1. Home
  2. Why TalkingDot
Why TalkingDot

Talk to every visitor — without slowing your site down

Live chat, an AI assistant, pop-ups and a “What’s new” feed come in one small script and one dashboard. Here is what you get, what published research says about each, and how fast the script really is — measured, not promised.

What you get with us

Several tools you would otherwise buy, install and pay for separately — with one script on your site and one place for your team to work.

Live chat and a shared inbox

A messenger on your website and one real-time inbox for the whole team, with notes, assignment and saved replies. Learn more about Live chat and a shared inbox →

An AI assistant

Answers from your website, help center and documents, and hands over to a person when it isn’t sure — on your own Claude or OpenAI key. Learn more about An AI assistant →

Pop-ups, bars, corner cards and inline forms

Collect emails, start chats or run a quick survey — after a delay, on scroll, on exit, after inactivity or from your own buttons. On every plan; spin-to-win and countdowns come with paid plans. Learn more about Pop-ups, bars, corner cards and inline forms →

A “What’s new” feed

Announcements in the messenger, behind your own button with an unread badge, as a one-time pop-up, by email and on a public changelog with RSS. With Starter, Pro, Lifetime and Reseller accounts. Learn more about A “What’s new” feed →

Targeting

Show pop-ups, posts and auto messages only to visitors who match: their plan, tags, visits or any attribute your site sends. Learn more about Targeting →

WhatsApp, Telegram and email

Customers write where they already are; every message lands in the same inbox. Learn more about WhatsApp, Telegram and email →

A help center

Articles your visitors can search from the messenger, before they need to ask. Learn more about A help center →

Chatbots and automation

Collect details, route to the right team and answer the common questions automatically. Learn more about Chatbots and automation →

An API, webhooks and a JavaScript API

Connect your own systems, identify logged-in users and open the messenger from your own buttons. Learn more about An API, webhooks and a JavaScript API →

One light script — measured

We loaded the same page with and without the TalkingDot messenger, as a first-time visitor on a slow phone connection (Chrome with the CPU slowed 4× and “slow 4G”, like Google’s mobile test), 11 times each, and took the middle result. The files were the ones we ship: minified and compressed.

9.2 KBon the critical path: the small loader that shows the chat button
55.0 KBthe rest of the messenger (chat, AI, pop-ups and “What’s new”): loaded after the page, or right away when something must show, then kept in the browser’s cache
0 msblocking time with the standard snippet in our test, and no layout shift
0.69 suntil the chat button is ready, on that slow connection
On a slow phone connectionWithout the messengerStandard snippetAfter-load snippet
Page content appears (FCP)240 ms236 ms232 ms
Main content shown (LCP)240 ms236 ms232 ms
Layout shift (CLS)000
Blocking time (TBT)0 ms0 ms5 ms
Browser’s “load” event232 ms446 ms233 ms
Chat button ready—0.69 s0.74 s
Script running time: the loader—50 ms55 ms
Script running time: the rest, after the page—22 ms30 ms

What visitors see is the same with or without the messenger: the first and main content appear within a few milliseconds of each other — about the difference between two runs of the same page — and in our test the page didn’t jump or freeze. With the standard snippet the browser’s “load” event waits for the loader (0.21 s on this slow connection); with the after-load snippet it doesn’t move at all. One small request (12.5 KB) fetches your settings, and the rest of the messenger (55.0 KB) follows after the page, or right away when something must show (an unread reply, a pop-up that is due). Measured November 19, 2026 on version 1.30.0, with a simple test page; your own page’s numbers will be different.

The fastest way to install it

Paste this instead of the standard one-line snippet, just before </body>. The messenger then loads after your page has finished, and your own code can already call TalkingDot(…). Your workspace key is in Settings → Installation.

<script>
  (function (w, d) {
    w.TalkingDot = w.TalkingDot || function () { (w.TalkingDot.q = w.TalkingDot.q || []).push(arguments); };
    function load() {
      var s = d.createElement('script');
      s.src = 'https://talkingdot.com/widget.js';
      s.async = true;
      s.setAttribute('data-key', 'YOUR_WORKSPACE_KEY');
      d.body.appendChild(s);
    }
    if (d.readyState === 'complete') load(); else w.addEventListener('load', load);
  })(window, document);
</script>

Why it works: what the research says

Studies and reports by other companies, each with its year and a link to the source. They describe averages over many websites — they aren’t promises about yours. Where a figure comes from a vendor’s own customers, we say so.

Live chat: People expect an answer now

83%

of customers expect to interact with someone immediately when they contact a company.

Salesforce, State of the Connected Customer, 5th edition · 2022Survey of 13,020 consumers and 3,916 business buyers in 29 countries.
88%

of customers expect faster responses than a year ago, and 74% expect customer service around the clock.

