blend immediately, and the cost of being caught doing arithmetic across incompatible definitions is far higher than the benefit of a bigger headline number.

The number that actually sizes your pipeline

Market value in dollars is a weak input for a founder, because you cannot sell to a dollar. Device count is the better driver. IoT Analytics counted 21.1 billion connected IoT devices in 2025, up 14% year on year, and forecasts 39 billion by 2030, a 13.2% CAGR, with the 2025 figure landing about 300 million connections below their own September 2024 forecast because of capex deferrals and softer demand in China IoT Analytics, State of IoT 2025.

That downward revision is the useful part: demand here is cyclical and tied to industrial capital spending, which is precisely the risk a lender will probe. Build the model from devices under management times revenue per device per month, then show what happens if onboarding runs two quarters late. A plan that survives that slip reads as competent; a smooth exponential curve reads as untested.

The UK and European picture

The UK IoT market is sized at $24.61 billion in 2025 growing to $36.78 billion by 2030 at 8.4% MarketsandMarkets, UK IoT market. Note the gap between that 8.4% and the 23.5% global analytics CAGR. They are not contradictory: the UK installed base grows slowly while the analytics layer sold on top of it grows quickly. Say that out loud in a UK plan, because a reviewer who spots two growth rates used interchangeably will discount everything downstream.

What It Costs to Stand This Up

Our planning composite for a connected device analytics launch is $60,000 to $435,000 (£48,000 to £340,000). The spread is wide because two very different businesses sit under the same keyword: two engineers on a managed platform serving one vertical, and a venture-funded team building its own ingest pipeline with enterprise security evidence from day one.

Line by line, and why each number moves

• Cloud ingest and data platform, $4,800–$42,000: managed platform list pricing sets the floor. ThingsBoard Private Cloud starts at $1,499 per month for up to 5,000 devices and runs to $3,999 per month for up to 50,000 ThingsBoard pricing. Your own stack on raw cloud is cheaper per device at volume and far costlier in engineering time before it.
• Engineering build, $22,000–$150,000: BLS puts the median annual wage for data scientists at $120,230 as of May 2025 across 275,600 jobs, with 35% projected growth to 2035 BLS Occupational Outlook Handbook. For computer and information research scientists the median is $140,300, with the 90th percentile above $230,630 BLS, research scientists. Four months of one senior contractor is the realistic unit, not a team.
• Security and compliance, $12,000–$65,000: a penetration test and SOC 2 Type I readiness stop being optional the moment you sell into an enterprise. Omitted from most first plans, they arrive as an unbudgeted month-five invoice driven by the customer's procurement team.
• Legal, $3,000–$22,000: incorporation, IP assignment from any contractor who touched firmware, data processing agreements, and the data-access terms the EU Data Act now requires. Most template legal packs do not cover the last item.
• Pilot hardware and gateways, $3,500–$38,000: you need devices to demonstrate on. Borrowing a design partner's fleet looks cheaper and adds weeks to every demo because you cannot break their hardware.
• Brand, documentation site and collateral, $2,500–$23,000: the documentation site is the sales asset here. Embedded engineers evaluate by reading docs at 11pm, not by booking a demo.
• Working capital, $12,000–$95,000: industrial accounts run six to eight months to close plus one to three months of deployment per site. Six months of fixed costs is the minimum buffer, and it decides whether you reach your first renewal.

The three cost mistakes that get plans sent back

First, budgeting the build and forgetting the runway: a plan that funds the product to completion then runs dry during a nine-month enterprise cycle has been wished, not modelled. Second, hiring senior data science before a paying pilot exists; at a $120,230 median, one premature hire burns more than the entire pilot hardware budget. Third, treating hyperscaler services as free infrastructure. AWS ended support for AWS IoT Analytics on 15 December 2025, after which the console and resources are no longer reachable, and AWS IoT Events shuts down on 20 May 2026 AWS migration guide. Google retired Cloud IoT Core on 16 August 2023, five years after launch The Stack, 2022. Migration cost belongs in your model as a named risk with a number against it.

