Are B2B homepages all starting to sound the same?

B2B software homepages have been drifting toward the same message for a decade. We read the headline and subheadline of 42,045 homepage snapshots from 7,633 B2B software companies, one per company per year from 2014 to 2026. 57.4% of 2024-2026 homepages sit closer to their industry's typical copy than the median homepage of 2014-2016 did. If nothing had changed, that figure would be 50%.

The word that spread fastest is "platform": it is in 8.62% of hero copy in 2026, against 2.02% in 2014. And companies whose copy stands out did not grow faster than companies whose copy blends in.

How we measured sounding the same

The companies are 10,000 B2B software firms drawn at random from the company records Kenso uses for sales and outreach, headquartered in the US, the UK, Canada, Australia, Ireland or New Zealand. For each, we took one homepage snapshot a year from Kenso's archive of company websites and kept the hero copy: the main headline and the line under it. We dropped parked pages, pages not in English, template text ("coming soon", site-builder sample copy) and copy shared word for word by three or more companies in the same year. Each company's own name was replaced with a placeholder, so names do not make companies look different.

We turned each hero into a list of numbers that captures its meaning (a text embedding). Two heroes that say similar things in different words get similar numbers. For every year, within each of three industry groups, we averaged the similarity of every pair of companies.

The average rose from 0.541 in 2014-2016 to 0.553 in 2024-2026, on a scale where 1 means identical meaning: a rise of 0.012 (95% interval 0.009 to 0.015). It went up in 10 of 12 year-on-year steps. This is a slow drift, not a collapse into one voice. The level itself means little, because the embedding rates any two texts as fairly similar. Only the change counts.

The same companies are drifting

Kenso's web archive holds more sites in recent years, so the mix of companies changes. We checked that three ways.

• Same companies. Among 310 companies captured in every year of both windows, the rise is the same size: 0.013 (95% interval 0.006 to 0.020).
• Each company against itself. Following 6,722 companies over time, each one's copy moves closer to its industry's average, by 0.00055 a year.
• Page format. Later homepages more often have a subheadline. The headline alone gives a rise of 0.015, and true H1 headlines alone give 0.012.

We had predicted a steeper rise from 2023, when AI wording spread. It did not appear: about 0.0014 a year from 2014 to 2022, and 0.0010 a year from 2022 to 2026. By group, the rise was 0.014 for software development, 0.013 for security, data and other software, and 0.004 for technology, information and internet, a broad label that also holds consumer apps.

Same claims, not the same words

The convergence is in meaning, not vocabulary. Compared word for word, similarity barely moved, and the interval includes zero. The 50 most common word pairs made up 5.18% of copy in 2014 and 4.92% in 2026. Companies are not copying each other's sentences. They are making the same kind of claim.

• "Platform" rose steadily, from 2.02% of hero copy in 2014 to 8.62% in 2026.
• "AI-powered", with AI-driven, AI-native and AI-first, was at 0.35% in 2022 and 2.41% in 2026.
• "All-in-one" rose from 0.22% to 1.19%.
• "Seamless" rose from 0.05% to 0.84%. We expected it to stay flat.

Standing out did not pay, or cost

Do companies that write differently grow faster? For each year from 2018 we took the quarter of companies whose copy was furthest from their industry's average, 2,757 company-years in all. We matched each with up to three typical companies of the same industry group, founding period, team size and prior growth, and followed headcount, rebuilt from work histories, for two years.

After 8 quarters, the median distinctive company had grown 0.0% and its matched typical companies 0.82%: a gap of -0.82 percentage points, with a 95% interval from -2.27 to 3.05. The two groups grew alike before the comparison, which is the check that the design works, and a regression across the full range of distinctiveness agrees: no measurable link.

We had predicted a small growth edge for distinctive copy. The data does not show one.

What this data cannot show

• Distinct is not good. Distance from the average measures difference in meaning, not quality. A sharp homepage and a confusing one can be equally far from the norm.
• Headcount is slow. Copy could still help pipeline or win rates in ways that do not show up in hiring two years later.
• The embedding's view. An embedding trained on recent web text may place recent marketing language closer together. Word overlap shows no rise, so the finding is about meaning as the embedding measures it.
• Coverage. The archive does not capture every site every year, the sample is companies that exist today, and industry labels do not separate B2B from B2C software.
• Association, not cause. The growth result is a matched comparison, not an experiment.

What it means for B2B teams

• Read your hero against your category. If your headline leans on "platform", "all-in-one" or "seamless", a growing share of your competitors says the same.
• Specific beats different. Standing apart carried no growth penalty and no measurable reward. Say what you do and for whom, rather than chasing novelty.
• The AI label is getting crowded. "AI-powered" and its variants went from 0.35% to 2.41% of hero copy in four years. On its own, it no longer sets you apart.
• Judge copy by outcomes. Distinctiveness alone did not predict growth, so test a new headline on demo requests or replies, not on how different it sounds.