What affects how much e-liquid you use

Summary

From 1 October 2026, UK Vaping Products Duty is charged at £0.22 per millilitre of e-liquid. The rate is flat: it does not vary with nicotine strength, with PG:VG ratio, or with the device the e-liquid is used in. A 10ml bottle carries £2.20 of duty whether it contains 6mg or 18mg of nicotine.

That makes one question unusually important: what actually determines how much 3-liquid a vaper gets through?

JAC Vapour examined this using order records from its own UK customers across two matched six-month periods in 2025 and 2026. Four findings:

1. Lower nicotine strength means more e-liquid, but less nicotine overall. JAC Vapour's analysis of 1,366 UK refill customers found that those buying only 6mg premixed e-liquid used 35 to 44% more volume than those buying only 12mg, while taking in 28 to 33% less nicotine per day.
2. Vapers only partly compensate. Full compensation would mean that halving the strength doubled the volume. It rose by about a third. A separate check on 45 customers who changed their own strength found the same pattern in the same people: nicotine intake fell by roughly 44% when strength was reduced.
3. PG:VG ratio has no independent effect on volume. A raw comparison suggested high-VG customers used more e-liquid. Once nicotine strength was accounted for, the difference disappeared, and high-VG customers used slightly less than their strength alone predicted.
4. Coil resistance has no independent effect either. Low-resistance users do get through more e-liquid, but they also vape roughly half the nicotine strength. Adjusted for strength, no effect by resistance remains.

Taken together: nicotine strength is the only product attribute that reliably changes how much e-liquid you use — and because duty is charged on e-liquid, it is the only one that changes what you pay. The two attributes the industry conventionally points to turn out to be standing in for strength.

How was this study carried out?

Source. Order line-items from JAC Vapour's own UK customers: 12,554 orders in the 2026 window and 13,352 in the 2025 window. Volume was computed per customer, then aggregated. All figures below are medians with interquartile ranges. Consumption is heavily right-skewed, so an average would describe a heavy minority rather than a typical customer.

Windows. 1 January 2026 to 30 June 2026, and 1 January 2025 to 30 June 2025 (181 days each). Both end before the duty change, so neither is affected by pre-duty stockpiling. The two windows cover the same calendar months, so seasonality is matched rather than confounded.

Cohorts are exclusive. A customer counted in the 12mg group bought 12mg premixed e-liquid and no other strength during the window. Customers buying across categories were excluded rather than assigned.

Exclusions. Customers with only one order in the window, since no reorder interval is observable; anyone buying DIY components (nicotine shots, flavour shots, base, mixing bottles), because mixers add their own base at home and purchased volume then understates consumption; cancelled orders and refunded items; samples and non-product lines. For the coil comparison, customers also buying pods.

Metrics. Volume per customer per day, and nicotine per customer per day, calculated as strength multiplied by volume.

What full compensation means. Nicotine intake is strength multiplied by volume. So a vaper holding their intake exactly constant would necessarily double their volume when halving their strength. That is the full compensation case, and it is the arithmetic benchmark every figure below is measured against. It is not a prediction anyone has made; it is the limiting case against which partial compensation can be quantified.

Samples. Strength: 873 customers (2026), 904 (2025). Ratio: 424 and 498. Coil: 464 and 544.

Within-customer check. A separate analysis covered a continuous eighteen months (1 January 2025 to 30 June 2026, 39,145 orders) to identify customers who made a sustained change of nicotine strength, defined as at least two consecutive orders predominantly at one strength, followed by at least two predominantly at another, with at least 30 days of orders on each side. 45 customers qualified. Rates exclude each period's final order, which is bought at the end of the period and not consumed within it. Results are reported under Finding 1.

Finding 1: Does lower nicotine strength mean vapers use more e-liquid?

Yes, but not proportionally — vapers still end up consuming less nicotine. JAC Vapour's analysis of 1,366 UK refill customers across two matched six-month periods in 2025 and 2026 found that customers buying only 6mg premixed e-liquid used 35 to 44% more volume than those buying only 12mg, while taking in 28 to 33% less nicotine per day.

Volume does rise as strength falls. It just rises nowhere near far enough to hold nicotine intake constant.

2026 window

2025 window

The 18mg median is identical across both years, and 6mg and 12mg are within a few per cent of each other. Set against the full compensation benchmark:

The clearest view is nicotine per day, which under full compensation would be flat across all three strengths. Indexed to 18mg = 100, the 2026 window gives 18mg 100, 12mg 87 and 6mg 58; the 2025 window gives 18mg 100, 12mg 83 and 6mg 60.

Customers at lower strengths end up taking in substantially less nicotine overall. They partially adjust their volume, then settle.

