'Continue' or 'Start Free Trial': which paywall CTA verb do apps actually use?
- By Ali Abouelatta
- n = 1,886
- Published
- Updated
- 2 min read
- paywall · trials · monetization · ux-patterns · design
Short answer
'Continue' is the single most common primary CTA at 21.2% (399 of 1,886), ahead of 'Start' verbs at 16.9% (318) and 'Subscribe' at 11.7% (220).source[1] The most common exact button text is bare 'Continue' — 350 rows across 66 distinct companies.source[2] Neutral verbs beat explicit trial verbs on raw frequency, so 'Start Free Trial' is a deliberate choice, not the default.
'Continue' leads paywall primary CTAs at 21.2% (399/1,886), vs 16.9% for 'Start' verbs — Lazyweb Research, July 2026.
Bar chart · Rows
Rows — The finding
One block is about 12.5 Rows
- continue
- 350
- subscribe
- 89
- start free trial
- 84
- try for free
- 74
- subscribe now
- 54
The finding
Among 1,886 primary paywall CTAs, the verb families break down as: continue 399 (21.2%), start 318 (16.9%), subscribe 220 (11.7%), upgrade 55 (2.9%), and unlock 46 (2.4%).source[1] At the exact-text level, bare 'Continue' dominates with 350 rows across 66 companies — far more companies than any single trial-led phrase.source[2]
| Exact CTA text | Rows | Distinct companies |
|---|---|---|
| continue | 350 | 66 |
| subscribe | 89 | 21 |
| start free trial | 84 | 22 |
| try for free | 74 | 16 |
| subscribe now | 54 | 14 |
How to apply it
The choice is between a neutral verb ('Continue') that keeps the trial promise in the offer copy above the button, and an explicit trial verb ('Start Free Trial', 'Try for Free') that puts the promise on the tap itself. 'Continue' is used by more distinct companies (66) than 'Start Free Trial' (22) and 'Try for Free' (16) combined, so if you are picking a safe default, neutral wins on prevalence.source[2]
But prevalence is not conversion. 'Start Free Trial' and 'Try for Free' together represent 158 rows from 38 companies — a large, deliberate cohort that has chosen to foreground the trial. Test the neutral-vs-explicit split against your own funnel rather than copying the mode.
Caveats
Verb families are counted with ILIKE substring matches, so a button can match one family only. Exact-text counts are normalized with lower(trim()).source[1]source[2] All role-based figures use the 1,886 cta_role='primary' rows; the 39% of rows with unknown role are excluded.source[1] Note that some high-row phrases are concentrated in one company — always check the distinct-company count before treating a phrase as an industry pattern.
Real in-market screens
Screens from real, in-market apps, tracked by Lazyweb Research.
Ai Arta1290 × 2796opens the full image 
Apple Fitness1179 × 2556opens the full image 
Gentler1179 × 2556opens the full image 
Lyft1179 × 2556opens the full image
The numbers
| # | Stat | Computed from |
|---|---|---|
| [1] | continue 21.2% (399), start 16.9% (318), subscribe 11.7% (220), upgrade 2.9% (55), unlock 2.4% (46) | cta_verb_families over 1,886 primary CTAs |
| [2] | 'continue' 350 rows/66 companies; 'subscribe' 89/21; 'start free trial' 84/22; 'try for free' 74/16; 'subscribe now' 54/14 | top_primary_cta_texts |
Methodology
Universe: 1,886 primary paywall CTAs across ~800 tracked apps, pulled July 2026. Verb families and exact texts counted with ILIKE and normalized lower(trim()) matches; unknown-role rows (39%) excluded.
Sources
- [1]
Lazyweb Research analysis of 1,886 primary paywall CTAs (~800 tracked apps), July 2026
Verb families via ILIKE substring match on cta_text; role-based cut excludes 39% unknown-role rows.
- [2]
Lazyweb Research analysis of 1,886 primary paywall CTAs (~800 tracked apps), July 2026
Exact-text counts on lower(trim(cta_text)) with distinct-company dedupe.
Source: Lazyweb Research — proprietary analysis of real, in-market app screens. Cite as Lazyweb Research, 2026-07-07.