Executive summary
The side hustle economy split in two.
The average side hustle in the US earns a record $1,242 a month. The typical one earns $200, and that number is falling. Both are true. One explains the headlines. The other explains why almost nobody is making any money.
The argument
The side hustle economy isn't getting worse. It's getting sharply bifurcated, and AI is accelerating it. What separates the top 6% from the bottom 94% isn't talent or work ethic. It's three things.
- Whether you own your audience.
- Whether you have someone watching your progress.
- Whether the ratio of your tool spend to your community spend is the right way up.
All three are versions of the same underlying problem: most side hustlers don't own the thing their revenue depends on. They rent attention, they rent feedback, they rent software. Almost nobody is doing all three. That's the opportunity. The rest of this report is evidence.
A note on geography
Headline figures and the K-shape chart are US-led, drawing on the four largest published US surveys. The UK deep-dive in §01 uses Finder UK 2026 and IPSE 2024, the two sources that publish demographic breakdowns at comparable depth. We hold dollars and pounds apart on purpose: we never convert one to the other or aggregate across them. When you see £, the sample is British. When you see $, it is American.
of US workers rely on any secondary income (broad definition)
of Brits have a money-earning side hustle
Median US monthly earnings
of Shopify stores do not survive past 90 days
of creators now use AI tools (global)
Creator economy market size by 2027
The big picture
Four claims this report defends.
Every section that follows is evidence for one of these. If you only read four paragraphs of this report, read these.
The market is K-shaped, not normal
The average US side hustle earns $1,242 a month, a record high. The median earns $200, and it is falling. Average up, median down, simultaneously. The 22% middle is being squeezed from both sides. This is not a downturn. It is the new equilibrium.
AI removed the moat, not the work
Pre-AI, technical execution was the moat. Now anyone can ship a product in a weekend. The floor rose. The ceiling did not. 84% of creators use AI. 12% see it in their revenue. AI did not make winning easier. It made losing cheaper, and there is more of it.
Distribution is the new product
75% of side hustlers operate on platforms they do not own. 11% have their own primary channel. The 6% who break through almost always have at least one owned audience (3.2× correlation). Platforms have figured out they do not need their sellers to succeed. They need them to be interchangeable.
The spending is six times backwards
Sub-$1k earners spend $47/month on SaaS, $8/month on community and mentoring. Community has a 3.8× correlation with crossing $1,000/month. Tools have roughly 1×. Not that tools are wrong. The ratio is. Fix the ratio and you fix most of what is broken.
The side hustle economy isn't getting worse. It is getting more sharply bifurcated. What separates the 6% from the 94% is structural, not personal.
About this report
How we put it together.
This report synthesises data from 25+ sources across government statistics, industry surveys, platform reports, and academic research to create the most comprehensive snapshot of the side hustle economy available in early 2026.
Government data
- Office for National Statistics
- Bureau of Labor Statistics
- Federal Reserve well-being
Industry surveys
- Bankrate 2025
- MyPerfectResume 2026
- Finder UK
- IPSE 2024
Platform data
- Etsy Q4 2025
- Shopify 2026
- Upwork 2026
- Amazon FBA 2026
Research & analysis
- Goldman Sachs creator economy
- Inc Magazine 2026
- Side Hustle Nation
- Henley Business School
A note on definitions. Different surveys define "side hustle" differently. Bankrate’s strict definition gives 27% US participation. MyPerfectResume’s broader definition gives 72%. The UK’s 46% comes from Finder’s broad definition. We’ve noted which definition each stat uses so you can judge accordingly. The truth for intentional side businesses is probably 30 to 40% in both countries.
How each source defines participation
| Source | Definition | Participation |
|---|---|---|
| Bankrate 2025 | Strict. Intentional secondary work | 27% US |
| LendingTree 2026 | Moderate. Regular secondary income | 33% US |
| Federal Reserve SHED 2024 | Official. Any gig activity prior month | 20% US |
| MyPerfectResume 2026 | Broad. Any secondary income | 72% US |
| Finder UK 2025 | Broad. Any additional income activity | 46% UK |
The full methodology, with the glossary and the four-track source breakdown, is in §10.
Section 01
Who’s building side hustles in 2026.
US first, then UK. The headline question. How many people are doing this. Gets very different answers in each country, and the rest of this section unpacks why.
United States
rely on any secondary income (MyPerfectResume 2026, broad definition)
- 33% under LendingTree’s moderate definition
- 27% under Bankrate’s strict definition
- 76.4m freelancers (36% of workforce)
- $1.5T in annual freelance earnings
United Kingdom
of Brits have a money-earning side hustle (Finder UK 2026)
- 70% have tried or considered one
- 1.32m workers have a second job (ONS)
- 4.4m self-employed individuals
- 68% of students now work part-time
The UK number depends entirely on the definition
The 46% Finder UK figure is the broadest possible read: "any additional income activity," including selling used clothing on Vinted and one-off eBay listings. Authoritative sources publish much narrower counts.