Zendesk, CX Trends 2026 · 2026Survey of 6,182 consumers in 22 countries.
1 in 4

chats were left in the queue before anyone answered (27.4%), across 1.68 billion chats.

LiveChat, Customer Service Report · 2024Data from LiveChat’s own customers.
64%

of customers would prefer companies didn’t use AI for customer service. Their biggest worry: not reaching a person.

Gartner press release · 2024Survey of 5,728 customers, December 2023.

How TalkingDot helps: Reply in seconds from the inbox or your phone, show your reply time and opening hours, and let the AI assistant answer out of hours — always with a way to reach a person.

Pop-ups: A well-timed pop-up turns visitors into contacts

3.09%

average conversion rate across 1.75 billion pop-ups; the best 10% averaged 9.28%.

Sumo, pop-up statistics (archived copy) · 2019Data from Sumo’s own customers; last updated April 2019.
4.82%

average conversion of email pop-ups. Shown after 11–15 seconds they converted at 6.45%, against 4.16% shown right away.

Wisepops, pop-up statistics · 2025Data from Wisepops’ own customers.
2.1%

average email pop-up conversion rate across 1.24 billion displays.

Omnisend, email pop-up performance · 2025Data from Omnisend’s own customers, calendar year 2025.

How TalkingDot helps: Pop-ups, bars and corner cards with delay, scroll and exit triggers, shown only to the visitors you choose (new, returning, no email yet), with results per pop-up. Vendors count conversions differently, so their averages range from about 2% to 5%.

Announcements: Most new features go unnoticed

80%

of features in the average software product are rarely or never used.

Pendo, Feature Adoption Report · 2019Data from 615 Pendo customers’ products over three months.
60.2%

of people who see an in-app message with a call to action engage with it, in the best-performing products (top 10%).

Pendo, Product Benchmarks · 2024Data from 6,800+ applications of Pendo customers.

How TalkingDot helps: Tell people where they already are: a “What’s new” tab in the messenger, an unread badge on your own menu button, a one-time pop-up for big news and an email to subscribers. Replies come into your inbox as chats.

Page speed: Every tenth of a second counts

+8.4%

more retail conversions when a mobile site got 0.1 seconds faster (travel: +10.1%).

Deloitte for Google, Milliseconds Make Millions · 202037 brands in Europe and the US; a statistical model, not a controlled test.
+32%

more likely that a mobile visitor leaves when a page takes 3 seconds to load instead of 1.

Google / SOASTA research (archived copy) · 2017A model trained on mobile site data.
53%

of mobile visits are abandoned if a page takes longer than 3 seconds to load.

Google (archived copy) · 2016Google Analytics data from 3,700 mobile sites, March 2016.
92%

of web pages load scripts from other companies — and chat and contact tools are among the heavier ones.

HTTP Archive, Web Almanac: Third Parties · 2024Independent research; home pages only.

How TalkingDot helps: One script for chat, AI, pop-ups and announcements instead of one per tool, built to stay out of the way. Our speed test ↑

Try it on this page

This page runs the real TalkingDot messenger, with our own release notes as the “What’s new” feed.

More about the announcements widget → · Every feature →

Frequently asked questions

Will a chat widget slow down my website?

Not in what your visitors see. In our test on a slow phone connection, the page’s first and main content appeared within a few milliseconds of the same page without the TalkingDot messenger — no more than two runs of the same page differ — with 0 ms of blocking time and no layout shift. Only 9.2 KB is on the critical path; the rest of the messenger (55.0 KB) loads after the page, or right away when something must show. Use the after-load snippet and even the browser’s “load” event stays the same.

Do I need separate tools for pop-ups and announcements?

No. Live chat, the AI assistant, pop-ups, bars, corner cards and the “What’s new” feed all run from the same script and the same dashboard. Live chat and pop-ups are on every plan, including Free; announcements come with Starter, Pro, Lifetime and Reseller accounts.

Where do the statistics on this page come from?

From studies and reports published by other companies — Salesforce, Zendesk, Gartner, Google, Deloitte, Pendo, HTTP Archive and pop-up tool makers — each linked with its year. Many come from the vendor’s own customers, which we note under each figure. They are averages across many websites, not a promise of what will happen on yours.

How did you measure the speed?

With a script that opens the same test page in Chrome with and without the messenger (the minified, compressed files we ship), as a first visit with an empty cache, the CPU slowed four times and a “slow 4G” connection (150 ms latency, 1.6 Mbit/s) — the conditions of Google’s mobile Lighthouse test — 11 times each, taking the median. The numbers on this page are from November 19, 2026.

One script. Every way to talk to your visitors.

Start free — live chat and pop-ups are on every plan, including Free.

Free plan foreverNo credit cardUnlimited teammates