Thirteen of the 20 B2B SaaS homepages we checked on 21 October 2026 name a product category buyers search for in their title tag. Only eight name one in the hero a visitor reads. The category word has moved from the page to the search snippet, and on most of these sites the headline is now a slogan about AI.
This is a fresh sample, separate from the 24 sites in our earlier 24 B2B SaaS Websites Scored on Clarity, Speed and Demo Path (2026). It keeps that scorecard's clarity and demo tests and adds search basics and lab Core Web Vitals, with thresholds from Nielsen Norman Group, web.dev and Google Search Central.
The rubric: 12 pass or fail tests
Each test passed scores one point, so each dimension scores 0 to 3 and the total is out of 12. Binary tests let anyone repeat them on their own site.
| Dimension | Test, one point each | Basis |
|---|---|---|
| Clarity | C: hero names a category buyers already use; coined labels fail | NN/g homepage guideline |
| Clarity | A: hero names who it is for: a role, team or company type | NN/g homepage guideline |
| Clarity | S: hero gives one concrete specific: named functions or a number | Our test |
| Search | T: title tag names a searchable category | Google title link guidance |
| Search | H: hero headline present in the server HTML, before JavaScript | Google JavaScript SEO basics; Vercel crawler data |
| Search | O: Organization structured data on the homepage | Google Organization markup guidance |
| Speed | Lab LCP 2.5 s or less | web.dev "good" LCP |
| Speed | Lab TBT 200 ms or less | web.dev TBT target |
| Speed | Lab CLS 0.1 or less | web.dev "good" CLS |
| Demo path | A demo, sales or trial form one click from the first screen | Our test |
| Demo path | Five fields or fewer on the first step | A round line, also our median |
| Demo path | Response time or call length stated on the form page | Our test |
Source: Nielsen Norman Group, 2002; web.dev, 2023 to 2025; Google Search Central, 2025 to 2026; Vercel, 2024
Clarity method. The hero is the H1 and the line under it, read in a 1,024 by 768 pixel window without scrolling. Navigation, logos, proof strips and the title tag do not count.
Search method. We fetched each homepage once with curl and read the raw HTML: title tag, H1s and JSON-LD. We did not measure rankings, which for someone else's domain are only modelled estimates. These tests cover what a crawler receives, the part a website team controls.
Speed method. We ran Lighthouse 12.8.2 locally with its desktop preset, three times per homepage between 09:49 and 10:35 UTC, and took each metric's median. Google's PageSpeed Insights API, which also returns field data, refused keyless requests with a 429 error that morning. A first pass with the mobile preset returned simulated LCPs of 5 to 48 seconds on a shared machine, so we discarded it.
Demo path method. We followed the most prominent demo or sales link on the first screen and counted visible fields on the first step. A phone number with a country picker counts as one; a consent box counts as none. We submitted nothing.
The scorecard: clarity and search
Letters show which tests a site passed: C, A and S for clarity; T, H and O for search.
| Site | Category | Clarity | Passed | Search | Passed |
|---|---|---|---|---|---|
| monday.com | Work management | 0 | none | 2 | HO |
| ClickUp | Work management | 1 | S | 2 | HO |
| Airtable | Work apps | 0 | none | 1 | H |
| Zapier | Workflow automation | 1 | C | 3 | THO |
| Calendly | Scheduling | 2 | CS | 3 | THO |
| Klaviyo | Marketing automation | 2 | CS | 3 | THO |
| Twilio | Communications APIs | 0 | none | 3 | THO |
| Brex | Spend management | 2 | CS | 3 | THO |
| Lattice | HR software | 2 | CS | 3 | THO |
| Outreach | Sales engagement | 2 | AS | 2 | HO |
| Apollo | Sales data and engagement | 1 | S | 2 | HO |
| Clay | Data enrichment | 0 | none | 2 | HO |
| Retool | Internal tools | 0 | none | 3 | THO |
| Snyk | Application security | 2 | CS | 3 | THO |
| 1Password | Access management | 2 | CS | 2 | TH |
| Pendo | Product analytics | 0 | none | 2 | HO |
| Okta | Identity | 0 | none | 3 | THO |
| Smartsheet | Work management | 1 | S | 3 | THO |
| Docusign | E-signature | 2 | CS | 2 | TO |
| Productboard | Product management | 2 | AS | 3 | THO |
Source: Octo Agency, browser read of each hero and one curl fetch of each homepage's HTML, 21 October 2026. Klaviyo and Outreach redirected us to klaviyo.com/uk and outreach.ai, which were scored
The scorecard: speed, demo path and totals
Speed columns are medians of three desktop Lighthouse runs. Demo path counts clicks from the first screen, first-step fields and whether timing is stated.