- ONS Labour Force Survey: ~1.32m UK workers report a paid second job. About 4% of the workforce.
- IPSE Self-Employed Landscape 2024: 2.046m UK freelancers, predominantly primary income.
- FCA Financial Lives 2024: tracks "informal income" with full-population weighting. Landed in the same low-single-digit range as ONS on a paid-second-job basis.
- Finder UK 2026: 46% on the broadest definition.
The spread is the story, not a problem. Paid contracted second jobs (ONS ~4%) and "do any of these earn money sometimes" (Finder 46%) are not the same population. The rest of this section uses Finder for the demographic breakdowns because Finder is the only UK source that publishes them at this granularity; we flag the wider context here so the comparison frame is honest.
A note on what follows
The next three breakdowns. By generation, by gender, by age. Are UK-only. Finder UK 2026 publishes this depth; US sources (Bankrate, LendingTree, Upwork) report participation but not the same demographic slices.
The patterns travel. Gen Z over-indexes on participation in both countries, men out-earn women in both, financial motivation is rising in both. But the numbers below are specifically British. Earnings figures are medians unless we say otherwise. §03 explains why we use median.
The UK picture
Participation by generation
% of each UK generation with a money-earning side hustle. Source: Finder UK 2026.
The generational breakdown
Gen Z (18-27)
66% participate · ~£140 median/mo
Top categories: content creation, freelancing, Vinted resale. Most likely to use AI tools. Highest failure tolerance but often lack business fundamentals.
Millennials (28-43)
62% participate · ~£230 median/mo
Top categories: freelance services, e-commerce, micro-SaaS. The sweet spot cohort. Enough experience to execute, enough energy to ship.
Gen X (44-59)
36% participate · ~£230 median/mo
Top categories: consulting, coaching, B2B services. Lower participation but the same typical earnings as Millennials. Monetising professional networks.
Boomers (60+)
23% participate · ~£180 median/mo
Top categories: consulting, crafts/Etsy, tutoring. Smallest cohort but fastest-growing. Domain expertise commands premium rates.
The gender gap
Finder UK 2026 publishes gender averages but not gender medians. We show the participation gap and the 1.8× earnings ratio. Both directly sourced. Without inventing a median figure. The ratio is unadjusted: it does not control for hours worked or category mix. Read it as a description of how the average pounds split today, not as an apples-to-apples pay gap.
Men
have a money-earning side hustle
Favour: investing, e-commerce, tech services
Women
have a money-earning side hustle
Favour: crafts, content, coaching, tutoring
Earnings ratio
Unadjusted monthly average. Not hours-controlled, not category-controlled. Most of the gap is mix (investing vs crafts), not pay-for-the-same-work.
Income by age group
Finder UK 2026, in pounds. The mean column shows why averages mislead. The median column is what the typical builder actually earns.
| Age | Mean monthly £ | Median monthly £ |
|---|---|---|
| 18-24 | £340 | £95 |
| 25-34 | £620 | £210 |
| 35-44 | £580 | £250 |
| 45-54 | £450 | £280 |
| 55+ | £310 | £180 |
Source: Finder UK 2026. Mean is skewed by high earners. Median reflects the typical experience.
Why people start side hustles
Finder UK 2026 by primary motivation. The breakdown closely matches US survey patterns (Bankrate, LendingTree). Financial motivation has overtaken passion as the #1 driver in both countries.
In 2023, "passion project" was the #2 motivation. By 2026, "financial freedom" has overtaken it. The cost of living crisis permanently shifted the conversation from "follow your passion" to "build financial resilience." 60% of side hustlers now cite financial motivation, up from 48% in 2023.
Takeaways
- Participation is massive in both countries. Definitions matter. 27% to 72% in the US depending on how strict the definition is; 46% in the UK. The true number for intentional side businesses is probably 30 to 40% in both.
- Gen Z over-indexes on participation. Millennials earn the most. UK: Gen Z 66% vs Millennials 62%, but Millennials and Gen X tie at median (~£230). Pattern holds in US survey data.
- The gender gap is 1.8×. And it’s about category, not capability. UK: men 53% / women 39% participation; men earn 1.8× more per month on average. Category mix (investing vs crafts) explains most of the gap.
- Financial motivation has overtaken passion. 60% now cite financial reasons, up from 48% in 2023. Pattern is consistent across US and UK surveys.
- The median tells the real story. Most side hustlers. US or UK. Are barely covering their tool subscriptions. §03 quantifies the gap between average and median.
Section 02
What they’re building.
Categories, platforms, and the shifting landscape of side hustle types in 2026.
