Starting with a small budget can be useful. It forces a better question than “what business could make the most money?”
The better question is: what is the cheapest credible test of a problem that somebody may pay to solve?
£500 does not buy a business. It does not buy demand, trust, skills, insurance, a laptop, a vehicle or the time required to learn. What it can buy is a disciplined first experiment: a clear offer, a basic way to reach customers and enough room to find out whether the model deserves more time.
I would not use a small budget to chase “passive income”. I would use it to test one defined customer problem, keep the fixed costs low and learn where the real constraint sits. Sometimes that constraint is customer acquisition. Sometimes it is delivery capacity, regulation, support or simply the fact that the work is less attractive once it becomes real.
What “under £500” actually means
The figures below are illustrative launch budgets for someone who already has a usable phone and computer. They exclude wages, a vehicle, stock commitments, working capital, advertising spend, insurance premiums, tax and any sector-specific licence, qualification or permit. A model can be cheap to start and still be expensive to run well.
For a UK sole trader, the £1,000 trading allowance is based on gross trading income, not profit. Once trading income exceeds that threshold, registration and record-keeping obligations can apply. VAT is a separate rolling turnover test, not a profit test. These rules change, so use the official links at the end of this article rather than treating this as individual tax advice.
The seven models, compared honestly
| Model | Lean test budget | First customer route | Real constraint |
|---|---|---|---|
| Productised freelance service | £15–£145 | Specific outreach and referrals | Founder time and scope creep |
| Local coordination service | £40–£197 before insurance | Local operators and direct relationships | Reliability, access and liability |
| Simple workflow automation | £40–£192 | One low-risk admin process | Data, maintenance and support |
| Digital product | £0–£200 | Existing audience or clear problem | Demand and distribution |
| Niche newsletter | £0–£200 | Useful public writing and opt-in | Editorial consistency and trust |
| Print-on-demand | £50–£350 | Defined community or creator partnerships | Margins, quality and IP |
| Niche content library | £50–£300 | Search, distribution and email | Slow feedback and sustained production |
1. Productised freelance service
This is the most straightforward model on the list. The key is to sell a defined outcome to a defined type of buyer, rather than presenting yourself as available for every kind of work.
Examples could include a monthly Google Business Profile update pack for local trades, a fixed-scope website-copy refresh for independent accountants or a recurring reporting spreadsheet for a small service firm. The offer needs a clear deliverable, a revision limit, a turnaround time and a price.
The initial budget can be modest: a domain and email, a simple portfolio of three representative samples, basic proposal and invoice templates, and a small amount of direct outreach. The difficult part is not setting these up. It is finding a problem that a buyer recognises and trusts you to solve.
First test: choose one buyer type, build three relevant samples and offer one paid, tightly scoped project. Do not give away the entire service as a “trial”.
Where it breaks: bespoke requests, endless revisions and a pipeline built only on chasing strangers. Productisation only works when the work genuinely becomes repeatable.
2. Local cleaning or routine-maintenance coordination
This is not a generic “start a cleaning business” suggestion. A more defensible version is a narrow local coordination service: for example, turnover cleaning and minor-maintenance coordination for a specific group of serviced accommodation hosts or small offices.
The operator’s value is in a reliable process: finding vetted providers, agreeing a job specification, managing access, collecting a photo report and dealing with exceptions. Early on, the cleanest commercial arrangement may be for the client to pay the provider directly while the coordinator invoices a transparent management fee.
First test: secure two or three dependable providers before advertising, then offer a defined pilot for one local property type.
Where it breaks: no-shows, key handling, quality disputes, scattered geography and unclear responsibility when something goes wrong. This is operational work, not passive income. It also becomes much more complex if you take client funds, move waste, employ people or arrange regulated trades.
3. Simple workflow automation service
There is a real opportunity in helping small businesses tidy up repetitive, low-risk administration. There is far less credibility in claiming to sell a vague “AI agent” that replaces staff or makes important decisions without oversight.
A useful first offer might connect an enquiry form to a spreadsheet or CRM, send a simple acknowledgement, create an internal task and produce a weekly summary. The client should own the relevant software accounts and payment details. That keeps the ongoing cost visible and makes a proper handover possible.
First test: map one real process with a clear trigger, output, human approver and failure path. Use dummy data for demonstrations.
Where it breaks: poorly understood client processes, live personal data, changing software permissions and support requests that outgrow the original setup fee. If personal data is involved, the client and provider need to understand their data-protection roles and document the arrangement appropriately.
4. Digital products
Digital products can have attractive marginal economics because there is no physical inventory to replenish. That does not make them easy. A generic template or short guide with no obvious buyer problem is simply a file on the internet.