| Site | LCP (s) | TBT (ms) | CLS | Speed | Clicks | Fields | Timing | Demo path | Total |
|---|---|---|---|---|---|---|---|---|---|
| monday.com | 2.57 | 0 | 0.00 | 2 | 1 | 9 | No | 1 | 5 |
| ClickUp | 9.47 | 90 | 0.69 | 1 | 2 | 5 | No | 1 | 5 |
| Airtable | 5.29 | 0 | 0.00 | 2 | 1 | 9 | No | 1 | 4 |
| Zapier | 1.20 | 0 | 0.05 | 3 | 1 | 3 | No | 2 | 9 |
| Calendly | 2.78 | 180 | 0.00 | 2 | 1 | 5 | No | 2 | 9 |
| Klaviyo | 2.54 | 0 | 0.03 | 2 | 1 | 6 | No | 1 | 8 |
| Twilio | 8.75 | 0 | 0.25 | 1 | 1 | 7 | Yes, 1 day | 2 | 6 |
| Brex | 1.59 | 38 | 0.00 | 3 | 1 | 5 | No | 2 | 10 |
| Lattice | 2.67 | 0 | 0.00 | 2 | 1 | 3 | No | 2 | 9 |
| Outreach | 2.53 | 39 | 0.04 | 2 | 1 | 1 | No | 2 | 8 |
| Apollo | 2.06 | 0 | 0.00 | 3 | 1 | 1 | No | 2 | 8 |
| Clay | 4.18 | 0 | 0.00 | 2 | 1 | 4 | No | 2 | 6 |
| Retool | 1.82 | 0 | 0.00 | 3 | 1 | 2 | No | 2 | 8 |
| Snyk | 1.21 | 0 | 0.00 | 3 | 1 | 8 | No | 1 | 9 |
| 1Password | 3.47 | 48 | 0.00 | 2 | 1 | 9 | No | 1 | 7 |
| Pendo | 1.21 | 0 | 0.04 | 3 | 1 | 4 | No | 2 | 7 |
| Okta | 2.04 | 0 | 0.10 | 3 | 1 | 8 | No | 1 | 7 |
| Smartsheet | 6.66 | 97 | 0.04 | 2 | 1 | 1 | No | 2 | 8 |
| Docusign | 6.25 | 3 | 0.36 | 1 | 1 | 7 | No | 1 | 6 |
| Productboard | 1.26 | 0 | 0.00 | 3 | 1 | 4 | No | 2 | 10 |
Source: Octo Agency, Lighthouse 12.8.2 desktop preset, median of three runs, and a browser read of each form page, 21 October 2026; no form was submitted. Okta's CLS of 0.097 rounds to 0.10 and passes
Clarity: the category word moved to the title tag
Nielsen Norman Group's first homepage guideline asks for a tagline that "summarizes what the site or company does", and applies it to well-known companies too. Nobody here passed all three clarity tests, against five of 24 in our earlier scorecard.
The sites know their category, because they put it in the title tag for Google: "Meeting Scheduling Software" for Calendly, "Work Management Platform" for Smartsheet, "Agentic Product Management System" for Productboard. Five sites then leave it out of the headline. Twilio's title lists APIs for SMS, email and voice; its hero says "The platform for conversations in the AI era". Retool's title says internal software; its hero, on the day we read it, said "Secure your vibe-coded apps". Okta's hero was a takeover for a livestream.
AI replaced the category, not the product. Sixteen of 20 heroes mention AI, agents or "agentic". Calendly, Klaviyo, Snyk and 1Password put AI next to a plain category word and passed. Six of the seven zero scores used AI or agents where the category used to be.
Audience was the rarest pass, again. Only Outreach ("for Revenue Teams") and Productboard ("product makers") named a buyer. "Teams", "people and agents" and "builders" describe everyone; our B2B SaaS Website Playbook covers the fix.
Brex shows the shortest route. "Modern cards, banking, expenses, accounting, and more, in 120+ countries" names four products and a number in one line under a short headline.
Search basics: near universal, with two lessons
The technical basics separated almost nobody. Nineteen homepages send the hero headline in the server HTML, and 18 carry Organization structured data, which Google says helps it understand a company's details and tell it apart from others in results. Airtable and 1Password were the two without any JSON-LD in their homepage HTML.
Docusign's homepage HTML, as we received it, had no H1 until JavaScript ran. Google renders JavaScript with a current version of Chromium, so this matters less for Google than it once did. Google's own guidance still calls server-side or pre-rendering a good idea because not every bot runs JavaScript. Vercel's December 2024 analysis of its network found that none of the major AI crawlers, including OpenAI's and Anthropic's, rendered JavaScript; Gemini, through Google's infrastructure, did. A headline that only exists after a script runs is invisible to most of the crawlers feeding AI answers.