Categories: growth and decline. US (YoY)
US category participation is the cleanest data globally because Upwork, Amazon FBA, Etsy, and TikTok Shop all publish category breakdowns. UK data is reported differently. See the next panel for what's actually known about the UK mix.
| Category | YoY | US participants |
|---|---|---|
| Content creation & social | +18% | 12.4m |
| E-commerce & dropshipping | +14% | 9.8m |
| AI-powered services | +42% | 3.2m |
| Freelance professional services | +6% | 15.1m |
| Teaching & coaching | +11% | 5.7m |
| Gig delivery & driving | −3% | 8.9m |
| Handmade & crafts | −7% | 4.1m |
Sources: Upwork freelancing statistics 2026, Amazon FBA seller statistics 2026, TikTok Shop statistics 2026, Etsy marketplace statistics 2025.
The fastest-growing category, AI-powered services, grew 42% YoY from a small base of 3.2m participants. Meanwhile, the largest category (freelance professional services at 15.1m) grew just 6%. The market is splitting into two tiers: established, slow-growth categories where competition is fierce, and emerging categories where early movers are capturing disproportionate opportunity.
How the UK mix differs
UK sources (Finder UK 2026, IPSE 2024) report category share and total freelancer counts. Not US-style absolute participant counts per category. Here's what's actually known.
UK side hustler share by category
Finder UK 2026 (n ≈ 2,000 adults). The figures Finder publishes. Not every category has a clean number.
Finder's top three by popularity (no published %)
- Selling used clothing (Vinted, Depop)
- Part-time second job
- Social media influencing (TikTok)
UK freelancer total by occupation
IPSE Self-Employed Landscape 2024. 2.046m UK freelancers total. Top occupational groups (SOC1 to SOC3).
Two things to notice. First: Finder's top three UK categories. Selling used clothing (Vinted), part-time second jobs, and social media influencing. Bear almost no resemblance to the US top three. UK side hustles skew significantly toward asset-light resale and platform labour. Second: artistic, literary, and media work is the single largest UK freelance occupation (16%, 318k people), supporting the picture that the UK's side hustle economy is disproportionately creative-services led.
The platform landscape
Platforms: floor enabling, ceiling suppressing
Platforms make this market possible at all. Without Amazon, Etsy, TikTok Shop, Upwork, and the rest, the median is closer to $0 than $200. Read the participation numbers as evidence that the floor exists because of platforms, not in spite of them.
The trap is the ceiling, not the floor. 75% of side hustlers operate primarily on third-party platforms where they don't own the customer relationship. Only 11% run their own website or app as their primary channel. Algorithm changes, fee increases, or policy shifts can wipe out months of effort overnight. The 6% who break through almost always use platforms as a bootstrap and then migrate the customer relationship somewhere they own. The failure mode is staying on a platform past escape velocity. Not being on one in the first place.
Three platforms, three different dependencies
The 75% number above flattens three very different dependencies. It is worth pulling them apart, because the failure mode is not the same in each.
Aggregators
TikTok · Amazon · YouTube · Meta
Own the user relationship. Commoditise suppliers. The algorithm decides who gets seen.
How to read it: The actual trap. The side hustler is a substitutable input; the platform captures the customer and most of the margin.
Marketplaces
Etsy · Upwork · Fiverr · eBay
Facilitate matching. Take-rate present but bounded. Buyer-seller relationship exists.
How to read it: Mixed. Seller can build repeat-buyer relationships off-platform; the platform extracts but does not fully intermediate.
SaaS rails
Shopify · Substack · Beehiiv · Gumroad
Take-rate light. Seller owns the customer list, the storefront, the relationship.
How to read it: Arguably fine. Dependency is on a vendor, not on a gatekeeper. The customer is yours; switching costs are operational, not existential.
The rise of AI-native side hustles
+68% growth
AI content agencies
One-person agencies using GPT/Claude to deliver blog posts, social media, and email copy at scale. Typical revenue: $500 to $3k/mo.
+54% growth
AI automation consulting
Building Zapier/Make workflows and custom GPTs for small businesses. Growing fastest in professional services.
+89% growth
AI-generated digital products
Colouring books, planners, print-on-demand designs made with Midjourney/DALL-E. Low barrier, high competition.
+31% growth
AI tutoring & coaching tools
Custom chatbots for niche coaching, exam prep, and language learning. Early but promising unit economics.
Takeaways
- AI-powered services are the fastest-growing category. +42% year on year growth.
- 75% operate on platforms they don’t control. Only 11% run their own website.
- The market is splitting into two tiers. Established vs emerging categories.
- AI-native categories emerged from nothing in 18 months.
Section 03
The money. This is the chart.
If you only look at one number in this report, look at the next two together. The average is at a record high. The median is falling. Both numbers come from credible 2026 surveys. Both are correct.
The K-shape: average up, median down, at the same time
The news headlines you've read this year quote the average. It's the bigger number, and it keeps going up. The average US side hustle earned $473/month in 2022. $1,242/month today. That looks like a side hustle boom. It isn't. The median, the typical builder, went the other way over the same period. The gap is not closing. It is widening every year.
US monthly side hustle earnings, 2026
Average (mean)
The number you'll see in headlines. Record high, up from $473 in 2022. Source: LendingTree 2026.
Median (the typical builder)
The actual middle of the distribution. Down from $250 in 2024. Source: Bankrate 2025.
