Not one of 20 marketing technology homepages we opened on 29 October 2026 said how long a demo call lasts or when sales will reply. Seven of the 20 heroes scored zero on clarity, and only six named a product category a buyer would type into a search box. Yet 19 of 20 put a demo, sales or sign-up form one click from the first screen.
Below are the nine tests, every measurement, and what the top scorers, Klaviyo, HubSpot, Mailchimp and Amplitude, do that others could copy. The tests match the ones in our cybersecurity scorecard, 20 Cybersecurity Websites Scored on Speed, Clarity and Demo Path (2026), so the two samples compare directly.
The rubric: nine pass or fail tests
Each test passed scores one point, so each dimension scores 0 to 3 and the total is out of 9.
Clarity, three tests. C: the hero names a product category buyers already use. Coined labels fail. A: it names who the product is for, as a role, team or company type. S: it gives one concrete specific, such as named functions, channels or a number. The basis is Nielsen Norman Group's first homepage guideline, which asks for a tagline that "summarizes what the site or company does".
Speed, three tests. Time to first byte of 0.8 seconds or less, web.dev's "good" line. HTML transfer of 155 KB or less, and of 78 KB or less: the 90th and 75th percentiles of mobile home page HTML in the Web Almanac's 2025 page weight chapter.
Demo path, three tests. A demo, sales or sign-up form one click or less from the first screen; five fields or fewer on its first step; and a stated call length or response time.
Clarity method. The hero means the headline, the line under it, any label directly above and any proof line directly below, read in a headless Chrome window at 1,024 by 768 pixels without scrolling. Cookie banners, promotion strips and pop-ups were ignored, so Mailchimp was scored on the hero behind its quiz.
Speed method. Five curl requests per homepage between 16:48 and 16:49 UTC from one connection in Serbia, taking the median time to first byte and the median compressed size of the HTML document as served (gzip or Brotli). 6sense answered curl with a bot-protection page (HTTP 403), so its speed is n/a.
Demo path method. We followed the most prominent demo, sales or sign-up action and counted visible text boxes, dropdowns, radio groups and text areas on the first step, in a rendered browser at 1,280 pixels wide. Consent and newsletter boxes do not count. We submitted nothing.
The scorecard: clarity and speed
Letters show which clarity tests a hero passed. Platform comes from signatures in the served HTML.
| Site | Product | Platform | Clarity | Passed | TTFB (s) | HTML (KB) | Speed |
|---|---|---|---|---|---|---|---|
| HubSpot | CRM and marketing platform | HubSpot CMS | 2 | AS | 0.06 | 102 | 2 |
| Mailchimp | Email marketing | Not identified | 2 | CS | 0.13 | 52 | 3 |
| Klaviyo | Email and SMS, ecommerce | Astro, Sanity | 3 | CAS | 0.28 | 80 | 2 |
| Braze | Customer engagement | Next.js, Sanity | 1 | A | 0.15 | 169 | 1 |
| Iterable | Cross-channel marketing | WordPress | 1 | C | 0.10 | 104 | 2 |
| Customer.io | Messaging automation | Next.js, Sanity | 0 | none | 0.15 | 65 | 3 |
| ActiveCampaign | Marketing automation | Craft CMS | 1 | S | 0.08 | 94 | 2 |
| Attentive | SMS and email marketing | Webflow | 1 | S | 0.05 | 55 | 3 |
| Semrush | Search visibility | Not identified | 0 | none | 0.33 | 21 | 3 |
| Similarweb | Digital market data | WordPress, headless | 0 | none | 0.87 | 307 | 0 |
| Hootsuite | Social media management | Next.js, Contentful | 1 | C | 0.14 | 75 | 3 |
| Sprout Social | Social media management | Next.js | 1 | C | 0.24 | 35 | 3 |
| Brandwatch | Social listening | WordPress | 1 | C | 0.33 | 136 | 2 |
| 6sense | ABM and intent data | WordPress | 0 | none | n/a | n/a | n/a |
| Demandbase | ABM | WordPress | 0 | none | 0.06 | 39 | 3 |
| Mixpanel | Product analytics | Framer | 0 | none | 0.66 | 118 | 2 |
| Amplitude | Digital analytics | Next.js, Sanity | 2 | AS | 0.88 | 55 | 2 |
| Contentsquare | Experience analytics | Next.js, Contentful | 1 | S | 0.15 | 68 | 3 |
| Hightouch | Composable CDP | Next.js, Sanity | 0 | none | 0.13 | 118 | 2 |
| Jasper | AI marketing content | Webflow | 1 | A | 0.05 | 102 | 2 |
Source: Octo Agency, headless browser read and five curl runs per homepage, 29 October 2026; raw file OB-0169-martech-scorecard-2026-10-29. Mixpanel's time includes a redirect to /home/.
