When Should a Side Project Become the Main Business? Unsplash's 20,000-Signup Test
Short answer: Separate a side project when demand continues without promotion, users value it independently of the parent business and supporting it now takes resources from the parent. Give it a distinct budget and accountable owner for 90 days before a legal split. Do not spin it out from one traffic spike, and do not keep it merely because it sends leads to the original business.
A successful side project creates two tempting stories. It may be the cheapest marketing channel the main business has found, or it may be a larger business being constrained by its parent.
Traffic cannot decide between them. You need to know whether the project creates independent value, has a plausible economic role and deserves scarce attention more than the business it was built to support.
Unsplash outgrew its original job
Unsplash began shortly after the creative-services marketplace Crew as a way to help Crew grow. Mikael Cho's account of the eventual separation says the first version was a Tumblr page with Dropbox links and barely worked. Even so, 20,000 people signed up on the first day.
After several months, it had 80,000 members and more than one million downloads. The team selected ten photographs every ten days, taking about an hour, and nobody worked on the project full time. It also referred more customers to Crew than any other source.
Continued growth led Crew to assign one co-founder full time and rebuild Unsplash as a standalone site. The team grew from one to five people over two years. Eventually the two operations competed for focus.
Cho wrote that the companies spent six months assessing a split. Crew had about US$5 million of the US$8.5 million raised two years earlier still in the bank and was close to profitable. In January 2017, Crew and Unsplash became independent companies with separate teams and budgets.
This was not a decision to abandon a failing parent for a popular page. Crew was operating, the side project was its strongest referral source and both had claims on the same resources.
My view is that attention conflict, not audience size, is the decisive separation signal. If two propositions need different customers, teams and investment choices, keeping them together for convenience eventually makes both slower.
Use the Side-Project Separation Gate
The Side-Project Separation Gate tests four forms of independence.
| Gate | Evidence | Failure signal | |---|---|---| | Demand | Use continues without parent promotion | Spike fades after launch | | Identity | Users value the project for its own outcome | Most users only want the parent offer | | Economics | A credible payer or funding route exists | Growth creates only unbounded cost | | Attention | Project needs different priorities and ownership | Same team can support both cheaply |
Passing three gates justifies a ring-fenced test. Passing all four with stable evidence can justify a company split. Legal separation comes last.
Remove the parent's promotional help
Run a quiet period in which the parent stops featuring the project beyond existing links. Measure new users, repeat use, referrals and direct searches. If acquisition collapses, the project may still be a strong campaign, but it has not shown independent demand.
Look beyond registrations. Unsplash's early downloads demonstrated use, while word of mouth and continued growth with little staff time showed that the first-day attention was not the whole story.
Choose a natural cycle. A weekly service may reveal persistence in eight weeks. A seasonal project needs longer. Do not declare independence from a weekend of publicity.
Price the value sent back to the parent
A side project can be worth more as a channel than as a separate company. Attribute parent leads and customers conservatively. Record a single primary source and compare their conversion, contribution and retention with other routes.
Then calculate what the side project consumes: staff hours, hosting, support, brand risk and delayed parent work. A project sending £20,000 of annual contribution to the parent may still be unattractive if it consumes £35,000 of scarce capacity.
The alternative is not always separation. You can keep it as a bounded marketing property, reduce its scope, license it, sell it or close it.
Identify an independent economic customer
Users do not have to pay directly, but somebody must support the service without corrupting the value that created demand. Possible payers include users, commercial partners or customers buying an adjacent service. Each changes incentives.
Do not bolt on a revenue idea merely to justify the split. Test willingness to pay or partner commitment before forecasting. If no credible economic customer exists, the project may deserve a fixed marketing budget rather than a new company.
Worked example: MaterialIndex
Northbank Interiors runs a £720,000-a-year design consultancy. It publishes MaterialIndex, a free directory of low-waste building materials, to attract commercial clients. These figures are illustrative.
MaterialIndex attracts 18,000 monthly visitors after nine months. Only 300 arrive from Northbank's site, while 11,700 arrive through direct visits or search and 6,000 through referrals. Forty consultancy enquiries credit the directory over the year. Eight become projects with average contribution of £4,200, producing 8 x £4,200 = £33,600 for Northbank.
The directory consumes 55 staff hours monthly at £38, hosting and data cost £850, and support costs £420. Annual direct cost is (55 x £38 + £850 + £420) x 12 = £40,320.
As a lead channel alone, it loses £40,320 - £33,600 = £6,720 before central overhead.
