Scaling and Direct
When one pot stops being enough, and the numbers that decide the next step.
Scaling the Batch
When one pot stops being enough.
The kitchen batch that works at three vessels does not automatically work at thirty. Wax mass has thermal inertia: a 1.2 kg melt cools measurably faster than a 6 kg melt, so the pour temperature that gave you glossy tops last week gives you sink holes this week. Fragrance-oil dispersion is worse at higher volumes because the stir path is longer. And every extra pour is a chance for temperature drift between the first and last vessel of the run.
Scale in disciplined doublings. If today's proven batch is twelve vessels, the next tested batch is twenty-four, not the fifty you actually want to ship. Every doubling triggers a fresh four-burn protocol on the first and last vessel of the run to prove homogeneity.
- 1–12 vessels (hobby)
- Melt on a bain-marie. Pour temperature 55–60°C, tolerance ±2°C.
- 13–36 vessels (weekend maker)
- Thermostatic melter, 8–12 L. Pour temperature 55–60°C, tolerance ±2°C. Stir at 500 rpm for 2 min after fragrance drop.
- 37–100 vessels (workshop)
- Jacketed melter, 20–30 L. Pour temperature 55–60°C, tolerance ±1°C. Powered stirrer, timed 3 min. Pour within 12 min of fragrance drop.
- 100+ vessels (small-batch production)
- Two jacketed vessels in parallel plus a decanting pot. Pour temperature 55–60°C, tolerance ±1°C. Fill from a levelled decant pot to hold surface finish across the run.
- Batch doubled, not tripled or quadrupled in one step
- First and last vessel of the run tagged for four-burn protocol
- Melt-to-pour clock recorded, target under 12 min
- Ambient bench temperature logged for the whole run
- Cure shelf capacity confirmed before the pour begins
| Symptom | Likely cause | Fix | Prevent |
|---|---|---|---|
| Frosting appears on the last third of a scaled batch, not the first. | Melter temperature drifted below the target as pour progressed and no reheat was scheduled. | Hold the melter at pour temperature ±1°C with a jacketed vessel or return to the melter between pours; log temperature every 6 vessels. | |
| Sink holes appear only on the larger batch size. | Thermal mass of the bigger melt is cooling more slowly, so surface skins before the interior settles. | Drop pour temperature 2°C and pre-warm vessels to 40°C to slow the surface skin; retest the four-burn protocol. |
The Threshold
The UK VAT line, and how to cross it on purpose.
UK VAT registration is triggered by a rolling twelve-month turnover of $114,300. Cross it accidentally and your net margin on every candle drops by roughly 17 percent overnight, because you owe HMRC one-sixth of the gross on every VATable sale but your customers have not seen a price rise. Plan the crossing and it becomes a pricing decision instead of a margin collapse.
You do not accidentally register for VAT. You either plan for it or you get taxed for not planning.
- £1,000 turnover / year
- HMRC trading allowance exhausted. Self-assessment registration required.
- £30,000 turnover / year
- Insurer-required product-liability cover typically moves from starter to standard band. Book at least £2m cover.
- £85,000 turnover
- Old VAT threshold. Use as an early-warning trigger to model the crossing.
- £90,000 rolling 12 months
- VAT registration required within 30 days. Making Tax Digital for VAT applies immediately.
- 50 tonnes packaging / year OR £2m turnover
- UK EPR small-producer reporting duties begin. Track packaging materials by weight from day one.
- 200 tonnes packaging / year OR £2m turnover
- UK EPR large-producer duties: full reporting and fees per DEFRA schedule.
- Rolling 12-month turnover checked at the end of every month
- Reprice modelled and communicated 2 months before the projected crossing
- Making Tax Digital-compatible bookkeeping in place before crossing (not after)
- Packaging materials logged by weight per SKU from month one
- Product-liability policy limit reviewed at each threshold
| Symptom | Likely cause | Fix | Prevent |
|---|---|---|---|
| Accountant flags an unexpected VAT liability at year-end. | Rolling-12-month test was checked annually instead of monthly; the crossing happened silently in month eight. | Rebuild the last 12 months of turnover monthly, register within 30 days of the crossing month, and pay the back-VAT from reserves rather than repricing retroactively. | |
| EPR reporting demand arrives without warning. | Packaging tonnage was never logged; the threshold was crossed on primary and secondary packaging combined. | Weigh a representative sample of each SKU's packaging, back-calculate annual tonnage, and register on the DEFRA portal. |
DTC Economics
The decision matrix for owning the customer versus renting them.