The average is 6.2× the median. That's not a normal distribution. It's a long tail of a few high earners pulling the headline up. Most builders never see anything close to the average.
The gap is widening every year
This is the chart of the year. Monthly US side hustle earnings, 2022 to 2026. One line is the average. One is the median. They were $223 apart in 2022. They are $1,042 apart now.
Sources: LendingTree side hustle survey for the average (n≈2,049, "regular secondary income"); Bankrate side hustles survey for the median (n=2,400, "intentional secondary work"), with our 2026 estimate. Two different surveys, two different population definitions: read the slopes, not the dollar delta. The definitional differences are spelled out in §10.
This chart names the report. Every other section is evidence for what this graph shows. Who's making the money (§01). What they're building (§02). Why the gap is widening, not closing (§04, §05). What the 6% at the top are doing that the 94% at the bottom are not (§05, §06).
The lineage
The shape this chart describes is not new. David Autor's polarisation work (MIT, 2010s onward) named the same pattern in the US labour market: a hollowing middle, growth at the top and the bottom, technology accelerating the spread. Brookings' two-track research extended it to skills and geography. Frey & Osborne's automation-susceptibility paper (Oxford, 2013) is the same family of argument applied to job tasks. The contribution this report makes is the side-hustle-specific extension: same shape, same direction, same accelerant (now AI instead of robotics and offshoring), measured on the population of people running secondary income streams in 2026. The precedents make the finding more credible, not less. If the same pattern appears in full-time labour, contract labour, skills distributions, and now side-income distributions, the most likely explanation is that we are looking at one structural force.
Income distribution
Earn less than $250/month. Barely covering tool subscriptions.
Earn over $1,000/month. The "operator" tier with real traction.
Two economies, one label
The experimenters
- Earn under $500/month
- 7 hrs/week on average
- 1.2 revenue streams
- Primarily on third-party platforms
- Most quit within 8 months
- Motivation: extra income (68%)
- Rarely track P&L or metrics
The operators
- Earn over $2,500/month
- 18 hrs/week on average
- 2.8 revenue streams
- At least one owned channel
- Operating 2+ years on average
- Motivation: building a business (74%)
- Track revenue, CAC, and retention
The 22% in between, earning $500 to $2,500/month, are the most interesting cohort. They’ve proven market demand but haven’t cracked scale. This middle tier represents the highest-leverage opportunity.
A third group inside the 72%: the sustainers
Not everyone in the experimenter band is trying to climb. A real and under-counted slice of the 72% runs a single-product or two-product business in the $50 to $500/month range, considers it a success, and has no intention of scaling. Gumroad, Substack, and Patreon are full of them. The report treats this group as part of the experimenter cohort because the published surveys don't separate "small and stable" from "small and stuck." But the strategic question is different for each: one wants leverage on their distribution; the other wants permanence for the small thing they've built. When we say "the 6% break through," the implicit comparison is to climbers. Sustainers are a different evaluation.
Earnings by generation
Monthly average earnings by generation. Note the gap vs the medians in §01. Averages favour Millennials sharply because the operator tier skews younger.
Takeaways
- The distribution is bimodal, not normal. 72% experimenters earning under $500/month. 6% operators earning over $2,500/month. Only 22% in between. The curve has two humps, not one.
- The middle 22% is the opportunity, and it is shrinking. They have proven demand but have not cracked scale. AI-driven competition is pushing more of them down than up.
- The gap is widening, year over year. Average up 162% since 2022 (with a strong caveat: LendingTree's "regular secondary income" definition broadened through this period to include new gig categories, so part of the mean climb is definitional drift, not pure income growth). Median down 20% on Bankrate's stable strict definition. The direction holds regardless. Read the slopes, not the headline percentages.
- Half of side hustlers are running at a loss. 50% earn less than $250/month, which barely covers a typical SaaS tool stack ($47 average).
Section 04
AI removed the moat.
This section defends claim 02 from the big picture: AI didn't make winning easier, it made losing cheaper. The K-shape in §03 is what happens when the floor rises but the ceiling does not.
Pre-AI, the moat was technical execution. Building a website, writing copy, designing a product, shipping it. Those skills sorted the 22% middle from the 72% bottom of the distribution. AI commoditised every single one of them. The moat that made the middle possible is gone. And almost nobody has rebuilt one.
of side hustlers use AI tools regularly
report meaningful revenue impact
An independent read that converges
Brookings (Hui, Reshef, Zhou) studied Upwork freelancers in the 6 to 8 months after ChatGPT and DALL-E 2 launched. AI-exposed categories (copyediting, proofreading, text-heavy services, graphic design) saw roughly a 5% drop in monthly earnings and a 2% drop in new contracts. The authors characterise the magnitude as comparable to historical automation shocks like industrial robots, and note the effect was sharpest among the most experienced freelancers. Read it as independent evidence on the same direction this report points. The floor is rising for the platform; the ceiling is compressing for the people on it.