How the Money Works Per Device

Everything here reduces to two numbers: revenue per device per month and cost to serve per device per month. The gap between them, times devices under management, is your gross profit. Most plans state the first number and never state the second, which is why they fall apart in the second meeting.

What the market pays

Published list pricing gives you defensible anchors. Broad IoT platform licensing commonly sits at $1 to $5 per device per month Monetizely, IoT integration pricing. At the infrastructure end, ThingsBoard charges $0.10 per device per month above 5,000 devices, falling to $0.08 above 50,000 ThingsBoard pricing. AWS IoT Device Management with fleet indexing and job execution comes out at roughly $0.10 per device per month, about $1,000 per month at 10,000 devices AWS IoT Device Management pricing.

That spread, from eight cents to five dollars on the same nominal unit, is the whole commercial question. You are paid for how far up the stack your answer sits. A dashboard showing a device is online is worth cents. A model telling a manufacturer which firmware revision will fail in the field within 60 days, before the warranty claims arrive, is worth dollars.

A worked contract, end to end

Here is the arithmetic laid out the way a seed investor will want to see it. A regional water utility has 60,000 connected meters. You price at $0.42 per device per month plus a $14,000 annual platform fee.

Revenue: 60,000 devices times $0.42 times 12 months equals $302,400, plus the $14,000 platform fee, for $316,400 of annual contract value. Cost of goods: ingest, storage and query compute at $0.09 per device per month equals $64,800, plus 0.35 of a support engineer at a $132,000 loaded cost equals $46,200. Total cost to serve $111,000, giving gross profit of $205,400 and a gross margin of 64.9%.

Now the acquisition side. One account executive and one solutions engineer working a six to nine month cycle, plus travel and a proof-of-value deployment, puts fully loaded customer acquisition cost around $118,000. Divide that by gross profit and you get 0.57 years, or 6.9 months, of gross-margin-adjusted payback. That single figure will be checked before anybody reads your market size slide.

Show the margin curve, not the margin

The reason to build the model this way is that your margin is not a constant. At 60,000 devices your ingest cost is around nine cents. At 500,000 devices under management across a dozen accounts, volume commitments and better data retention tiering bring that toward four cents, which pushes gross margin into the high 70s without any price increase. State that explicitly with the three fleet sizes in a table, because it converts a flat assumption into a demonstrated scale effect. Our overall planning band is 62% to 78% gross margin and 14% to 32% net margin once annual recurring revenue passes roughly $1 million.

Separate three revenue layers in the forecast. The per-device subscription is the compounding base. The fixed annual platform fee covers onboarding, access control and reporting, and protects you on small fleets where per-device revenue will not cover support. Implementation and data-engineering services, at $120 to $220 per hour or a fixed fee, fund the early months but do not compound, so label them as services revenue rather than folding them into the recurring line.

Three Business Models Inside One Category

Connected device analytics is not one business. Market reports segment it by solution, into device management, data analytics, network management and security management, which is useful for sizing and useless for planning. From a founder's seat there are three distinct businesses here, with different buyers, cash cycles and funding routes. A plan that straddles two of them is the fastest way to lose an investor.

Margin bands and cycle lengths in this table are Avvale planning composites; the comparable companies are named for positioning reference only. Note the trade-off in the bottom two rows: the services model is easiest to fund with debt because cash arrives in 30 to 60 days, and hardest to raise equity against because a reviewer applies a services multiple however much software sits underneath. Founders often start in column three to fund themselves and then move to column one, and the plan should say that is the sequence rather than pretend services revenue is recurring.

Who Actually Signs the Contract

In this category the person who loves your product almost never controls the budget. An embedded engineer finds you, tries you and becomes your advocate; the money sits with a VP of Engineering, a Director of Service Operations or a plant manager, and the signature needs a security review. Name all three roles in the plan and show how a champion becomes a signature, because that path is your sales cycle and your sales cycle is your working-capital requirement.

Two notes for the operations section of your plan. The proof-of-value deployment is your real product demo and it costs real money: engineer time, gateway hardware, sometimes a site visit. Budget it as a sales cost with a number against it. And the hyperscaler retirements create a time-boxed buying trigger, because every team that built reporting on AWS IoT Analytics has had to move. A migration offer aimed at those teams converts better than a cold pitch.