The fee-stack table in Act 4 showed the numbers: marketplaces cost 11 to 20 percent of RRP, own-domain checkout costs around 3 percent after card fees. The gap is real and it compounds every month you run the shop. The friction people underestimate is not the money; it is traffic. Marketplaces bring shoppers to you; your own domain requires you to bring shoppers to it.
The volume floor where own-domain beats a marketplace on take-home is lower than most makers think, because the marketplace fees rise with turnover but the DTC platform fee is fixed. A ballpark: at around fifty orders a month on a $38 RRP, the Shopify + Stripe stack is already netting more per order than Etsy even before Offsite Ads erosion.
- Etsy Basic
- Fees ~£4.50/order = £450/mo. No platform fee. Traffic included.
- Amazon Handmade
- Fees ~£4.50/order = £450/mo. No platform fee. Traffic included.
- Shopify Basic + Stripe
- Fees ~£1.20/order = £120/mo + $39 platform. Traffic self-generated.
- WooCommerce + Stripe
- Fees ~£1.20/order = £120/mo + ~£20 hosting. Traffic self-generated.
- Marketplace baseline take-home documented per SKU
- Own-domain platform selected (Shopify or WooCommerce) with a public URL live
- Card processor connected, sandbox charge tested end-to-end
- Email capture live on every product page from day one
- First 6-month acquisition budget ring-fenced from the fee savings
- Repeat-customer flow (post-purchase email, refill nudge) mapped
| Symptom | Likely cause | Fix | Prevent |
|---|---|---|---|
| Own-domain traffic is a fraction of marketplace traffic after 3 months. | No traffic strategy: the store is live but nothing feeds it. | Ring-fence 5 to 10 percent of gross to acquisition (email list build, one paid channel, one content channel) and measure sessions weekly. | |
| DTC net margin is worse than the marketplace it replaced. | Fee savings were spent on discounts to seed reviews instead of on acquisition. | Hold RRP; move the discount budget into a documented acquisition line and evaluate CAC after 90 days. |
The First Hire
Fulfilment, ops, or nobody, decided with numbers.
Every hour spent packing candles is an hour not spent making, marketing, or planning the next range. The economics of the first hire are simple: at UK real-living-wage rates (call it $15/hour), a four-hour fulfilment shift costs $61. If that shift covers thirty orders that otherwise ate a full working day, you have bought back six hours at a net of a few pounds per hour after employer costs. That is the trade you keep making until the shop hits a real ceiling.
- Fewer than 30 orders / week
- Do not hire. Compress packing to one afternoon per week with a batched-shipping calendar.
- 30–75 orders / week
- First hire is a fulfilment assistant, 6–10 hours / week. Keep making and admin on your desk.
- 75–150 orders / week
- Fulfilment assistant to 15–20 hours / week or a 3PL trial for a subset of SKUs. First maker apprentice for pouring days.
- 150+ orders / week
- 3PL for standing SKUs, in-house team of 2 for new launches and wholesale, you focus on product and channel.
- Order volume tracked weekly, not monthly
- Fulfilment time per order measured with a stopwatch, not estimated
- Employer's National Insurance and pension costs modelled before hiring
- Written pick-pack-ship SOP in place before the assistant starts
- 3PL quotes gathered at the 75-orders/week band, held on file for later
| Symptom | Likely cause | Fix | Prevent |
|---|---|---|---|
| Assistant is hired but the founder is still packing evenings. | No written SOP: the assistant packs the easy orders and hands the awkward ones back. | Write a one-page pick-pack-ship SOP covering fragile vessels, gift wrap, and address exceptions; hand every order to the assistant for one full week. | |
| 3PL trial produces more customer complaints than in-house fulfilment did. | 3PL is picking generic bubblewrap and skipping the care card. | Move the care card and any inserts into a pre-kitted mailer that the 3PL cannot deviate from; audit the first 20 orders visually. |