Productboard ships template placeholders. Its server HTML holds three H1s, two of them leftovers: "Primary title number two" and a personalisation tag that reads "See how [qualified_company default="your organization"] can ship what matters". A visitor sees the right headline; a crawler reads all three.
Title length. Google sets no character limit on titles and truncates them to fit the screen. Docusign's, at 74 characters, and Airtable's, at 71, are the longest here and the likeliest to be cut.
Speed: blocking time is solved, the largest element is not
Total Blocking Time separated nobody: all 20 medians were under web.dev's 200 millisecond target. That is partly the desktop preset, which does not slow the processor; a mid-range phone would block for longer.
Largest Contentful Paint is where the sample split. web.dev calls 2.5 seconds or less good. Eight homepages met it, led by Zapier at 1.20 seconds and Snyk and Pendo at 1.21. Five more, Outreach, Klaviyo, monday.com, Lattice and Calendly, landed between 2.53 and 2.78 seconds, close enough that a single hero image or font could decide it. The slowest were ClickUp at 9.47 seconds, Twilio at 8.75 and Smartsheet at 6.66.
Weight alone did not predict it. Median transfer was 9.2 MB per homepage, from 3.4 MB for 1Password to 21.3 MB for Clay. Productboard sent 14.9 MB and still painted in 1.26 seconds; Docusign sent 5.8 MB and took 6.25. What matters is whether the largest element in the first screen is waiting on scripts, fonts or a video behind everything else.
Layout shift failed on three sites. web.dev calls 0.1 or less good and over 0.25 poor. ClickUp scored 0.69, Docusign 0.36 and Twilio 0.246. Okta passed at 0.097, a whisker under the line.
Demo path: one click everywhere, then silence
Nineteen of 20 sites had a demo or sales form one click from the first screen. ClickUp was the exception: "Get a Demo" opens an on-demand video page, and the form sits behind a second "Book a demo" button.
First-step length ran from one field to nine. Apollo, Outreach and Smartsheet ask for an email address only; Smartsheet's leads to a recorded demo rather than a call. Zapier shows three fields and a Next button. monday.com, Airtable and 1Password ask nine on one screen. The median was five, as in our earlier sample; the Demo Request Conversion Rates 2026 post covers what field count does to completion.
Only one form says what happens next. Twilio's sales page promises contact "within one business day", twice. Okta and Clay say someone will be in touch, but only in the message shown after submitting. The other 17 say nothing about timing or call length. HockeyStack's December 2025 study of 80 B2B SaaS companies, vendor data from an analytics company, found demo pages get 0.91% of site traffic with a 70% bounce rate and a 5.5% conversion rate. Few visitors reach the form and most leave, so one sentence about what happens next is cheap insurance.
What ranks and what converts: what 20 sites can and cannot show
A scorecard of 20 sites cannot prove that any test moves rankings or revenue. These brands rank on brand searches and years of links whatever their hero says.
The outside evidence points the same way as the tests. Portent's 2022 study, an agency's analysis of about 100 million page views across 20 sites, 14 of them B2B lead generation, found B2B pages loading in one second converted three times better than pages loading in five. Google's guidance asks for descriptive titles and markup that identifies the company. NN/g asks for a tagline that says what the company does.
A smaller company cannot afford the same gap: a first-time visitor from an advert or an AI answer does not arrive knowing that "Build systems to grow revenue" means data enrichment.
What this means for a B2B SaaS website team
Run the 12 tests on your homepage and two competitors; each site takes under an hour.
1. Put the category back in the hero. If the title tag names it, the headline or the line under it should too, with who it is for.
2. Read your server HTML. Check that the H1 is there and that no template placeholder survived.
3. Run Lighthouse three times and take the median. Lighthouse's own documentation lists A/B tests, ads, routing and the test machine as causes of run-to-run swings, and Calendly's LCP ranged from 2.6 to 5.9 seconds across our three runs. If you have field data in Search Console, it outranks any lab run.
4. Walk the demo path. Count clicks and fields, then say when the visitor will hear back, if you can keep the promise.
A disclosure: we build marketing sites in Webflow and applications in Next.js. Four of these 20 homepages run on Webflow (monday.com, Lattice, Outreach and Clay) and 11 show Next.js markers in their HTML; none of the 12 tests depends on the platform. We would run these same 12 tests on a staging site before launch, and our marketing website service explains how we scope that work.
