Comparison
Comparison section examples
The open-tabs section. An honest grid with conceded rows: a table you win 10–0 reads as rigged.
261 comparison sections browsable in the library, each cropped from a real page at its measured position.
How we counted
Source: 261 comparison sections in this category, counted at build.
No sections match these filters.
Comparison insights
A count, not a recommendation. Every figure carries its denominator: these count comparison sections measured on crawled pages, and the 261 listed above are the ones that crop to a showable image.
- 13.2%of the page tall, typicallyacross 262 comparison sections measured
- 129words in the median comparison sectionacross 261 comparison sections measured
- 5%carry no-credit-card note2 of the 43 SaaS comparison sections measured
Comparison questions, answered from the corpus
How many pages have a comparison section?
305 of the 4,069 fully measured pages carry at least one comparison section, which is 7%. A page counts once however many it holds.
Where does a comparison section sit on the page?
The median comparison section starts 37% of the way down the page (median of 262 comparison sections measured), counted from the top edge of the page.
How big is a typical comparison section?
The typical comparison section takes 13% of its page’s height (median of 262 measured) and holds 129 words (median of 261 measured).
Which tactics show up most in comparison sections?
Among SaaS comparison sections, no-credit-card note appears in 2 of the 43 measured (5%); rating appears in 2 of the 56 measured (4%). A tactic counts only when its text is on the band.
Does a comparison section open with an h1?
1% of the 253 comparison sections whose heading level we could measure open with an h1.
This gallery shows 261 real comparison sections, cropped from 236 different sites, each cut from a live page’s full-page screenshot at the band’s measured position. 305 of the 4,069 fully measured pages carry at least one comparison section, which is 7%. Among the comparison sections measured, the typical comparison section starts 37% of the way down its page (median of 262 measured), takes 13% of its page’s height (median of 262 measured) and holds 129 words (median of 261 measured). 1% of the 253 comparison sections whose heading level was measured open with an h1. Classification is deterministic, from the band’s own DOM signals, with no AI judgment. Every card opens the full measured band, and the broader measured range is on the matching research page. The corpus is seeded from launch directories, so read these counts as what UXXRAY measured, not as a sample of the whole web.
How this is measured
Each card is the actual comparison band, cropped from a crawled page's full-page screenshot at its measured position — the highest-scoring funnels first.
Comparison insights
- Typically 13.2% of the page tall across 262 sections
- Usually starts 37% of the way down across 262 sections
- Carries a median of 129 words across 261 sections
- Holds a median of 4 links across 96 sections
- Most often on a nav → hero → comparison → faq → footer page 63 of 262 pages
- comparison comes before faq 95% of the time 176 of 186 pages carrying both
- No-credit-card note — 5% 2 of the 43 comparison sections we measured in SaaS funnels
Conversion tactics measured in comparison sections
Share of comparison sections we measured in SaaS funnels
A count, not a recommendation
Every rate is a lower bound: where the extractor found a tactic elsewhere on the same capture, that section leaves the denominator rather than counting as a miss, because only one position per tactic per page is recorded.
Page shapes, corpus-wide
34 page shapes cover half of the 3,405 pages we've measured. The most common is nav → hero → content → footer: 385 pages (11%). Measured across the funnels we've walked: SaaS 84% · E-commerce 11% · Affiliate 4% · Lead gen 1%.
How this is counted
The top 10 shapes cover 34% of pages. Its content element is the classifier’s terminal fallback (a band nothing else matched), not a named section. With that unnamed fallback removed from every sequence, 486 distinct shapes remain (down from 1017 raw) and the top 10 cover 51%. Raw distinct counts move with detector coverage, not with design, so concentration is the number worth citing.