Fitted across the three strengths, volume rises approximately as the square root of the nicotine reduction: an exponent of −0.50 in 2026 (95% CI −0.57 to −0.34) and −0.53 in 2025 (CI −0.63 to −0.40). Full compensation would require −1.0. The two windows are statistically indistinguishable from one another.

The interquartile ranges are wide and overlap between strengths. Individual variation is large; these are typical values, not narrow ones.

Does it hold within the same person?

Yes. Among 45 JAC Vapour customers who changed their own nicotine strength, the 19 who halved theirs from 12mg to 6mg went on to use 24% more e-liquid while taking in 40% less nicotine per day. The 11 who increased their strength moved the other way on both measures.

That matters because the comparison above is between different groups of people, which leaves an obvious objection — perhaps customers who use 6mg are simply lighter nicotine users by disposition, rather than people who reduced. To test it, JAC Vapour identified customers who made a sustained change of strength during an eighteen-month span and measured each one's consumption before and after their own change. 34 reduced their strength; 11 increased it.

Nicotine intake. Under full compensation the nicotine ratio would be 1.00, because volume would rise or fall to hold intake constant. It did not:

Across all 34 reductions, nicotine intake fell by a median of 44%. This is the part of the finding that no longer depends on comparing different people.

Volume. It moved as predicted too, though less steeply. The 12mg to 6mg change was the largest group and the most common change in the data:

The two reduction figures are far short of full compensation and close to one another. And the 11 customers who increased their strength moved the other way, as the mechanism requires: 8 of the 11 used less e-liquid afterwards. That symmetry rules out an alternative reading — if only reductions showed an effect, it would suggest something else about those customers was changing at the same time. Both directions moving as predicted is harder to explain that way.

How to read these figures. The volume response within individuals is shallower than between them, a fitted exponent of −0.18 against −0.50. The most likely explanation is composition: the between-person figure includes the effect of who chooses which strength, and the within-person figure strips that out. 24 of the 34 reductions produced an increase in volume, which is a clear majority but not unanimity. And across all reductions pooled the volume increase was smaller, at +14%, because that figure averages changes of different sizes — an 18mg to 12mg move is a much smaller nicotine reduction than a halving, and responds accordingly, so the halvings are the meaningful comparison.

The increase arm is small for a structural reason rather than by chance: 18mg is the strongest e-liquid in the refill range, so customers already there cannot step up.

Finding 2: Does high-PG e-liquid last longer than high-VG?

No. JAC Vapour's analysis of 424 customers buying from a single product range found no independent effect of PG:VG ratio on premixed e-liquid consumption once nicotine strength was accounted for — high-VG buyers did use about 9% more e-liquid, but they were also vaping a weaker strength on average, a volume-weighted mean of 6.4mg against 9.9mg.

High-VG e-liquid produces more vapour per puff, so it is commonly said to be consumed faster, and a high-PG e-liquid to last longer. The raw comparison appears to confirm it. Both groups below come from the same product range, with the same flavours at the same price, differing only in ratio.

Averaged across customers, high-VG buyers used 3.92 ml/day against 3.61 for high-PG, about 9% more. But the two groups are not comparable on strength: the volume-weighted mean strength was 9.9mg for high-PG and 6.4mg for high-VG.

Applying the strength relationship from Finding 1, with no ratio term at all, a group that much weaker should use 1.24 times the volume. The observed difference was 1.09 times, less than strength alone accounts for. At the median, the two groups are indistinguishable.

So once strength is controlled for, no independent effect of PG:VG ratio on volume remains. If anything, high-VG customers use marginally less e-liquid than their strength predicts.

Finding 3: Does coil resistance change how much e-liquid you use?

No. JAC Vapour's analysis of 464 customers found no independent effect of coil resistance on premixed e-liquid consumption once nicotine strength was accounted for — 0.5Ω users did get through more e-liquid than 1.2Ω users, but they were also vaping around half the nicotine strength.

Lower-resistance coils draw more power and vapourise more e-liquid per puff, so resistance should predict consumption. Superficially it does:

Low-resistance users do get through more e-liquid. But look at the final column: they also vape barely half the nicotine strength. The resistance groups are, to a large extent, strength groups.

Adjusting each cohort for its own strength using the relationship from Finding 1, and expressing the result as observed volume divided by strength-predicted volume:

Every cohort lands close to what strength alone predicts, with no gradient by resistance in either window. The lowest-resistance group consumes slightly less than predicted, not more.

An independent purchase signal agrees. Coil consumption ran at a median of one coil every 12 to 18 days across all four groups. A cohort genuinely vapourising substantially more e-liquid would burn through coils faster. None did.

