Should a discount say '50% off' or 'save 50%' or 'discount'?

'Off' dominates: it appears in 325 discount CTAs across 60 companies, versus 111 CTAs (34 companies) for 'save' and just 41 CTAs (10 companies) for the literal word 'discount'.[1] Buyers overwhelmingly see percentage-off framing, not the word 'discount.' If you're picking copy, lead with 'X% off' — it's the market's default vocabulary.

'Off' appears in 325 discount CTAs (60 companies) versus only 41 for the literal word 'discount' — Lazyweb Research, July 2026.

Lazyweb Research · n=4406 · Published 2026-07-07

paywallpricingmonetizationux-patternsmobileexperiments

The finding: 'off' is the default vocabulary

Splitting discount CTAs by which keyword they use shows a clear hierarchy.[1]

FramingCTAsCompanies
'…off' (e.g. '50% off')32560
'save' (e.g. 'save 50%')11134
literal 'discount'4110

'Off' framing is used by 60 of the 80 discounting companies — three times as many as use the literal word 'discount' (10 companies).[1] The word 'discount' itself is rare enough that headlining on it would put you outside the norm.

How to apply it

Default your primary CTA and headline to 'X% off' — it's what buyers are trained on. Use 'save' framing as a secondary or when a detected experiment swap makes sense: Snapchat's detected experiment shifted a price-focused headline to 'Save 60% on Snapchat+' with a 'SPECIAL OFFER' badge, and Drops' detected diff swapped a 'GET 50% OFF' button to 'SAVE 50%'.[2][3] Both moves keep the discount dominant while varying the verb. Avoid leading with the bare word 'discount' — only 10 companies do, so it reads as off-pattern rather than distinctive.[1]

Caveats

These are detected before/after UI diffs with inferred rationale, not measured A/B lift — the Snapchat and Drops swaps tell you what teams tried, not which won.[2][3] Keyword buckets can overlap (a CTA can contain both 'save' and 'off'), so the numerator counts are not mutually exclusive. Detection is text-only; a strikethrough price with no discount word won't be captured under any of these buckets.

The numbers

StatComputed from
'off' in 325 CTAs (60 companies); 'save' in 111 CTAs (34 companies); literal 'discount' in 41 CTAs (10 companies)keyword_split_off_discount_save
Snapchat detected experiment shifted to 'Save 60% on Snapchat+' with a 'SPECIAL OFFER' badgenamed_discount_example_snapchat
Drops detected diff swapped 'GET 50% OFF' button copy to 'SAVE 50%'experiment_learning_drops
Methodology. Universe: 4,406 paywall CTAs across 252 tracked mobile apps, July 2026. Method: keyword occurrence split over CTA + surrounding copy, deduped by company; experiment examples from 795 detected before/after diffs. Caveat: buckets overlap and experiment framing is inferred, not measured.

Sources & citations

  1. [1] Lazyweb Research analysis of 4,406 paywall CTAs (252 tracked paywall companies), July 2026. Keyword occurrence split ('off','save','discount') over combined CTA + surrounding copy, with distinct-company counts.
  2. [2] Lazyweb Research analysis of 795 detected paywall-CTA experiments (252 tracked paywall companies), July 2026. Named detected experiment: Snapchat 'Save 60% on Snapchat+' with 'SPECIAL OFFER' badge; inferred rationale, not measured lift.
  3. [3] Lazyweb Research analysis of 795 detected paywall-CTA experiments (252 tracked paywall companies), July 2026. Named detected diff: Drops 'GET 50% OFF' -> 'SAVE 50%'; inferred rationale, not measured lift.

Source: Lazyweb Research — proprietary analysis of real, in-market app screens. Cite as Lazyweb Research, 2026-07-07.

Related questions

Explore the underlying screens, flows, and A/B tests inside Lazyweb. More research