My Cheaper Packaging Saves ₹20 per Unit but Breakage Rises by 3% — How Do I Calculate Whether the Change Actually Saves Money?

The formula, term by term

The two numbers in your title are your figures: a ₹20.00 saving per unit from the cheaper packaging, and a 3% rise in breakage. To compare them they must be in the same unit, rupees per unit shipped. The saving already is. The breakage rise becomes rupees when you multiply it by what one broken order costs:

Net saving per unit = S − Δb × C, where S is the packaging saving per unit (₹20.00), Δb is the rise in broken units per unit shipped (3 points = 0.03), and C is the full cost of one broken order. Rearranged, the change breaks even when C = S ÷ Δb: ₹20.00 ÷ 0.03 = ₹666.67. If a broken order costs you less than ₹666.67, the cheaper pack saves money; more, and it costs money. You can also turn it round: for a given C, the largest rise the saving can pay for is S ÷ C. On the worked replacement cost below, ₹20.00 pays for a rise of up to 4.70 points.

Which 3%? "Breakage rises by 3%" can mean two very different things. Read as percentage points (say 2 broken per 100 becomes 5 per 100), it is the large rise this page works through. Read as 3% of your current rate, 2 per 100 becomes 2.06 per 100, a rise of 0.06 points: it costs ₹0.26 a unit and the break-even C would be ₹33,333.33, so the saving is safe. Check which one your trial measured before you go further.

Per unit or per order? If customers often buy two or three candles, a broken parcel may mean one broken jar out of three. Keep both numbers in the same unit: breaks per unit shipped with cost per broken unit, or breaks per order with cost per broken order. Mixing them overstates or understates the rise by your average units per order.

Before reading the worked numbers, set up the calculation with your own. Every row below is something you can pull from your invoices, courier statements and claims log in an afternoon; the worked column is only there to show the shape.

What one broken order really costs

Everything turns on C, so build it line by line. Here is a worked example: a 200 g candle at 8% (16 g of oil, 184 g of wax, load as a share of total weight) with CSI wax, oil and wick, your own jar and label, sold at a worked ₹549.00. British Rose's page states "candles (up to 10% soy, up to 6% paraffin)" and Luxury Soy Wax Chunks states "up to 13% fragrance load capacity", so 8% sits inside both; your burn tests confirm it.

Packaging falls from 17.1% of COGS to 10.7%. Now the cost of a broken order. There are two routes, and they cost very different amounts.

Replacement route. You send a new candle. The cost is the candle's materials and packing, a second courier charge and your time, plus whatever goodwill you offer. The selling price is not lost: the buyer still paid it. That makes C ₹425.91 here.

Refund route. You refund the buyer. Compared with a parcel that arrived safely, you lose the whole selling price, because the materials and first courier charge were spent either way. On a marketplace, a damage claim may also bring a fee, a penalty or a hit to your seller metrics; that is a reader input, shown here as a worked ₹100.00. C becomes ₹649.00. Where a buyer is offered the choice, a replacement is usually cheaper for you whenever the price is higher than the replacement cost, as it is here.

The lost customer. A buyer whose first parcel arrived broken may never order again. If a worked 25% of them leave and each would have brought ₹600.00 of future margin, every broken order also costs ₹150.00. That line is a judgement, not a measurement; put it in only if your repeat-purchase data supports it, and test the answer with and without it.

At 1%, 2% and 3%: the sensitivity grid

Your 3% may come from a short trial and could really be 1 or 2 points, and your C depends on the claims route. So run the formula across both. Each cell is ₹20.00 − rise × C.

Read across the bottom row first. At a 3-point rise, every route except the replacement route leaves you with little or nothing, and the worst case loses ₹3.97 a unit. At a 1-point rise every route still saves money. In other words: the cheaper pack is a good change if it raises breakage by a point, a doubtful one at two, and at three it depends entirely on how your broken orders are settled. The chart above draws the same thing: each line crosses zero at its break-even C.

Per unit, per month, per year

At a worked 1,000 units a month (12,000 a year), a 3-point rise means 360 extra broken orders a year.

