Catalogue Sales vs Shops: Why Lush Rebuilt After Cosmetics to Go Collapsed
Short answer: Rebuild only when you can show that customers valued the product but the old channel, cost structure or execution destroyed the economics. Preserve the smallest proven asset, such as a product, customer problem or repeat-purchase pattern, remove every obligation that made added sales consume cash, and run a paid test through a materially different operating model. Do not restart by restoring the old volume, because demand without positive contribution recreates the failure faster.
Failure can hide good demand. It can also make founders sentimental about a business customers never valued enough.
The distinction is observable. A viable core produces repeat purchases, referrals or full-price sell-through. A broken model loses cash while delivering that value. If neither side is present, there is nothing to rebuild.
Use the Demand-Economics Separation Tree
The Demand-Economics Separation Tree diagnoses a failed business in sequence. It separates what customers proved from what the operating model assumed.
| Branch | Test | Preserve when | Remove or change when | |---|---|---|---| | Customer problem | Did buyers repeatedly pay to solve it? | Payment and repeat use survived weak execution | Interest depended on discount or novelty | | Product | Did the item perform its promised job? | Refunds and complaints were controlled | Quality failures drove rework or churn | | Channel | Did each fulfilled order leave contribution? | Revenue exceeded variable channel cost | More orders increased losses | | Capacity | Could operations meet demand reliably? | Lead times and errors stayed controlled | Growth created mistakes and compensation | | Fixed structure | Could realistic volume cover overhead? | Break-even volume was attainable | Premises, payroll or debt required fantasy growth |
My view is that a founder should not “bounce back” into the same business. A credible rebuild is narrower and structurally different. If you cannot name the changed economic mechanism in one sentence, you are restarting, not rebuilding.
What the Cosmetics to Go failure proves
Before Lush, Mark and Mo Constantine and colleagues operated Cosmetics to Go, a creative mail-order cosmetics business. Mark Constantine has repeatedly described strong demand alongside destructive economics. In one founder interview about the period, he said the company could lose roughly £1 on an order. Forbes separately reported his recollection that the company lost money with every order.
The company failed. The team later founded Lush around shops and products whose theatre, scent and freshness could be experienced directly. The Guardian's history of the transition confirms that Cosmetics to Go went bust before Lush was formed, while a Basenotes interview describes the mail-order company's runaway demand and poor planning.
These are retrospective founder accounts, not complete management accounts. They do not prove that shops are inherently superior to mail order. They show that a full order book can coexist with negative unit economics, and that a rebuild must change the mechanism responsible.
Establish whether customers or discounts created demand
Recover order, refund, complaint and repeat-purchase records where lawful and available. Divide buyers into full-price, promoted and replacement orders. A customer who reordered without a discount is stronger evidence than a catalogue recipient who placed one subsidised order.
Look for products that sold through repeatedly, generated unsolicited referrals and survived price changes. Then identify fulfilment errors, free replacements and service work by product. The apparent hero may have created most customer enthusiasm and most losses.
If records no longer exist, do not reconstruct precise numbers from memory. Interview former customers about what they bought, why they returned and which alternatives they now use. Use that only to design a new paid test. Memory is a lead, not audited evidence.
Rebuild the order contribution line by line
Recreate an old order from sale price to collected cash. Include product cost, packing labour, packaging, pick errors, postage, payment fees, discounts, returns, replacements and customer service. Allocate only costs that change with the order at this stage.
Then ask which amount behaved badly as volume grew. Bulk postage may fall per parcel, but rushed packing can increase mis-picks. A larger catalogue can lift average order value while raising stock, print and forecasting costs. Economies of scale are hypotheses until the ledger shows them.
The article on [gross profit versus net profit](/business-finance/gross-profit-versus-net-profit/) explains why a healthy product margin can disappear after delivery and overhead. For a rebuild, contribution per order is the first gate because negative contribution makes every additional customer a cash liability.
Choose a channel that changes the failure mechanism
A new channel should solve the diagnosed problem, not merely feel fresh. Shops can remove parcel postage and display products sensorially, but they introduce rent, fit-out, staffing and local footfall risk. Online sales can reduce fixed retail cost, but they preserve picking, packing, delivery and returns.
| Failure in old model | Possible channel response | New risk introduced | |---|---|---| | High parcel cost on low-value orders | Minimum order, bundles or local retail | Lower order frequency or fixed rent | | Product needs touch, scent or fit | Demonstration, market pitch or shop | Labour and location dependence | | Excess range creates stock errors | Narrow assortment or made-to-order | Lost choice or longer lead time | | Returns destroy contribution | Better trial, sizing or local collection | More service cost | | Demand exceeds fulfilment capacity | Order cap and staged expansion | Slower revenue growth |
Test one channel at a time. A new shop, website, wholesale account and product range launched together make it impossible to learn which change repaired the model.
Worked example: Quarry Bath Co
Quarry Bath Co closed a catalogue operation after eight months. Customers loved its £18 bath set, but the founder had treated a £10 product gross margin as profit.
The last 1,000 shipped orders actually looked like this:
| Item per order | Amount | |---|---:| | Sale price | £18.00 | | Ingredients and product labour | (£8.00) | | Catalogue and discount allocation | (£1.70) | | Packing labour and materials | (£2.20) | | Postage and payment fee | (£4.45) | | Error, replacement and refund allowance | (£2.10) | | Contribution per order | (£0.45) |
The business lost 1,000 x £0.45 = £450 before £12,000 of fixed overhead. More orders would not have rescued it.