What they're using AI for: low-leverage tasks
This is the most diagnostic chart in the section. The highest-adoption use cases (content, research, images) are the lowest-leverage. The highest-leverage use case (pricing and strategy) is at 8%. People are using AI to do faster what they were already doing badly.
AI didn't make winning easier. It made losing cheaper, and there is more of it. The floor rose. The ceiling did not.
The four moats AI can't fake
"The new moat is what AI cannot fake" is too compressed to be useful. It covers four different kinds of moat, each with different economics, evidence in the data, and business shape. The 6% operator class typically has one of these. The strong operators have two. Nobody we have studied has all four.
01
Data
The side hustler holds the data the model needs.
Proprietary inputs the foundation models do not see at training time. Fine-tuning data from a niche client, transcripts of conversations with a specific audience, internal taxonomies inside a vertical. Cheap to deploy if you have access, impossible if you do not. The freelancer with five years of client emails has a data moat. The freelancer with a ChatGPT subscription does not.
Where it shows up: Shows up in §02 as the +42% growth of AI-powered services. The ones surviving margin compression are the ones with proprietary inputs the buyer does not have.
02
Distribution
The audience you own. Not the platform you publish on.
The 3.2× owned-audience correlation in §05 is this moat by another name. An email list, a paid community, a podcast subscriber base, a Discord you administer. AI does not erode it because AI does not change who the customer trusts. The algorithm does. Owning your distribution removes the algorithm from the equation. The audience is yours; the channel cannot deplatform you.
Where it shows up: Mapped throughout §02 and §05. The 11% who run their own primary channel are the survivable cohort. The other 89% are renting attention.
03
Taste
The judgement to know which AI output is good.
AI produces volume; taste decides which slice ships. This is the moat designers, editors, art directors, and curators have always had, and the one AI amplifies rather than erodes. The agency that runs Claude Opus and ships work that wins clients has the same model the freelancer running on a free tier does. The difference is taste applied per piece. Cannot be subscribed to; can only be developed.
Where it shows up: The inverse of the 72% content / 8% strategy chart above. High-volume low-leverage use is taste-free use. Low-volume high-leverage use is taste-applied use.
04
Trust
A human relationship the customer is paying for as much as the work.
High-trust services (coaching, advisory, named-expert consulting, therapeutic work, gallery-represented art) are the only category where AI produces near-zero competitive pressure because the buyer is not buying output, they are buying the person. AI cannot replicate the relationship because the relationship is the product. Expect this category to widen and become more expensive through 2027 as AI pushes the price of generic output toward zero.
Where it shows up: §07 shift 02 (commercial-use policy) reinforces this: as the model layer commoditises, the trust layer is what does not.
The 6% operator class has at least one of these. The 94% has none of them and is buying more tools. That is what "the floor rose, the ceiling did not" actually looks like at the level of an individual builder's strategy.
Takeaways
- AI is the accelerant on the K-shape, not a separate story. The floor dropped (anyone can start). The ceiling did not (winning still requires what AI cannot do). The result is the bifurcation in §03.
- 84% adopt. 12% see revenue impact. The gap is structural. People are using AI to produce more of what already does not sell. Faster bad output is still bad output.
- The highest-leverage AI use case (pricing, strategy) is at 8%. The lowest-leverage one (content production) is at 72%. People use AI on the wrong tasks.
- The new moat is what AI cannot fake. Taste, audience trust, distribution, accountability. The 6% operator class has at least one of these. That is the story of §05 and §06.
Section 05
What’s working and what isn’t.
The failure rate is high, but the patterns of success are remarkably consistent.
The failure funnel
What happens to 100 side hustlers over 24 months.
Directional, based on common attrition patterns across the cited surveys. Not a single longitudinal cohort study. Read the shape, not the decimals.
25% quit in first 3 months
23% never make a sale
21% give up mid-year
17% plateau
6% quit late
The 5 factors that predict success
Ranked by correlation with earning over $1,000/month.
How to read these multipliers
Directional, internal sidething estimates derived from cross-referencing the public surveys listed in §10 with our applicant-pipeline conversations. They are not yet drawn from a single primary survey under one methodology. Treat them as well-evidenced patterns, not point estimates. A primary n=500 survey is in scope for the 2027 edition; the multipliers will be re-stated against that data and either confirmed, tightened, or revised.
The rank order is also directional. Community and owned-audience sit at the top because they show up most consistently. The causal arrow between them is unsettled: it is plausible (and a Kit/ConvertKit-style reading supports it) that owned audiences come first and the community is downstream of the list. Ranking them 1/2 is editorial; treat them as a pair rather than a podium.
Community & accountability
Side hustlers with structured accountability partners or groups are 3.8× more likely to cross $1,000/month. The single strongest predictor.
Owned audience channel
Having at least one direct channel (email list, own website, owned community) vs pure platform dependency. The practical breakpoint isn’t subscriber count, it’s revenue per subscriber per month: above roughly $0.50/sub/mo a creator is durable. Below, the list is technically owned but the audience still behaves like a borrowed one.
