saas · Sep 25, 2026

gpt-realtime.ai

We found more conversion tactics here than on 67% of the 1,350 multi-step SaaS funnels we have measured.

6 conversion tactics detected

2 pages walked · measured by Roentgen 1.27.0

Put this x-ray to work Brief your AI X-ray your funnel
How this was measured

Measured by Roentgen 1.27.0 on 2026-09-25. A capture is read once and its measurements freeze there, so the stamp says which ruleset read this one.

This is detection coverage, not a verdict — it counts the conversion tactics we could prove are present, out of the 11 that apply to a saas funnel. A low count means we did not detect those patterns, not that the funnel is weak.

Against all 1,781 multi-step funnels, more conversion tactics than 72% of them.

Walk 5 links on the entry page, 1 walkable. 1 matched a funnel step; we captured 2 pages in all, this one included.

The short version, in 5 parts

  1. Proven on the page

    6 conversion tactics detected, each backed by text on the page

    Answers buyer objections
    What does the credit pool mean? + Each paid plan includes one shared credit pool. Starter
    Offers a free trial or free tier
    Try It Free
    Shows customer logos or a trusted-by claim
    logos: Toolpilot, Powered by Startup Fast, Submit AI Tools, Listed on ToolRain, Dang.ai

    + 3 more with evidence

    See all 6

  2. Not seen

    5 of 11 applicable tactics we checked for and did not see

    Us-vs-them comparison
    A comparison table framing the brand against alternatives.
    Lead-capture form
    An on-page form collecting contact details (email / name / phone) outside a che…
    Proof above the fold
    Social proof rendered within the first viewport on at least one step.

    + 2 more

    See what we looked for

  3. Flagged

    5 of 16 build checks flagged, each with a fix

    Text contrast (WCAG AA)
    89% of measurable text passes WCAG AA. 4+ runs fail.
    Body text size
    Smallest sampled text is 10px, below the 13px floor this check uses.
    Named controls
    1 button/link has no accessible name (icon-only, no aria-label).

    + 2 more

    See the fixes

  4. Walked

    2 pages walked, Homepage / Lander to Pricing

    Step 1
    Homepage / Lander
    Step 2
    Pricing

    Walk it frame by frame

  5. In their words

    The headline: GPT Realtime AI Voice Generator

    Promise
    Try GPT Realtime for low-latency voice agents, speech-to-speech demos, image-aware support, SIP calls, and AP…

    Read every quote

Captured Homepage / Lander
Step 1 · Homepage / Lander1 / 2

What we recorded

Progress

3 scans

this funnel’s coverage score across measured re-scans

measuredextracted directly from the live page. a fact, not an opinion
Unchanged across 3 scans Same 6 conversion tactics detected since 2026-08-09.
  1. 6 tactics 2026-08-09
  2. 6 tactics 2026-08-15
  3. 6 tactics 2026-09-25

The funnel, frame by frame

2 steps walked

see their whole funnel without clicking through it. scored step by step

measuredextracted directly from the live page. a fact, not an opinion
Homepage / Lander
01

Homepage / Lander

▼ thinnest
5 tactics
CountdownPressFree trialLogo wallDemo / docs
Pricing
02

Pricing

▲ strongest
8 tactics

$19.9 +18 more  ·  save 20%  ·  subscription  ·  1 proof point

CountdownFAQPressFree trialLogo wallPricing tiersDemo / docsExpress pay

In their own words

15 quoted

the entry page’s own title, headline, buttons and prices — quoted, never paraphrased

measuredextracted directly from the live page. a fact, not an opinion
Page title
GPT Realtime Free: #1 AI Voice Generator for Calls
Meta description
Try GPT Realtime for low-latency voice agents, speech-to-speech demos, image-aware support, SIP calls, and API workflows. Start building voice apps free now.
Headline
GPT Realtime AI Voice Generator
Section headings
Realtime conversationWhat Is GPT RealtimeStart a realtime sessionSession setupGPT Realtime speech-to-speechGPT Realtime API workspaceGPT Realtime voice agentsGPT Realtime model testing
Buttons and links
Start for FreeStarter Voice (single-turn)Start Free

Sections

3 identified

structural bands identified on this funnel, cropped from its own capture

measuredextracted directly from the live page. a fact, not an opinion
Hero section of gpt-realtime.ai
Hero1% down the page
Logos section of gpt-realtime.ai
Logos65% down the page
Footer section of gpt-realtime.ai
Footer96% down the page

Elements

1 harvested

the atomic pieces — price anchors, ratings, guarantees — with their verbatim on-page text

measuredextracted directly from the live page. a fact, not an opinion
Price anchor element
Price anchor“$19.9”