Former records show that 38% of full-price buyers reordered, while one citrus set caused few complaints. The founder tests that set at a permitted weekend concession. Its price remains £18, product cost £8 and card fee £0.35. Retail staffing and pitch cost total £310 over two days. She sells 74 sets.
Sales contribution before the pitch is 74 x (£18 - £8 - £0.35) = £714.10. After £310, the test leaves £404.10, or £5.46 per set after channel cost. However, a permanent kiosk quoted at £2,800 a month plus staffing would need a separate footfall test. The pop-up proves a different channel can produce contribution; it does not prove a lease works.
Quarry keeps one product, the full-price customer and in-person discovery. It discards the broad range, catalogue printing and low-value parcel structure. That is a rebuild.
These figures are illustrative. Insolvency, creditor and director duties require qualified advice based on the actual entity and jurisdiction.
Keep the old liabilities separate from the new test
Do not move assets, customer data, stock or trading activity out of a failed company informally. Ownership, creditor rights, tax, data protection, insolvency restrictions and reuse of a company name can all matter. Requirements differ by jurisdiction and legal structure. In the UK, speak to a licensed insolvency practitioner and appropriately qualified legal and accounting advisers before taking action.
A new entity does not erase personal guarantees, intellectual-property ownership or duties owed by directors. Nor does it make an old customer list automatically available for new marketing. Obtain explicit advice and document every transfer at proper value.
Commercially, ring-fence the experiment as well. Set a maximum cash loss, order limit and review date. Do not use deposits for future orders to fund unresolved obligations from the former model.
Rebuild in 30 controlled days
In week one, reconstruct the last 50 to 100 representative orders and identify the first point where contribution turns negative. In week two, rank products by full-price repeat demand, complaint cost and operational simplicity.
During week three, sell one preserved product through one changed channel with a hard order cap. Collect payment and include your labour. In week four, compare actual contribution, failure rate and repeat intent with the old model. Continue only if the test produces positive contribution and the fixed-cost break-even volume is supported by observed demand.
Related guides
Frequently asked questions
How soon can I start another business after one fails?
There is no single waiting period. Commercially, restart only after you understand the failure and can test a changed model without using money needed for living costs, tax or old obligations. Legally, the answer depends on the country, entity, insolvency process, director status, company name and any restrictions imposed. In the UK, obtain advice from a licensed insolvency practitioner and qualified solicitor before transferring assets or resuming similar trade. Speed is not the objective. A two-week restart with the same negative order economics merely shortens the route to a second failure.
Does strong demand mean the failed business was worth saving?
No. Strong demand is one asset, not a complete business. Orders can be unprofitable, operationally impossible or dependent on discounts. Verify full-price repeat purchasing, contribution after delivery and the fixed volume needed to break even. If customers value the product but no available price or channel covers its cost, the demand is commercially unusable. Conversely, a small group of profitable repeat buyers can justify a narrow rebuild even when the original company was too large. Preserve evidence, not the emotional importance of the old venture.
Should I keep the same brand name?
Keep it only when you lawfully control it, customers retain trust and the name does not carry misleading associations with the failed entity. Check intellectual-property ownership, insolvency restrictions, domains, social accounts and customer expectations. A familiar name can retain useful recognition, but it may also import complaints and promises the new model cannot fulfil. Laws on reuse of company names and asset transfers vary, particularly after insolvency. Obtain qualified advice before trading under a similar name. The brand decision follows legal clearance and customer evidence, not nostalgia.
Is opening a shop safer than selling online?
No. A shop replaces some parcel and discovery problems with rent, fit-out, staffing, stock security and location risk. It is safer only when observed local contribution covers those fixed costs under a realistic downside. Test with a concession, pop-up or short agreement before a long lease where possible. Online may remain superior for high-value, low-return products or geographically dispersed customers. Compare both routes at the same sales volume and include founder time. The Lush history is evidence about one company's model, not a general rule that shops beat mail order.
What records should I keep from the failed business?
Preserve lawful copies of accounts, bank statements, invoices, order and refund data, supplier agreements, complaints, marketing results and product-cost records for the required retention period. Do not remove records or customer data that belong to an insolvency process or another entity. Access and retention rules vary by jurisdiction, tax and data type, so follow professional advice. Commercially, the most useful records link each order to price, channel, variable cost, error, refund and repeat purchase. Aggregated revenue alone cannot reveal which core deserves another test.
How much money should I risk on the rebuild?
Set the amount from a bounded experiment, not from what you have left. Cost the smallest batch, channel test, professional advice, refunds and a downside allowance. The loss must be affordable without borrowing against essential living costs or using restricted company money. A useful first gate is one complete sales cycle with a capped order volume. If contribution is negative, stop before fixed commitments. If positive, fund the next stage from a separate working-capital calculation. Personal and company finance implications require qualified local advice.
When should I decide there is nothing to preserve?
Stop when customers did not pay repeatedly at a viable price, the product failed its promised job, or every realistic channel leaves negative contribution. Also stop when the asset you hoped to preserve cannot be lawfully transferred or would require unaffordable capital to operate. Founders sometimes interpret attention, press or a long order list as demand, but none compensates for refunds and losses. Write the preservation claim before testing it. If one capped paid experiment cannot support that claim, closing cleanly may be the better decision.
Comments (0)