Startup Failure Rates by Industry in 2025: Which Sectors Fail Most and Why

Sep, 16 2025

You clicked because you want a straight answer: which industry eats startups for breakfast? Short answer: it depends on how you define “startup” and which country you’re in. If we’re talking about new small businesses, the US Information sector (software, media, telecom) and Transport & Warehousing usually see the worst survival. In the UK, Accommodation & Food Service and Transport & Storage are among the toughest. If we mean venture-backed tech startups, consumer hardware, crypto/media cycles, and deep capital-heavy plays tend to wash out the most. I’ll give you the hard numbers, the nuance, and a plan to de-risk your move, so you don’t become another statistic.

TL;DR: the highest failure rates by industry (with data)

Here’s the fast version first. Sources include US Bureau of Labor Statistics (Business Employment Dynamics, 2023-2024), UK Office for National Statistics (Business Demography, 2023-2024), CBInsights (2024 post-mortem analysis), and Cambridge Associates / Horsley Bridge (long-run venture returns).

• In the US, the Information sector often has the highest 5-year failure rate for new businesses (around 60-65% close by year 5), with Transport & Warehousing close behind. Accommodation & Food Service is high, but not always the worst.
• In the UK, Accommodation & Food Service and Transport & Storage show the weakest 5-year survival in most recent cohorts; Retail also struggles. Health and Professional Services tend to survive longer.
• For venture-backed startups, consumer hardware and pure-play media/publishing see the most wipeouts; cleantech had brutal vintages in the 2010s, crypto cycles spike failures; enterprise SaaS and vertical software post better odds when unit economics are sound.
• Across all sectors, the top reasons for failure don’t change much: no market need and running out of cash lead, followed by team gaps, pricing mistakes, and getting outcompeted (CBInsights 2024).
• Use sectors as a risk flag, not a verdict. Execution, capital timing, and local market depth matter more than the label on your Companies House form.

Ranges reflect recent cohorts and rounding; exact numbers move with cycles and methods. If you remember one thing, remember this: “Information” in the US and “Accommodation & Food”/“Transport & Storage” in the UK usually give founders the roughest ride.

What do we mean by “startup”? The definition changes the answer

Ask ten people what a startup is, you’ll get twelve answers. That’s why the “highest failure rate” claim bounces around the internet. Here’s how I split it so you compare like for like.

• New small businesses (sole traders, shops, contractors, small agencies): Measured by business survival. Data comes from BLS (US) and ONS (UK). Restaurants, couriers, and some media outfits live on thin margins and fail more.
• Venture-backed startups (designed for hyper-growth, equity funded): Measured by the chance of reaching Series B, profitability, or a meaningful exit. Data is patchier. Cambridge Associates and Horsley Bridge show most VC funds rely on a few winners; 65-75% of startup investments don’t return capital, and 40%+ can go to zero.
• Deep-tech/regulated plays (biotech, energy, aerospace): Many “failures” are scientific or regulatory outcomes rather than simple cash flow issues. Long timelines, binary risk. The win rate can be low but the winners are huge.

By sector inside the venture lane, failure has clusters:

• Consumer hardware: High BOM costs, channel risk, returns, and support. Failures pile up if gross margins are thin or certification slips. This bucket often has the most flameouts.
• Media/publishing and ad-funded apps: CPMs swing, platform dependency bites, user attention is a moving target. Lots of starts, few durable businesses.
• Crypto/Web3: Boom-bust. Vintage-dependent. High attrition in down cycles, then sudden resurrections when liquidity returns.
• Cleantech (early 2010s vintages): Many wiped out under cheap natural gas and slow procurement. The 2020s wave is stronger, thanks to subsidies and better unit economics, but hardware scaling risk is still real.
• Enterprise SaaS / vertical software: Better survival when there’s a tight ICP, high gross margin, and sub-12-month payback. Still plenty fail, but the base rates are kinder.

So when someone says “restaurants fail the most,” they’re not fully wrong, but they’re not fully right either. In the US data, Information often looks worse on 5-year survival than Accommodation & Food. In the UK, hospitality and transport are dragged by energy, rent, and wages. In both places, software shops that never find repeatable demand disappear quietly.