What we found

6 tactics

conversion tactics we could prove are present, with the on-page text that evidences each

measuredextracted directly from the live page. a fact, not an opinion

objection1 of 2

Answers buyer objections What does the credit pool mean? + Each paid plan includes one shared credit pool. Starter

proof2 of 4

Shows customer logos or a trusted-by claim logos: Toolpilot, Powered by Startup Fast, Submit AI Tools, Listed on ToolRain, Dang.ai
Shows press logos or an as-seen-in claim featured on

offer2 of 2

Offers a free trial or free tier Try It Free
Makes the pricing ladder explicit $15.92/mo, $31.92/mo, $79.20/mo, $239.20/mo

checkout1 of 2

Offers a demo or docs entry point docs link

What we didn’t find

5 of 11 applicable

applicable tactics we checked for and did not see on any step we reached

measuredextracted directly from the live page. a fact, not an opinion

What we didn’t see5 of 11 applicable

Us-vs-them comparison 441 of the 2,420 funnels we checked
Lead-capture form 437 of the 1,683 funnels we checked
Proof above the fold 381 of the 2,420 funnels we checked
Usage / adoption proof 289 of the 2,043 funnels we checked
Scarcity / urgency copy 125 of the 2,417 funnels we checked

Absence is a measurement, not a recommendation: we do not know why a tactic is missing, whether it belongs here, or whether adding it would change anything. Where a walk stopped early, a tactic living deeper would also read as not detected.

How well it’s built

16 run · 5 flagged

contrast, mobile, typography, accessibility, weight — each with what would fix it

measuredextracted directly from the live page. a fact, not an opinion

5 of 15 measured checks need work1 could not be measured — not counted either way

Text contrast (WCAG AA)

contrast:text-contrast-wcag-aa

89% of measurable text passes WCAG AA. 4+ runs fail.

worst: “Voice Lab:” at 1.03:1 (14px, #ffffff on #fffbfe). needs 4.5:1

FixRaise text/background contrast to ≥4.5:1 (≥3:1 for large text), especially on CTAs and body copy.

Body text size

type:body-text-size

Smallest sampled text is 10px, below the 13px floor this check uses.

FixKeep body copy ≥15-16px; reserve sub-13px for fine print only.

Named controls

a11y:named-controls

1 button/link has no accessible name (icon-only, no aria-label).

FixAdd aria-label to icon-only buttons and links.

Oversized images

weight:oversized-images

25 images are shipped at more than 2x the width shown.

“1.jpg” is 128px wide, displayed at 40px

FixServe responsive sizes (srcset/sizes) matching the rendered width.

Tap-target size (mobile)

mobile:tap-target-size-mobile

39 of 46 interactive targets are under 44×44px on mobile.

smallest: “Dismiss banner” at 32×32px

FixGive tappable controls a ≥44×44px hit area. the touch minimum.

Worth knowing7

Line length

9 paragraphs run past ~95 characters per line.

FixCap measure at ~60-75ch for comfortable reading.

type:line-length
Form labels

1 form field relies on placeholder text as the only label.

FixBack placeholders with a real <label>. placeholder text vanishes on input.

a11y:form-labels
Label in name

1 control has an aria-label that does not contain its visible text, so voice-control users cannot activate it by saying what they see.

FixMake the accessible name start with the visible label text.

a11y:label-in-name
Render-blocking resources

4 stylesheets/scripts in <head> block first paint.

FixDefer non-critical JS (defer/async) and inline only critical CSS.

weight:render-blocking-resources
Keyboard reachability

16 elements look clickable but cannot be reached by keyboard.

FixUse <button>/<a>, or add role + tabindex="0" and a key handler.

structure:keyboard-reachability
Autocomplete hints

No autocomplete attributes across 1 field. browsers cannot autofill.

FixAdd autocomplete="email"/"tel"/"name" to identity fields.

forms:autocomplete-hints

Not measurable1

Page weight

At least 1.1 MB across 73 measurable resources, but 43 cross-origin resources did not expose a size. The true weight cannot be determined from the page.

weight:page-weight

Clear3

Mobile viewport · Language declared · Image alt text

Brand kit

5 colour roles

palette by on-page usage · type rendered live

measuredextracted directly from the live page. a fact, not an opinion
full palette ordered by on-page usage %
accent#f3edf7
bg#f3edf7
surface#f7f2fa
text#49454f
See the roles applied
accent

Body copy in their text role, on a surface card.

the roles, doing their job
The quick brown fox60px
Jumps over the lazy dog48px
A subheading in their voice36px

Body copy set at their measured paragraph size. This is how their product reads to a first-time visitor. same family, same scale, same weight.

  • List item one, at body scale
  • List item two
  • List item three
Type scale, weights & families
SIZE SCALE
12pxAa
13pxAa
14pxAa
15pxAa
16pxAa
18pxAa
24pxAa
36pxAa
48pxAa
60pxAa
WEIGHTS
Aa400
Aa500
Aa600
Aa700
Aa800
FAMILIES
ui-sans-serif
rendered in their own type system · webfont not loaded. system fallback if uninstalled

Measured, found none

1 of 10 layers

layers we ran on this funnel that came back empty

measuredextracted directly from the live page. a fact, not an opinion

Playbook match — which funnel play this page runs — measured, evidence-quoted

We ran this pass on the pages we captured and found nothing to show. That is a measurement of this funnel, not a recommendation — and it is a different statement from a layer we never ran, which does not appear here at all.

Similar crawls