The replacement route keeps ₹86,672.40 of the ₹2,40,000.00 gross saving; the rest goes on 360 replacements. With refunds and penalties it keeps only ₹6,360.00, which is barely worth the risk to reviews. The worst case loses ₹47,640.00. A one-off trial of 30 parcels in the cheaper pack, at a worked ₹75.00 courier each, costs ₹3,090.00; on the replacement route (₹7,222.70 a month) that pays back in about 13 days, so the trial is cheap relative to the decision it settles.

A refinement for high rates. Replacements travel in the same pack and can break too. If breakage goes from 2 to 5 per 100 and replacements break at the same rate, replacements per 100 orders go from 2.04 to 5.26, an extra 3.22 rather than 3. The extra cost becomes ₹13.72 a unit, the net ₹6.28, and the break-even C falls to ₹620.67. At low rates the simple formula is close enough; at 5% and above, use the refinement.

The option between the two packs

The choice is rarely all or nothing. Most of a ₹20.00 packaging saving comes from the box and fill; most of the breakage comes from a missing pad or a jar that can move. So price a middle pack: the cheaper pack with one tissue sheet (₹6.50) crumpled into a bottom and top pad. It costs ₹34.50, saving ₹13.50 instead of ₹20.00. If your trial shows it holds the rise to 1 point (an illustration, not a prediction), it nets ₹9.24 a unit on the replacement route, more than the full cut's ₹7.22 at 3 points, with fewer unhappy buyers.

The same reasoning prices any add-back: put its per-unit cost into S and the trial's rise into Δb. The Air Bubble Roll could be a candidate for heavier jars, but its captured page states no roll length, so ask on WhatsApp before you can put it into the formula. How to build a pack that cuts cost without adding breakage in the first place (fit, pads, fill, drop tests) is the packaging-without-breakage post.

How sure is the 3%?

If your 3-point rise came from a trial of 100 parcels, it rests on a handful of broken jars, and chance alone can move it by several points either way. Before you act on a borderline result, either extend the trial or run both packs side by side over the same weeks; the statistics of test parcels (why zero breaks in 20 parcels proves little) are in the packaging-without-breakage post. A sensible practice is to decide on the grid's worst plausible cell, not its best.

A rough sense of scale helps plan the trial. At a worked 2 breaks per 100 today, 100 parcels in each pack would be expected to show about 2 breaks in the old pack and 5 in the new one if the 3-point rise is real; counts that small overlap easily by chance. At 300 parcels each, the expected counts are 6 and 15, a gap you can see. If you cannot ship 300 of each quickly, run the new pack on one sales channel or alternate days for a month and compare like with like. This is practical judgement, not a formal statistical test; the point is that a 3-point rise measured on a few dozen parcels is a hint, not a fact.

Three things commonly distort a breakage comparison. Season: festive peaks and summer heat change handling and wax softness, so compare packs in the same weeks. Mix: if one pack carried more multi-jar orders, it will look worse. Jar: a jar change at the same time as a pack change makes the trial meaningless; a lighter jar's breakage is a separate question, the same-price lighter-jar post. If incoming glass is arriving chipped, measure that separately as well; comparing jar suppliers by real cost per usable jar is the cheaper-jar-supplier-with-more-breakage post.

Packaging is outside the burn formula, so this change needs no burn retest: wax, oil, wick and jar stay as approved (which changes do count as a revised formulation is the which-changes-count post). What it needs is the trial and the log. Courier weight is a separate gain or loss: a lighter pack can drop a parcel into a lower slab, which is the heavy-jar courier post; and whether a premium box should take its share at all is the ₹45 box post.

What to buy, and what to skip

Four CSI items for building and pricing the cheaper and middle packs, chosen on page statements, pack prices and stock on 4 October 2026. CSI publishes no crush, burst or drop data; nothing is ranked on protection performance.

Skip for this decision: 100-packs bought for a discount that is not there, and any protective add-on you cannot yet price per unit, because the formula cannot judge what it cannot cost.

Frequently asked questions