Multiple revenue streams
Operators average 2.8 revenue streams. Diversification protects against platform risk and compounds growth.
Metrics tracking
Those who track revenue, costs, and conversion rates make better decisions. 74% of operators track. Only 12% of experimenters do.
Strategic AI usage
Using AI for strategy and analysis (not just content production) correlates with higher revenue. Only 8% use it this way.
A fair objection: is this just selection?
The most honest counter to the 3.8× community correlation is that people who pay for accountability groups have the disposable income, the conscientiousness, and the budget to show up weekly. That same trait-set is the thing that makes them ship. The community didn't cause the income. The income-shaped person caused both. We take this seriously. Our read is that selection explains some of the gap but not all of it: structured accountability shows up in our applicant interviews as the variable people credit when their cohort moves and theirs doesn't, even at comparable income and discipline. A primary cohort survey is in scope for 2027 to separate selection from effect.
The #1 predictor of success isn’t the idea, the market, or the tools. It’s whether you have someone checking in on your progress.
Section 06
The tools & community gap.
This is not an anti-tools argument. Tools are essential, and we sell some. It is an argument about ratio. Sub-$1k earners spend roughly six dollars on software for every one dollar on the relationships their revenue depends on. Fix the ratio and you fix most of what is broken.
Where the money goes (under $1k/mo earners)
The community spending gap
Side hustlers spend 6× more on tools than on community.
Yet community is the #1 predictor of success (3.8× correlation). The right software, in the wrong proportion, is still a misallocation. This is the single biggest one in the side hustle economy.
Work entirely alone with no accountability
Have never paid for community or mentoring
Success correlation with structured community
Section 07
Three shifts since the last edition.
Not 2022 takes about AI 'commoditising things.' Three dynamics that are new in 2026 specifically, that we expect to define what wins between now and the end of 2027. The ones we'd argue about with another founder, not the consensus.
Token economics: pay-to-think-better is the new moat.
The cost-quality frontier on AI output shifted in 2025-2026 from free to expensive.
In 2023 a $20/month ChatGPT Plus subscription was the ceiling. By mid-2026, Claude Opus, GPT-5 Pro, Gemini Ultra, and the Cursor/Devin agent stack all sit at $100 to $400/month each. The outputs are visibly different at that price. Solo builders on free tiers ship work that looks competent and gets nowhere; solo builders paying enterprise prices ship work that wins clients. The per-task cost gap is now an order of magnitude wide. Operators we talk to are buying $300/month in Claude Opus credits and treating it as their lowest-ROI expense rather than their highest. Expect this to keep diverging through 2027. The new tooling underclass is anyone trying to do this on the free tier.
What this means for builders
The cheapest line item to upgrade in your stack right now is the model. It has more leverage than another SaaS subscription, by a wide margin.
Commercial-use policy is now competitive substrate.
Anthropic and OpenAI's contracts on commercial use shape which side hustles are viable.
Side hustlers building on top of LLMs have to read the contracts the way SaaS founders used to read Stripe ToS. Who owns the output. Whether the model provider trains on your prompts. Whether your use case is in policy or out. Anthropic's and OpenAI's commercial-use terms diverged meaningfully in 2025-2026, and both providers have started enforcing data-handling restrictions on lower-priced plans that the workspace and enterprise tiers do not face. A content agency on a personal plan and the same agency on a team plan are not the same business anymore. The assumption that "the APIs are roughly the same" was true in 2024. It is no longer true in 2026.
What this means for builders
If your side hustle ships AI output to clients, your model-provider tier is now a business decision, not a personal one. Read the commercial-use clause before scaling.
The platform schism is visible now, not hypothetical.
Aggregator and marketplace dependence is being penalised in 2026. SaaS-rail dependence is not.
Three concrete data points from the last twelve months. Etsy's June 2025 "creativity standards" overhaul (the platform quietly removed the "or using a templated design or pattern" language and ramped automated enforcement, suspending shops uploading more than 20-30 new listings per day with similar designs); a Change.org petition gathered thousands of signatures from frustrated sellers seeing original work flagged while obvious template product stayed up. Substack, despite the creator-economy gloss, lost about 3,000 paid creators to Beehiiv and Ghost in 2025 over its 10% fee (Grit Capital alone migrated 360k subscribers), and only just launched a formal sponsorship beta in late 2025 to replace the missing growth lever. Shopify, by contrast, posted seven consecutive quarters of GMV growth above 20% (Q3 2025 revenue +32% YoY, H1 2025 GMV $162.6B, +27%), and the typical Shopify-native DTC operator did not feel either of the above. Read these as the §02 three-platforms framing playing out: aggregator and marketplace dependence is getting more expensive, SaaS-rail dependence is fine. Expect the spread to widen by end of 2027 as more aggregators move to take-rate-plus-ad models and more rails-native businesses migrate to direct.