De-risk your bet: a step-by-step playbook that works across sectors

You can’t control the base rate, but you can control your odds. This is the playbook I wish more founders in Manchester followed. It’s not glamorous. It works.

1. Define your unit: are you a small business or a venture-backed bet? Decide your funding path and growth speed now. A bootstrapped agency doesn’t need venture-scale margins; a VC-backed fintech does. Your capital plan sets your survival metrics.

2. Prove demand with hard signals, not optimism. Use the 40/20 test: if 40% of target customers say they’d be “very disappointed” without your product, and 20% give you money or a firm LOI, you have signal. Collect two months of real preorders or signed pilots before building for six more.

3. Set non-negotiable unit economics. Use simple thresholds by model:

   • SaaS: CAC payback < 12 months (SMB) or < 18 months (mid-market). Net revenue retention ≥ 100%. Gross margin ≥ 75%.
   • Consumer subscription: 3-month retention ≥ 50%. LTV/CAC ≥ 3 within six months.
   • Hardware: Gross margin ≥ 40% at launch, path to 55%. Inventory turns ≥ 6/year. Warranty returns ≤ 5%.
   • Restaurants: Prime cost (food + labor) ≤ 60%. Occupancy cost ≤ 10% of sales. Break-even covers 3 slow months of cash.
   • Marketplaces: Take rate ≥ 10% or clear path to value-added services. Liquidity (successful matches) ≥ 70% within 30 days.

4. Design moats early. Choose two: distribution lock-in (partnerships, embedded channels), data advantage (unique dataset or workflow), switching costs (integrations, stored value), or capex barrier (hardware, supply chain). No moat, no mercy in high-failure sectors.

5. Run a 90-day kill plan. Every quarter, score your runway and traction:

   • Kill Score = (Months of runway at current burn) - (Months to next major proof). If < 6, cut burn or narrow scope.
   • Milestone proof = revenue milestone, retention proof, or regulated approval. Vanity metrics don’t count.

6. Use the “one-channel focus” rule until $1m ARR or £80k monthly revenue. Find one channel with CAC stability and scale it before adding another. Spray-and-pray marketing is how good products die.

7. Model downside like an adult. Baseline, downside, upside. If downside wipes you out in six months, change the plan. Your worst case should be survivable without heroics.

Formulas you’ll actually use:

• LTV = ARPU × Gross Margin × Average Customer Lifespan (months) ÷ Churn (monthly). Keep it simple and conservative.
• CAC Payback (months) = CAC ÷ (ARPU × Gross Margin - COGS variable per customer). If you don’t know COGS, you don’t know the business.
• Runway (months) = Cash in bank ÷ Net burn. Add a 20% buffer. Always.

Common traps that inflate failure rates:

• Building a “nice-to-have” in a crowded category. The market won’t save you.
• Relying on ad-funded models without hedges. Platform changes can erase you overnight.
• Underestimating certification and supply chain timelines in hardware or food. Weeks slip into quarters.
• Ignoring working capital. Fast growth kills firms that have to pay suppliers before customers pay them.
• Hiring ahead of product-market fit. Headcount is concrete. Revenue is fog.

Quick tools: examples, checklists, FAQs, and next steps

Examples (based on real patterns I see around Manchester):

• Consumer hardware fitness gadget. BOM is £62, MSRP £149, wholesale £75. Gross margin at wholesale is ~17% before returns and support. This will fail unless you raise price, redesign for cheaper components, or sell direct. You need ≥ 40% gross margin day one.
• Ghost kitchen in a busy student area. Deliveroo fees bite, food costs are up, part-time staff churn is constant. Fix by pushing combo bundles, renegotiating platform rates, and using off-peak specials to raise utilization from 30% to 45%. If you can’t keep prime costs ≤ 60%, pause.
• B2B vertical SaaS for UK trades. Two pilots signed, both pay after 90 days. Cut the custom work, ship a tight product, and get to 10 logos with monthly payments. Target CAC payback < 12 months with field reps + federated WhatsApp groups.
• Crypto portfolio app during a down market. Don’t bank on ad spend. Ship a compliance-grade tax report feature and partner with accountants. If MAU drops 30% for two months and you can’t monetize, freeze spend and switch to tools with durable demand.