The scorecard: demo path and totals
| Site | Form | Clicks | Fields | Timing stated | Demo path | Total |
|---|---|---|---|---|---|---|
| HubSpot | Demo, email first | 1 | 1 | No | 2 | 6 |
| Mailchimp | Sign-up via pricing | 2 | 2 | No | 1 | 6 |
| Klaviyo | Sign-up | 1 | 2 | No | 2 | 7 |
| Braze | Demo | 1 | 6 | No | 1 | 3 |
| Iterable | Demo | 1 | 2 | No | 2 | 5 |
| Customer.io | Demo | 1 | 5 | No | 2 | 5 |
| ActiveCampaign | Demo | 1 | 4 | No | 2 | 5 |
| Attentive | Demo | 1 | 7 | No | 1 | 5 |
| Semrush | Free sign-up | 1 | 1 | No | 2 | 5 |
| Similarweb | Demo, multi-step | 1 | 3 | No | 2 | 2 |
| Hootsuite | Demo | 1 | 8 | No | 1 | 5 |
| Sprout Social | Demo | 1 | 9 | No | 1 | 5 |
| Brandwatch | Demo | 1 | 6 | No | 1 | 4 |
| 6sense | Demo | 1 | 4 | No | 2 | 2 of 6 |
| Demandbase | Book a meeting | 1 | 4 | No | 2 | 5 |
| Mixpanel | Demo | 1 | 5 | No | 2 | 4 |
| Amplitude | Contact sales | 1 | 5 | No | 2 | 6 |
| Contentsquare | Demo | 1 | 7 | No | 1 | 5 |
| Hightouch | Demo, email then scheduler | 1 | 1 | No | 2 | 4 |
| Jasper | Demo | 1 | 4 | No | 2 | 5 |
Source: Octo Agency, rendered browser read of each form, 29 October 2026; no form was submitted. Fields counts the first step only. Iterable's page renders its two-field form twice; we counted it once. Attentive's channel checkboxes count as one question.
Clarity: coined categories and promotion strips
The most common failure is a vendor-made category. HubSpot's label reads "Agentic customer platform". Klaviyo's reads "The autonomous B2C CRM". 6sense calls itself "Intelligence for Agentic GTM", Mixpanel sells "AI-powered product intelligence" and Contentsquare opens with "Every journey. Human loved. Agent approved." A buyer searching for ABM software, a CDP or product analytics finds none of those words in the hero.
The odd part is that the same companies use the plain words elsewhere. 6sense's title tag says "The ABM Platform". Demandbase's says "ABM Platform for B2B Sales & Marketing Success". Hightouch's says "Customer Data & AI Platform for Marketers (CDP & AI Agents)". The search snippet is clearer than the first screen.
The market explains some of this. Chiefmartec's 2025 landscape lists 15,384 marketing technology products, up 9% in a year, and removed 1,211 that were acquired or shut down. In a category that crowded, a vendor wants to sound different. But a coined label only works for buyers who already know the brand.
Klaviyo shows how to keep both. It keeps its coined line, "Your AI marketing agent, always a step ahead", but the label above names a CRM for B2C brands, and the proof line below the buttons gives a number: "205,000+ relationship-driven brands across 100 countries". That is all three tests in about 20 words.
Amplitude names the buyer. "A new era for product teams" fails the category test on its own, but the subheading lists "analytics to feedback to code". That passes the audience and specifics tests. HubSpot does the same by naming marketing, sales and service in its subheading.
Social tools are the plainest. Hootsuite ("Rated #1 in social media management"), Sprout Social ("Powerful tools for social media management") and Brandwatch ("The most intelligent social suite") all pass the category test, and nothing else. None names a team or a number.
AI is in 13 of 20 heroes. That alone did not decide anything. Klaviyo mentions AI and scores 3; Hightouch, Mixpanel and 6sense mention it and score 0, because AI is the only concrete word on their first screen.
First screens are crowded. At least eight carried a promotion strip above the hero: a Black Friday guide at Attentive, a webinar at Amplitude, a product launch at Klaviyo and Hightouch, a discount at Mailchimp. Mailchimp's first load also opened a quiz asking for industry and goals, which covered the hero entirely. Hootsuite opened an AI chat agent over its hero image. Each may lift its own metric, but a first-time visitor has to close or read past them before learning what the product is.