What this means for builders
If your distribution lives on an aggregator (TikTok, Amazon, big-platform marketplaces), the cost of staying is rising and the urgency to bootstrap to a rail you own is higher than it was in 2024.
Section 08
Predictions, on the record.
We'll be graded on these next year. Confidence labels say how strongly the data points each way. We'd rather be specifically wrong than vaguely right.
The K-shape gets sharper. The median falls again.
US median monthly earnings drop below $175 by May 2027. The average crosses $1,400. The gap widens for the fifth consecutive year. Nothing in the underlying data points to a reversal.
AI writing rates compress 40% or more.
Content writing is the most commoditised AI use case (72% adoption, lowest leverage). Per-piece rates on Upwork and Fiverr drop sharply. Expect the floor to be near-zero by Q4 2026.
Community-led builders outperform solo builders by 2×.
The 3.8× community correlation in §06 is a leading indicator, not a coincidence. As accountability tools mature, the gap between supported and solo widens. Expect this to be the most-cited finding from this report next year.
Amazon, Etsy and TikTok Shop all raise take rates.
Platform consolidation continues. The "interchangeable seller" model gets more aggressive. Independent storefronts win share for the first time in five years.
Micro-SaaS overtakes content as the highest-leverage category.
AI coding tools (Claude, Cursor, Lovable) let non-technical founders ship niche software. It is the one category where AI is a force multiplier rather than a commoditiser. Small audience, high price, recurring revenue.
Spending on community grows 3× faster than spending on tools.
The 6× misallocation in §06 starts to correct, slowly. Expect "accountability stack" to become a category builders budget for. Expect a wave of tools to be repositioned as "community" without actually being it.
A major newsletter or platform declares the "side hustle bubble".
Media outlets will overcorrect. They will read "median falling" as "the economy is dying" and miss that the operator class is bigger and more profitable than ever. This report exists to prevent that misreading.
Operator-class infrastructure becomes a recognised category in 2027.
The three failures recapped in §09 are unsolved by anything currently sold to side hustlers. Expect that to change in 2027: at least one operator-focused "thin layer" product, distinct from SaaS-tooling and from community-Slacks, becomes widely cited as the prototype. We have a directional bet about who builds it (us). The prediction stands either way.
The community infrastructure that wins is list-plus-cohort, not high-touch.
The migration toward "community as moat" does not land on premium-priced groups of five. It lands on list-plus-cohort hybrids (email plus Circle/Geneva/Skool patterns) at $20 to $50/month that scale to thousands. The unit economics of pure mastermind-tier coaching are too high-touch to absorb the volume the data points to. Expect at least three operator-focused communities to cross 10k paying members on the list-plus-cohort pattern by Q4 2027.
Section 09
The missing infrastructure.
Three structural failures, recapped one more time. Then the question the report ends on: what would actually solve them, and who builds it.
The three structural problems, one more time
Problem 01
Nobody is watching
Community is the #1 predictor of crossing $1,000/month (3.8×). 76% of builders have no accountability partner. 89% have never paid for community.
Evidence: §05, §06.
Problem 02
Nobody owns their audience
75% of side hustlers operate on platforms they do not own. Owning at least one channel is a 3.2× predictor of crossing the threshold. 11% have one.
Evidence: §02, §05.
Problem 03
The money is going to the wrong place
Sub-$1k earners spend $47/month on SaaS and $8/month on community. Community has a 3.8× revenue correlation. Tools have ~1×. The spending is six times backwards.
Evidence: §06.
The 6% who break through don't have better ideas. They have someone watching, an audience they own, and a tool-to-community spend ratio that isn't six times backwards.
What filling that hole would have to look like
Nothing on the market today addresses all three structural problems at once. Tools address one (and arguably make problem 03 worse). Communities address one. Courses address none. The missing infrastructure is something different: a thin, persistent layer that puts a small group of stage-matched peers around every builder, gives them a place that compounds rather than a Slack to drown in, and shifts the budget from per-tool subscriptions to per-relationship value. It would feel less like software and more like a co-working space that happens to be on the internet.
We have an opinion about who builds this, and we have a directional bet (we develop it in §08). We are publishing this report because the data has made the bet feel more like an obligation. But the report does not need that to be true to be useful. The more important point is that the three structural failures in §05 and §06 are unsolved, and the 94% will keep being the 94% until someone, us or otherwise, actually builds for it.
About the publisher
sidething is a London-based startup building infrastructure for the operator class. sidething.com.
The call to action
Stop building alone.
Join the community that won’t let you quit. AI-powered roadmaps. Real accountability. A structured path from idea to revenue.
Apply to join sidethingMethodology
Methodology & sources.
This report synthesises data from 25+ sources across government statistics, industry surveys, platform reports, and academic research. Every number links back to its original source.
Government & official data
- 2025Office for National Statistics (ONS): Labour Force Survey
Used for the narrow "paid second job" rate (~1.32m UK workers, ~4%).