Cheat-sheet: fast checks before you commit

• Market demand: Can I get 10 prepaid orders or 3 contracted pilots in 30 days?
• Margins: Do I hit the minimums (SaaS 75% GM, hardware 40%+, restaurant prime ≤ 60%)?
• Cash: Do I have 12 months runway after a realistic sales ramp, not a dream ramp?
• Distribution: Do I own one repeatable channel that scales, with proof (CPL, CAC)?
• Moat: What gets stronger as I grow-data, network, capital barrier, or workflow lock-in?
• Downside: If sales slip 30%, do I survive without high-interest debt?

Decision rules (simple and brutal):

• If you can’t find 10 real buyers without paid ads, the problem isn’t marketing; it’s the offer.
• If your CAC payback is over 18 months and you’re not enterprise SaaS, your model is too heavy.
• If hardware gross margin is under 35% at pilot, delay launch. Redesign or change channel.
• If your “moat” is your brand and you’re under £1m revenue, you have no moat yet.
• If a regulator can stop you, plan for double the time and double the cash.

Mini-FAQ

• So which single industry has the highest failure rate for startups?
  For US-style small businesses, “Information” often shows the worst 5-year survival. In the UK, “Accommodation & Food” and “Transport & Storage” are often the weakest. Among venture-backed startups, consumer hardware and media/publishing see the most failures across cycles.
• Are restaurants the worst?
  They’re tough, but not always the worst on 5-year data. Media firms and couriers can fail faster. Restaurants fail loudly; software shops close quietly.
• Does the year matter (2025)?
  Yes. Energy prices, rates, and platform rules shift survival. Post-2022 rate hikes hurt capital-heavy and ad-funded models. Hospitality felt energy and wage shocks in the UK. Always check the latest cohort data.
• What causes most failures?
  No market need and running out of cash top the list (CBInsights 2024). Execution beats sector label.
• How do I sanity-check a venture bet?
  Target LTV/CAC ≥ 3 within 12-18 months; CAC payback ≤ 12-18 months; a clear wedge into a large niche; and at least one defendable moat.

Next steps

• Define your model (small business vs venture) and write your 90-day kill plan today.
• Run demand tests: collect 10 prepayments or 3 signed pilots in 30 days. If not, change the offer.
• Build a quick unit economics sheet with your real costs. If you don’t know a number, estimate high.
• Pick one channel and push it to a stable CAC before adding a second.
• Schedule a brutal review in 90 days: keep, narrow, or kill.

Troubleshooting by sector

• Information / media startup with ad revenue only. Add paid features now: power analytics, export, or team seats. Shift to owned audience (email/SMS). If ARPU won’t clear £5/month, pivot to B2B tools.
• Transport / courier service burning cash. Tighten zones, raise minimum order, push scheduled deliveries, and improve route density. If you can’t hit 40% route utilization by week 8, pause scaling.
• Hospitality with rent pressure. Negotiate turnover-based rent, expand off-peak events, and push high-margin add-ons (drinks, desserts). If rent stays above 10% of sales for 3 months, consider relocation or a cloud-only model.
• Hardware startup facing long lead times. Switch to a constrained MVP: one SKU, one channel, one geography. Run batch preorders to validate the next production round. Your enemy is inventory risk, not competitors.
• SaaS with flat trial-to-paid. Cut onboarding time in half, remove non-core features, and move top value into the first 15 minutes. If conversion doesn’t improve, your ICP is off-tighten the niche.

One last bit from my home in Manchester: I see founders worry too much about the “worst” industry and not enough about their proof and margins. Eleanor rolls her eyes when I say it again at dinner, but it’s true: pick a hard problem, prove real demand, and keep cash honest. That’s how you beat the base rate-no matter what sector you’re in.

If you need a crisp mental anchor, here it is: the startup failure rate by industry tells you where the potholes are. Your unit economics and proof of demand decide whether you hit them.