Speed: quick servers, a few heavy documents
Time to first byte separated almost nobody. 17 of 19 measured homepages answered within 0.8 seconds, with a median of 0.15 seconds. Similarweb (0.87 seconds) and Amplitude (0.88 seconds) were just over. web.dev notes that TTFB is not a Core Web Vitals metric, so this measures the server, not the full experience.
Document weight is where the spread sits. The median homepage sent 80 KB of compressed HTML, against the Web Almanac's 33 KB median and 78 KB 75th percentile for mobile home pages. Nine of 19 came in at 78 KB or less. Two went past 155 KB, and we decoded the five heaviest to see why.
| Site | Compressed HTML | Uncompressed HTML | Largest inline parts |
|---|---|---|---|
| Similarweb | 307 KB | 919 KB | 515 KB inline app data object, 274 KB inline SVG |
| Braze | 169 KB | 720 KB | 423 KB inline Next.js data, including 124 KB of base64 images |
| Brandwatch | 136 KB | 612 KB | 253 KB inline SVG |
| Mixpanel | 118 KB | 1,268 KB | 314 KB inline CSS, 266 KB inline SVG |
| Hightouch | 118 KB | 809 KB | 226 KB inline CSS |
Source: Octo Agency, curl fetch of each homepage and byte count by element type, 29 October 2026
The causes are ordinary. Large SVGs belong in image files, images should not be base64 strings inside a data payload, and a stylesheet that is linked can be cached for the next page. Our Website Speed Statistics 2026 post covers what page weight does to bounce and revenue; here the point is that none of these sites needs the extra weight.
Platform did not decide weight. Attentive and Jasper both run on Webflow, which we build on, so read this with our bias in mind. Attentive sent 55 KB and Jasper 102 KB. Mixpanel, on Framer, sent the most uncompressed HTML in the sample. The build choices, not the platform, set the size.
Demo path: short forms, no promises
Martech vendors keep forms short. The median first step asked 4 fields, against 7.5 in our cybersecurity sample. HubSpot, Hightouch and Semrush ask for an email address alone, and Klaviyo and Mailchimp ask for an email and a password. Similarweb splits its form and asks for a name and work email first.
The long forms belong to the social and analytics suites. Sprout Social asks nine questions on one screen, including company size, country and whether you are an agency. Hootsuite asks eight, with four of them dropdowns. Contentsquare and Attentive ask seven. For a sense of the cost, Factors.ai, a vendor, reports a 51% success rate for forms with four to six compulsory fields and 30.60% for seven or more, across a filtered sample of more than 50,000 sessions. Our post on web form conversion statistics sets that against other field-count studies.
There is a reason some keep the questions. Chili Piper, which sells scheduling software, reports that in marketing software "only 77% of form fills even get the option to book a meeting", so roughly a quarter are screened out. Company size and phone questions are how that screening happens. The trade is real, but a two-step form does the same screening after the visitor has already committed an email.
Nobody said what happens next. Not one of the 20 forms stated a call length or a reply time. In our B2B SaaS scorecard, 20 B2B SaaS homepages scored on the same demo tests, only Twilio did. Chili Piper's data says booking immediately after the form roughly doubles the rate at which qualified requests become meetings, from about 30% to 66.7%. Hightouch is the only site here that goes straight from a work email to a scheduler, which is the closest thing to a stated next step.
Three lead with self-serve. Klaviyo, Semrush and Mailchimp put a sign-up, not a demo, in front. That suits products with a free or low-cost plan. Mailchimp is the one that costs itself a click: its hero sign-up button opens the pricing page, and the form sits behind a plan choice.
What this means for a martech website
Run the tests on your homepage and two competitors; each site takes about 30 minutes.
1. Put the plain category next to the coined one. Keep the brand line if you like it, but add the search word, CRM, ABM, CDP or product analytics, in the label or the subheading. Your title tag probably already has it.
2. Name the team. Lifecycle marketers, product teams, ecommerce brands: two words cover the audience test that 15 of 20 failed.
3. Clear the first screen. One promotion strip at most, and no quiz or chat window over the headline on the first visit.
4. Weigh the HTML. Anything over 155 KB usually hides inline SVG, inline CSS or a data object that belongs in a separate file.
5. Say what happens after the form. A call length or a reply time is one sentence, and no site in this sample has it.
When we build a martech vendor's marketing site in Webflow, we would run these nine tests on staging before launch, as we would for any B2B site. Our marketing website service explains how we scope that work.