The strongest starting point is an asset that comes from specific experience or repeated customer questions: a spreadsheet, checklist, database, planning tool or guide that saves a buyer time or reduces an avoidable mistake. Platforms can keep fixed costs low, but transaction and discovery fees still affect the economics. Gumroad, for example, charges a fee on each sale, while other platforms use different payment and merchant-of-record models.
First test: describe the outcome before creating the full product. Offer a sample, speak to likely users and see whether anyone will exchange an email address, feedback or money for it.
Where it breaks: treating creation as validation, ignoring refunds and support, or assuming a platform will create distribution for you. Consumer rules also matter. For digital content supplied immediately, cancellation rights can only be handled in a specific way with the buyer’s express consent and acknowledgement.
5. Niche newsletter
A newsletter is best understood as an owned media asset. It can eventually support subscriptions, sponsorships, products or services, but those outcomes depend on a reader trusting that each edition is worth opening.
A strong niche has a recurring information need and a reason for readers to hear from you rather than a larger publisher. It might be a local business sector, a profession with changing rules, a practical hobby or a carefully defined research beat. Publishing platforms can be free to start, but the real cost is the repeated work of research, writing, list hygiene and audience service.
First test: write several useful public pieces on the same narrow topic and invite readers to opt in for a focused recurring briefing.
Where it breaks: an overly broad promise, no distribution plan and the temptation to buy an email list. Marketing-email rules require a lawful route to contact individuals and a real unsubscribe process. Trust is the asset.
6. Print-on-demand
Print-on-demand removes the need to buy a garage full of stock. It does not remove the need for a distinctive idea, good product samples, customer service and enough margin after fulfilment, shipping, platform fees and refunds.
This can work when the product is designed for a defined community rather than a broad marketplace. A small collection for a niche sport, local identity or professional in-joke may have a more credible acquisition route than a catalogue of generic motivational designs.
First test: order samples, test quality and delivery before asking others to buy. Check intellectual-property rights before using any phrase, image or reference that is not your own.
Where it breaks: thin margins, supplier failures, returns and a design that is either unremarkable or legally risky. No inventory does not mean no operational responsibility.
7. Niche content library
A content site is the slowest model here. I would not build one around the hope that display advertising creates easy income. Search visibility takes time, platforms change and a useful archive demands updates.
The better use of a niche content library is as an audience and trust layer. It can support a newsletter, a service, a digital product or carefully disclosed affiliate recommendations. That makes the economics more coherent than relying on early advertising revenue.
First test: publish a small cluster of genuinely useful pages around a narrow question and measure whether people read, subscribe, reply or ask for help.
Where it breaks: producing a large volume of thin content, mistaking impressions for demand and waiting too long to speak to actual readers.
How I would choose between them
There is no universally best model. The useful choice is the trade-off you can sustain.
- If you have a valuable skill and can tolerate selling, start with a productised service.
- If you enjoy systems and understand low-risk operations, test a simple workflow-automation offer.
- If you have local relationships and can manage exceptions, a tightly defined coordination service may be viable.
- If you have a specific asset or audience insight, test a digital product.
- If you can reliably research and write for a narrow group, build a newsletter or content library over time.
- If you have a genuine community insight and an eye for product quality, test a small print-on-demand collection.
In every case, I would keep the first test reversible. Avoid annual software commitments, large inventory orders, complicated corporate structures and broad paid advertising before a real customer has shown that the problem matters.
Where I could be wrong
A low launch budget can make a model look more accessible than it is. Someone may have the cash but not the credibility, sales ability, qualifications, local network or appetite for uncertainty that the model requires. Equally, a higher-cost business can be the better fit if it solves a real problem and the operator understands the market.
The goal is not to start something because it can technically be launched for under £500. The goal is to learn, cheaply and honestly, whether there is a repeatable exchange of value worth building on.
Useful next steps
- Choose one model, one buyer and one specific problem.
- Write down the smallest paid test you could run in the next 30 days.
- List the assumptions that have to be true for it to work.
- Speak to prospective customers before committing serious capital.
- Keep records from the first transaction and review the tax, consumer-law and data-protection implications that apply to your model.
This is general educational information, not legal, tax, financial or business advice tailored to your circumstances. Rules, fees and thresholds can change. Check the official sources below and take professional advice where the risk or commitment is material.
Sources and further reading
- GOV.UK: Register as a sole trader
- GOV.UK: VAT thresholds
- GOV.UK: Online and distance selling
- ICO: Electronic mail marketing rules
- Gumroad: Fees
- Substack: Costs
- Etsy: Fees
- HSE: Starting a cleaning business
Related: The Business of Financial Freedom, Financial Independence and the £1m Path Calculator.