- 2024Financial Conduct Authority (FCA): Financial Lives Survey 2024
Nearly 1,300 questions; full-population weighting. Used as a methodologically stronger triangulation source against Finder UK.
- 2025Bureau of Labor Statistics (BLS): Alternative Work Arrangements
- 2025Federal Reserve: Economic Well-Being of US Households (SHED)
20% of US adults performed gig activity in the prior month (Oct 2024 survey). 13% sold items, 9% short-term tasks, 2% rentals. Our official-source triangulation row.
Industry surveys
Platform data
- 2026Upwork: Freelancing Statistics
76.4m US freelancers · $1.5T earnings
- 2026Amazon FBA: Seller Statistics
9.7m sellers · $830B GMV
- 2026TikTok Shop: Statistics
$64.3B GMV · 475k US shops
- 2025Etsy: Marketplace Statistics
8.76m sellers · $2.88B revenue
- 2026Etsy: Q1 Earnings
GMS +5.5% · fees +10.5%
- 2026DemandSage: Upwork Platform Statistics
18m+ freelancers · $1B+ revenue
- 2026Shopify: Commerce Trends
- 2025Shopify Q3 2025 Results
Q3 2025 revenue +32% YoY. H1 2025 GMV $162.6B (+27%). Seven consecutive quarters of GMV growth above 20%. Used in §07.
- 2025Etsy: Creativity Standards (June 2025 update)
The policy revision that catalysed the 2025 AI-listing enforcement wave. Used in §07.
- 2025Substack: 2025 Creator-Economy State
Fee-driven creator defection to Beehiiv/Ghost (~3,000 paid creators in 2025; Grit Capital migrated 360k subscribers). Formal sponsorship beta launched late 2025. Used in §07.
- 2025InboxReads: The State of Newsletters 2025
77% of submitted newsletters in 2025 interested in sponsorships (vs 72% prior). First year paid-subscription growth flattened. Used in §07.
- 2026Upsella: Shopify Store Survival Rate
Only 10% survive past 90 days
- 2026Jobbers: Fiverr Freelancer Trends
Research & analysis
- 2025Goldman Sachs: Creator Economy Research
$480B by 2027
- 2025Brookings Institution: Is Generative AI a Job Killer?
- 2026DemandSage: Gig Economy Statistics
Cited for gig-market sizing only. Distinct from Goldman creator economy ($480B by 2027), which is the TAM used in this report.
- 2025eMarketer: TikTok Shop & Social Commerce
- 2025LendingTree: Business Failure Rate Analysis
22.1% fail year one
- 2026Inc Magazine: Side Hustle Growth Analysis
- 2025Side Hustle Nation: Annual Survey
- 2024Henley Business School: Side Hustle Economy Report
- 2026AutoFaceless: Freelancer Economy Statistics
- 2026Jobbers: Freelance Benchmark Report
A note on definitions
Different surveys define "side hustle" differently. That's most of why the participation numbers swing so wide.
| Source | Definition | Participation |
|---|---|---|
| Bankrate 2025 | Strict. Intentional secondary work | 27% US |
| LendingTree 2026 | Moderate. Regular secondary income | 33% US |
| Federal Reserve SHED 2024 | Official. Any gig activity prior month | 20% US |
| MyPerfectResume 2026 | Broad. Any secondary income | 72% US |
| Finder UK 2025 | Broad. Any additional income activity | 46% UK |
The truth for intentional side businesses is probably 30 to 40% in both countries.
Key definitions
Side hustle
Income-generating activity pursued alongside primary employment. Definitions vary by source. See participation table below.
Experimenter
Side hustler earning under $500/month, typically with under 12 months of activity.
Operator
Side hustler earning over $2,500/month, typically with 2+ years of activity and multiple revenue streams.
Revenue threshold
The $500 to $800/month point at which side hustles tend to accelerate rather than stagnate.
Creator middle class
The emerging cohort earning $500 to $2,500/month. Proven demand, not yet at scale.
Platform dependency
Reliance on third-party platforms (Amazon, Etsy, TikTok) where the side hustler doesn’t own the customer relationship.
Cite this report
APA (7th ed.)
sidething. (2026). The State of Side Hustles 2026 (Issue 1). The sidething Annual. https://sidething.com/state-of-side-hustles-2026/full
MLA (9th ed.)
sidething. "The State of Side Hustles 2026." The sidething Annual, no. 1, May 2026, sidething.com/state-of-side-hustles-2026/full.
Chicago (author-date)
sidething. 2026. "The State of Side Hustles 2026." The sidething Annual, no. 1, May 2026. https://sidething.com/state-of-side-hustles-2026/full.
The publisher name "sidething" is lowercase by editorial style; the report title uses standard sentence case. Cite as published.
Methodology
Synthesis of 25+ public datasets and surveys, weighted by sample size and recency. Definitions noted per stat throughout the report. Caveats on cross-source comparison surfaced inline next to each load-bearing figure.
Disclaimer
Informational only. Not financial or business advice. All projections are estimates based on current trend data. © 2026 sidething.