Three manufacturers offer £9,000 each annually for clearly labelled enhanced company pages, provided editorial inclusion cannot be bought. Potential commercial revenue is 3 x £9,000 = £27,000.
| Test | Evidence | |---|---| | Independent demand | 98.3% of visits originate beyond Northbank's site | | Independent use | 4,600 people return at least monthly | | Economic route | £27,000 of written partner intent | | Attention conflict | 55 monthly hours delay client work |
Northbank creates a separate 90-day budget, assigns one owner and charges one partner. It does not transfer staff or form a company yet. The gate is £6,000 collected partner revenue, directory contribution at least break-even and no fall in user repeat rate. A successful test would justify separation. A failed test would return MaterialIndex to a smaller, explicitly funded marketing role.
Ring-fence before you restructure
Give the project a separate internal profit and loss account, owner, objectives and resource cap. Allocate shared staff at replacement cost. Record any parent referrals in both directions.
This exposes whether the side project survives when it pays for what it uses. It also tests leadership. A project that works only when the parent founder makes every decision has not achieved operating independence.
Company, tax, intellectual-property, employment, investment and data-transfer consequences vary by jurisdiction. A legal split can be costly and difficult to reverse. Obtain qualified local legal, tax and financial advice before moving assets, people, contracts or funding.
Protect what made the project grow
Separation often introduces pressure to monetise quickly. Write the non-negotiable user value before adding revenue. For a community or content business, that may include editorial independence, access or attribution.
Measure whether the new economic model reduces repeat use, supply or trust. One sponsor cheque is not success if the core community begins leaving.
Also protect the parent. Agree brand use, shared services, customer ownership and referrals. Friendly informal arrangements become ambiguous when priorities or investors change.
Make the call in 90 days
In days one to 15, separate traffic, costs, staff time and parent referrals. Define the four gates and stop moving historical numbers between the two stories.
For the next 60 days, give the project a bounded budget and one accountable owner. Reduce parent promotion, test one credible revenue route and measure repeat use.
In the final 15 days, choose one of four outcomes: separate, retain as a channel, sell or close. If separating, obtain professional advice and plan contracts, staff, data and cash. Do not let “temporary” shared ownership continue indefinitely.
Related guides
Frequently asked questions
How much traffic proves a side project should be a business?
No traffic number alone proves it. Ten thousand users with no repeat use, independent need or economic route can be less valuable than 500 specialist users who pay. Measure where users come from, what they do, whether they return and what supporting them actually costs. Compare activity during a period with reduced parent promotion. Traffic is evidence of attention, not of a stand-alone business. Set thresholds from the project's likely model and service burden rather than copying Unsplash's first-day sign-ups.
Should I move full time onto the side project?
Only after a bounded test shows independent demand and a better use of your time than the parent business. Price the contribution and learning forgone in the parent. You can assign a co-founder or employee temporarily, as Crew did before full separation, while maintaining a defined budget and review date. Do not divide yourself indefinitely between two urgent operations. If neither can progress without you, the immediate issue is ownership and systems, not which project has the more exciting growth chart.
What if the side project is my best marketing channel?
That is a strong reason to keep funding it, not necessarily to spin it out. Calculate attributable parent contribution after the project's full cost. Then test whether independent monetisation would improve or damage that channel. Separation may introduce commercial incentives that reduce trust and parent referrals. Keep it as a marketing property when its highest value clearly returns to the parent and attention conflict is manageable. Give it a budget and owner so “free leads” do not conceal expensive work.
Does the side project need revenue before separation?
Not always, but it needs a credible funding path and a commercial reason that path should work. Written partner interest, paid pilots or strong willingness-to-pay evidence is better than a general plan to monetise later. A venture-funded project may separate before revenue, but that does not remove the need to understand who eventually pays. If the user value depends on being free, test an aligned payer carefully. Growth funded only by the parent's available cash is dependence, not stand-alone economics.
Can I run both businesses under one company?
Yes. Separate management accounts, accountable owners and explicit priorities can provide much of the decision clarity without immediate legal restructuring. One company may simplify shared staff and contracts, while separate entities can clarify investment, liability or ownership. The right structure depends on jurisdiction, tax, regulation, intellectual property and funding. Obtain qualified legal and tax advice before changing it. Operational separation should come first because a new legal entity by itself will not fix unclear customers, shared attention or weak economics.
When should I close the side project instead?
Close or reduce it when demand fades without repeated promotion, users do not complete a valuable action, costs exceed its strategic contribution and no credible economic route appears within the test limit. Preserve obligations to customers, staff and partners, and handle data lawfully. A project can be popular and still be the wrong use of scarce attention. Closing it is not a judgement that the work had no value. It is a decision that another use of the same money and time has stronger evidence.
Comments (0)