Data Rights, Device Security & Rules You Inherit

There is no licence to practise analytics. There is instead a set of obligations attached to the devices and the data you build on, and they have changed enough since 2024 that plans written before 2025 are now materially wrong. Handled properly, this is also the strongest differentiation argument available to a new entrant.

European Union: the EU Data Act, applicable since 12 September 2025

The EU Data Act has applied since 12 September 2025 and it is the most consequential rule in this category. Users of connected products have a right to access the data generated by their use of those products. Manufacturers must design products so that data is accessible to users by default, directly and free of charge. Where direct access is not possible, the data holder must make it available without undue delay, free of charge, in a structured, commonly used and machine-readable format, and on the user's request must transmit it to a nominated third party on the same technical terms Greenberg Traurig, September 2025. Products already on the EU market on that date must still allow access to data they continue to generate, without requiring a full redesign Skadden, June 2025. Enforcement is active across member states with penalties reaching €20 million or 4% of global annual turnover.

Read what that does to a business plan. If your moat is described as exclusive control of device data, the Data Act has just converted that moat into an obligation to hand the data over, free, in a machine-readable format, to the customer and to whoever the customer nominates. Any plan that still says "proprietary data asset" about EU connected-product telemetry will be marked down by any investor who has read the regulation.

The flip side is a commercial opening. Somebody has to build the access interface, the consent and authorisation layer, the audit trail and the third-party transmission path, and device makers largely do not want to. Positioning your product as the Data Act access layer, with analytics on top, turns a compliance cost line in your customer's budget into your revenue line.

United Kingdom: PSTI, in force since 29 April 2024

Part 1 of the Product Security and Telecommunications Infrastructure Act 2022, together with the Security Requirements for Relevant Connectable Products Regulations 2023 (SI 2023/1007), has applied since 29 April 2024, enforced by the Office for Product Safety and Standards on behalf of DSIT. It binds manufacturers, importers and distributors of consumer connectable products placed on the UK market, and imposes three duties: passwords must be unique per product or set by the user, there must be a published route for reporting security issues, and the manufacturer must publish a minimum security-update support period with an end date. OPSS can issue enforcement and recall notices, and penalties reach £10 million or 4% of global turnover, with daily penalties for continuing breaches Fieldfisher, PSTI regime, techUK, PSTI explained.

That published support-period duty is quietly a product requirement for you. A manufacturer who has committed in writing to a support end date needs to know, continuously, which firmware versions are live in the field and which devices have fallen out of support. That is a connected device analytics question, and it now has a statutory deadline attached to it. Put the PSTI support-period report in your feature list and name it in the plan.

Alongside PSTI you will carry the ordinary obligations: Companies House registration or sole-trader registration, ICO registration, UK GDPR and PECR treatment of device identifiers, and a data processing agreement with every customer whose telemetry you handle.

United States: the Cyber Trust Mark and a procurement deadline

The FCC's U.S. Cyber Trust Mark launched as a voluntary cybersecurity labelling programme for wireless consumer IoT products in January 2025, covering devices like smart thermostats and fitness trackers while excluding smartphones, wired devices, medical devices and business-only products FCC, U.S. Cyber Trust Mark. Administration has been unsettled: UL Solutions withdrew as Lead Administrator effective 19 December 2025, and the ioXt Alliance was named Lead Administrator effective 13 April 2026, with the FCC to announce when product applications open Cybersecurity Dive.

Procurement is what turns voluntary into mandatory. Executive Order 14306 directs the Federal Acquisition Regulatory Council to require Cyber Trust Mark labelling on all federally procured connected devices after 4 January 2027 BakerHostetler analysis. Any device maker selling to a federal agency therefore has a dated obligation, and the label carries a QR code pointing to a registry that states the support period and whether updates are automatic. Evidence of secure update delivery is, again, telemetry. On top of this sit state privacy statutes led by California's CPRA, which drive data minimisation and retention design rather than licensing.

Write this part of your plan as a product roadmap with three dated requirements across three jurisdictions, not as a legal appendix. A reviewer who sees 12 September 2025, 29 April 2024 and 4 January 2027 handled as features knows immediately that you understand the category.