Why don't PG:VG and coil resistance affect consumption?

Findings 2 and 3 appear to contradict established industry understanding. We think they refine it rather than overturn it, and the mechanism is the same in both cases.

Neither ratio nor resistance acts on the e-liquid directly. High-VG e-liquid is normally paired with a sub-ohm, high-power, direct-lung device; low-resistance coils are normally fitted to high-wattage mods. Both predict consumption because of the hardware they usually accompany, and, as this study shows, because of the nicotine strength that tends to go with them.

JAC's customer base is overwhelmingly low-power mouth-to-lung: simple refillable devices with fixed or near-fixed output. A customer running a 0.5Ω coil here is not running a high-power device; they are running the same modest device as everyone else with a different coil fitted. Remove the variation in device power, control for strength, and the two proxies carry no information of their own.

The defensible statement is therefore not that ratio and resistance never matter. It is that they predict consumption through device power and nicotine strength rather than in their own right, and in a low-power refill setup, what remains is strength.

How much will vape duty cost per year?

Duty is charged per millilitre of e-liquid, not per milligram of nicotine. That has a direct consequence: a 6mg vaper pays three times as much duty per milligram of nicotine as an 18mg vaper.

That much follows from the flat rate alone. What this study adds is that the absolute burden runs the same way, because vapers do not fully compensate. Comparing the two ends of the refill range, a typical 6mg customer takes in 42% less nicotine per day than a typical 18mg customer, and pays 75% more duty.

Figures are duty only, at £0.22/ml. Including VAT charged on the duty, the amounts a customer actually pays are approximately £373, £277 and £213 a year. The ranges are wide, and deliberately shown: individual consumption varies enormously, and the median is a typical case rather than a prediction for any particular person.

What does reducing your nicotine strength cost under the duty?

Less than proportional scaling would suggest, but not nothing. If volume rose in proportion to the strength reduction, halving from 12mg to 6mg would add £230 a year in duty. The observed figure is £80.

Two things that do not affect your duty bill, on this evidence: which PG:VG ratio you choose, and which coil resistance you fit. Both are worth choosing on flavour, throat feel and personal preference, not on cost.

Limitations

We publish these because they define what the findings can and cannot support.

1. These are purchases, not consumption. E-Liquid held at home, gifted, spilled or discarded is invisible. Purchase volume is a proxy.
2. Individual variation is large. The interquartile ranges overlap substantially between strengths. The medians describe typical customers; they do not predict any individual.
3. Rates assume activity across the whole window. Volume is divided by 181 days. The median customer's first-to-last-order span was 127 to 130 days, and the lowest quartile around 87 days, so rates are understated for customers who joined or lapsed mid-window. A sensitivity using observed active span raises all levels but changes no ordering or ratio.
4. DIY volume is unmeasurable from purchase data, because mixers add their own base. DIY customers were excluded rather than estimated.
5. Single-brand sample. JAC's customers are legacy, refill-oriented and predominantly low-power mouth-to-lung. Findings 2 and 3 in particular may not generalise to sub-ohm direct-lung or disposable users.
6. Coil resistance is what was purchased, not necessarily what was fitted.
7. Strength, ratio and resistance are inferred from purchase mix, not declared by the customer.
8. Observational, not experimental. The main cohort comparisons are between different groups of people. The within-customer check partly addresses this, but those customers chose to change their own strength, so the result still cannot support a prediction for any individual. Nobody was assigned a strength.
   
   The within-customer sample is also small: 19 customers in the largest cell, and 11 in the arm that increased their strength, where the volume change is not statistically significant although the nicotine change is. It corroborates the main finding; it does not carry it. The constraint is structural rather than a matter of collecting more data. Of 4,644 customers buying premixed e-liquid, only 1,204 had the four orders a before-and-after comparison requires, and 50 of those showed a clean sustained change.
9. 3mg was excluded. The cohort was small (41 and 61 customers) and unstable, sitting above 6mg in one window and below it in the other. The within-customer check pointed the same way: the five customers who moved from 6mg to 3mg showed a slight fall in volume. Something at the bottom of the strength range does not follow the pattern seen above it. We report no figure for 3mg.
10. 0mg was excluded as too few customers buy it exclusively.
11. A small upward bias affects 2025. Migrated refund records carry no product detail, so 83 partially refunded orders could not be netted off. Fully refunded orders were removed.
12. One customer moving from 3mg to 0mg was excluded from the within-customer nicotine figures. A move to zero nicotine is a different behaviour, not a step on the strength ladder, and it cannot be expressed as a ratio.
13. Sample sizes are stated with every figure.