Before you blame the supplier

Your complaint is real: the same product that held a room in December is weak and half-empty by the third week of May. What is not established is which material did that. A reed diffuser's rate of liquid loss is set by the exposed reed surface, the air moving past it and the air's temperature; the oil contributes, but so does the base, the reed count, the fill, the bottle's neck and whether the customer put it on a windowsill. CSI's own fibre reed pages say two useful things here in their own words — "avoid placing in direct sunlight or near heat sources which accelerate evaporation" and "use 10+ for maximum throw in larger rooms — the base will evaporate faster at higher reed counts" — and both of those point away from the oil.

So run the free comparisons before you run the expensive one. Each row below is a single-variable test you can do with material you already have, and each one is capable of making your supplier switch unnecessary.

There is one more reason to be sceptical of the supplier hypothesis. "Fades" conflates two different complaints that need different fixes. A bottle that empties fast has lost liquid, and that is a weighable fact. A bottle that still has liquid but smells of less has not lost liquid, and weighing it will tell you nothing. Decide which one your customers are reporting before you shortlist anything, because the first is an evaporation problem and the second may be a load, reed or oil-selection problem. The 60-days-to-25-days post works through the first; this page assumes you have already separated them.

The five gates, and what each can conclude

A gate is worth running only if it can eliminate a candidate. That sounds obvious and it is routinely ignored: most "testing" in this category produces observations nobody acts on, which is the same as not testing. So write down, before you spend, what each gate is allowed to decide — and, just as importantly, what it is not.

Gate 1: the screen that costs nothing

Set the run first. Five hundred bottles at 120 ml and the published 25% ratio is 500 × 30 g = 15,000 g of oil and 500 × 90 ml = 45,000 ml of base. Now take five plausible replacements from CSI's fresh and green shelf and add up every rung's live quantity. The arithmetic below is October 2026 stock, and it eliminates a candidate before anything is ordered.

Read the Blue Ocean row carefully, because it is the shape of a Gate 1 failure and it is not about quality. The oil is listed, in stock at every rung, and a CSI customer calls it "fresh and soothing". It simply cannot fund one 500-bottle run from the quantities currently on the shelf, and the attempt would cost 2.12% more per gram than its own best rate while leaving you nothing in reserve. That is a supply disqualification, it took ten minutes of reading, and it cost nothing — which is exactly the point of putting this gate first. For a smaller run it may be the right answer; the verdict is a function of your order, not of the oil.

What a disqualification costs at each gate

Here is the programme priced gate by gate on the four candidates that survive the paper screen — Citrus Lemon, Zesty Lemon, Eucalyptus Spearmint and Verbena — with the base, the bottle, the reed count and the load held constant throughout. The candidate counts are an illustration of a five-into-one funnel; the money is live.

Three things in that table are worth arguing with your own numbers about. First, Gate 2 is where the instruments get bought, so its ₹3,079.80 flatters later gates — if you already own a scale and a thermometer the gate is ₹2,371.80 and the staircase gets steeper, not shallower. Second, the ₹2,230.77 average is arithmetically correct and operationally misleading: budgeting four eliminations at ₹2,230.77 each prices a paper screen at two thousand rupees and a weighed trial at two thousand rupees, and both numbers are wrong. Third, the reed line is the single clearest argument for one-variable testing. Sixty-four reeds cost ₹636 at Gate 4; if you decide halfway through to change the reed count as well, you buy a second set, and on a 120 ml bottle a second set takes hardware from ₹170.20 to ₹234.20 — a permanent 37.6% increase against a one-off reformulation test.

What to buy to run the funnel

Four lines carry the whole programme. One oil that is cheap enough to be the baseline, one that rewards volume hard enough to change your cost structure, the base you hold constant, and the vessel that makes a retained reference possible at all. Every price below is live and every stock figure is per variant.

Approving a replacement, and what you may claim

Say Gate 4 leaves you one candidate. Before it becomes your production oil, two more things need writing down: what it costs you per bottle at every point of the published window, and what you are permitted to say about it. The first is arithmetic. The second is a discipline, and it is the part that creates refund conversations when it is skipped.

Now the honesty that makes the switch defensible. A slower-evaporating base is not automatically a better base. CSI's Depth page is the only one claiming to "slow evaporation" and Longevity the only one claiming "extended cold diffusion"; both are the pages' own words and neither is a measurement, so they cannot rank two bases numerically. More to the point, if a base restricts wicking or release then the bottle lasts longer and smells of less — which is a worse product with a better-looking day count. Every base swap in this funnel therefore gets judged on two readings taken from the same bottles: grams lost per day, and a coded intensity note against the cool-stored control. If consumption falls and throw falls with it, you have bought days by selling scent. Note too that CSI's own Depth page lists "Fresh citrus, light florals, aquatics" under Avoid and routes them to Lift — so for a lemon formula, the slow-evaporation base is the one the page steers away from.

Frequently asked questions