The Stack Diligence Will Ask About

Technical diligence on a connected device analytics company is narrow and predictable. An investor's technical adviser will want to know where the data enters, where it lands, what computes on it, what the customer sees, and what happens when a vendor in that chain disappears. Name real components in the plan. Vague architecture reads as an unbuilt product.

• Ingest and connectivity: MQTT brokers such as EMQX or HiveMQ, or managed ingest through AWS IoT Core or Azure IoT Hub, quoted from around $10 per unit per month on the Basic tier at 400,000 messages a day.
• Time-series storage: InfluxDB, TimescaleDB on Postgres, or ClickHouse for high-cardinality device data. This sets your cost-per-device curve more than any other decision, so justify it with a retention and cardinality estimate.
• Stream processing: Apache Kafka for transport, Apache Flink for stateful processing when alerting has to happen in seconds rather than on a batch schedule.
• Device-side telemetry: Memfault, now part of Nordic Semiconductor, for crash and metric capture on constrained devices; Golioth and Particle for fleet management; Zephyr RTOS as the common embedded base.
• Platform in a box: ThingsBoard, Losant, Datacake or PTC ThingWorx if you want to sell outcomes before building infrastructure. Starting here and migrating later is more credible than claiming a bespoke platform you have not built.
• Presentation: Grafana for operational dashboards, Metabase or Apache Superset for business-side reporting, plus an embedded layer if customers white-label what they see.
• Security tooling: SOC 2 evidence collection through Vanta or Drata, secrets management, and a documented vulnerability disclosure route, which PSTI makes a legal requirement.
• Vendor exit planning: assume every managed IoT service in your chain has a retirement date you have not been told. A one-page migration plan per critical dependency lands well in diligence.

One point most founders miss while still choosing components: the cost that decides your margin is retention, not compute. Storing every raw sample from every device forever turns a 75% gross margin into a 50% one over three years. Decide at the architecture stage what gets downsampled, when, and who is contractually entitled to the raw record, because the EU Data Act may oblige you to produce it.

Sample Business Plan Preview

An extract from a connected device analytics plan built on this template. Company, numbers and location are illustrative; the structure is the one we file for clients raising seed equity or applying for SBA credit.

What's Inside the Template

The template is an editable Word document with prompts written for this category rather than generic software-business prompts. Every section carries the specific question a lender or investor asks about telemetry businesses.

• Executive summary with a funding ask framed against devices under management, not account count
• Product definition that forces you to state which of the three business models you are in
• Market sizing worksheet with single-source discipline and space for the device-count driver
• Buyer and champion mapping covering the engineer who adopts and the executive who signs
• Per-device unit economics table with revenue and cost to serve at three fleet sizes
• Cost-to-serve and retention policy section, including downsampling decisions
• Startup cost schedule across the seven lines in this guide, in USD and GBP
• Data rights and compliance roadmap for the EU Data Act, UK PSTI and the FCC Cyber Trust Mark, with dates
• Technology architecture section with a vendor-exit plan per critical dependency
• Go-to-market plan built around proof-of-value deployments and platform-migration triggers
• 5-year financial model: profit and loss, cash flow, balance sheet, break-even and contracted-versus-booked revenue split
• Funding section mapped to SBA 7(a), Start Up Loans, SEIS and EIS, with the NAICS code stated
• Risk register including hyperscaler service retirement, silicon-vendor bundling and capex deferral
• Appendices: pricing calculator, design-partner agreement checklist, security questionnaire prep

If you would rather not write it yourself, our market research and content service builds the research and narrative in three to four days, and our bespoke business plan service delivers the full document with a five-year model in ten to fourteen days. You can also start from the free business plan template or read how our business plan writers work.

Frequently Asked Questions

Related Guides

If your business sits next to this category rather than inside it, these guides go deeper on the adjacent models.

• IoT business plan template for a hardware-plus-service model rather than a pure analytics subscription
• Industrial IoT platform business plan template if you are selling the platform itself into manufacturing
• Big data business plan template if your data sources are not devices
• Free business plan template to start from the generic structure