I Have Enough Raw Materials for 500 Candles but Sales Are Growing Unpredictably — How Much Safety Stock Should I Maintain?

Split the trend from the swing

Here are eight weeks of candles poured by an example business. They are illustrative numbers, chosen to look like a real small business growing unevenly: a market weekend, a slow week, a stockist reorder, a quiet patch, then a jump.

The average is 110 a week and the standard deviation of the raw numbers is 29.9. Many guides would plug those straight into the safety-stock formula. For a growing business that is a double mistake. The average understates next week: the trend line, fitted by ordinary least squares, rises about 9.6 candles a week and puts week 9 at 153, not 110. And the raw standard deviation overstates the swing, because part of the spread between week 3 and week 8 is simply growth. Measured around the trend line, the swing is σ = 19.8.

You do not need statistics software for this. A spreadsheet's trend function (or a chart with a linear trendline) gives the trend values, and the standard deviation of the "swing" column gives σ. If that feels like too much, the simpler methods in the next section still work; they just do not adapt as neatly to how jumpy your sales are.

Five ways to size the buffer

Here are five common methods, each run on the same example, with the week 9 forecast of 153 candles as expected demand and a 1-week example lead time. The last two use σ = 19.8 around the trend. The example also assumes deliveries have taken as long as 9 days and vary by roughly ±2 days; both are illustrations, not CSI delivery times.

The spread between methods is wide, from 33 to 88 candles, and that is the point. The simple methods are guesses about variability; the statistical ones measure it. Notice what the last line reveals: in this example, delivery spread matters more than sales spread. At 153 candles a week, two days of delivery uncertainty is worth about 44 candles on its own, more than the 19.8-candle weekly swing. If your deliveries are erratic, fixing that, or measuring it honestly, does more than any amount of sales forecasting.

Choosing a service level

The z value in the formula is a choice about how often you are willing to run out during a lead time. It is a business decision, not a technical one, and it should differ by material: running out of wax stops everything; running out of a C-class scent disappoints a few customers.

Each step up in service level costs more stock for less extra protection. Going from 90% to 95% adds about a third to the buffer; from 95% to 98% adds about a quarter again. A reasonable starting pattern is 95% for shared materials (wax, wicks, jars) and your A-class scents, and 90% or a one-sealed-bottle rule for slow scents. My Candle Business Uses 20 Different Fragrance Oils — How Do I Set Separate Reorder Levels for Fast-Selling and Slow-Selling Scents? sets out the scent classes.

The service level here is the chance of no stock-out during a lead time, the simplest definition and the one the z values above describe. It is not the same as the share of customer orders filled, which is usually higher. Keep the definitions straight if you compare notes with other makers.

From candles to kilos and rupees

The example candle is 200 g at an 8% test load: 184 g of Coconut Soy Wax CSI 468 at ₹637.00 a kilo (5 kg rate), 16 g of fragrance priced here at Cinnamon Vanilla's 500 g rate of ₹4.60 a gram, one Eco Thin C1 wick at ₹5.90 and one White Matte Glass Jar at ₹70.80. That is ₹267.54 of materials per candle, excluding labour, packaging, freight and GST.

The buffer that includes delivery spread costs ₹12,574.38 more than the sales-only buffer. That is the price of not knowing when deliveries will land. If you can narrow it, by ordering earlier in the week, by consolidating orders, or simply by measuring your real delivery times, the buffer shrinks and the cash comes back.

The same buffer does not have to be held in every material. Wicks are so cheap that holding a larger buffer costs almost nothing: 100 Eco Thin C1 wicks are ₹590.00. Jars are the expensive, bulky part: 80 White Matte jars are ₹5,664.00. Hold the statistical buffer for jars and wax, and round wicks up generously. I'm Constantly Placing Emergency Raw-Material Orders — How Do I Calculate Minimum Stock Levels for Wax, Fragrance, Wicks and Jars? explains why cheap items deserve bigger minimums.

What your 500 candles really cover

Back to the question. You have materials for 500 candles. At the week 9 forecast of 153 a week, that is 3.3 weeks of pouring, not the 4.5 weeks the old average suggests. With a 1-week lead time and 79 candles of safety stock (95%, with delivery spread), the reorder point is 233 candles' worth. So you can pour about 267 candles, roughly 1.7 weeks, before the reorder point fires.

By week 10 the forecast is 163, so the reorder point creeps up again. That is why the calculation belongs in a monthly routine, not a one-off. Add each new week to the data, drop the oldest, refit the trend and recompute σ. If growth slows, the trend flattens and the reorder point stops rising on its own. If a week lands far outside the swing, look for a reason (a one-off order, a stock-out that suppressed sales) before letting it into the data.

Safety stock answers one question: how much to keep for surprises. The other two parts of the stock picture, when to reorder and how much to order, are covered in I'm Growing Quickly but Don't Want to Run Out of Wax — How Do I Set a Reorder Point Based on Monthly Consumption and Supplier Lead Time? and My Candle Business Is Growing but All My Cash Keeps Going Back into Wax, Fragrance Oils and Jars — How Much Raw-Material Inventory Should I Actually Hold?. When cash is too tight to fund every buffer at 95%, I'm Scaling Candle Production but Don't Have Enough Cash to Bulk-Buy Everything — Which Raw Materials Should Receive Purchasing Priority? sets the order in which materials get it, and My Supplier Gives Better Prices for Larger Orders but I'm Worried About Dead Stock — How Do I Decide the Maximum Inventory Worth Holding? puts a ceiling on how much any line should hold.

Frequently asked questions