Most live chat rollouts do not fail because of the software. They fail because the team signs up on Monday, pastes the widget code on Tuesday, and then stares at an empty chat queue on Friday wondering why nobody is using it. The widget is the easy part. The hard part is knowing who should answer chats, when, what they should say, and how you will know whether any of it is working.
This checklist walks you through that hard part. It covers everything from signing up to your first real customer conversation, broken into eight phases you can realistically complete in one to two weeks. Each phase has concrete steps — no vague advice like “define your strategy.” Work through them in order, check each item off, and you will launch with a chat operation that actually earns its place on your site.
How to Use This Checklist
Read the whole thing once to see the shape of the project, then execute phase by phase. Phases 1 and 2 are planning and can be done in a single afternoon meeting. Phases 3 through 6 are technical setup and testing — the part your developer or technically inclined teammate handles. Phases 7 and 8 are the launch itself and the first round of measurement. If you are a team of one, budget a full day for setup and a week of soft launch. If you have a small support team, you can compress the whole thing into three or four working days.
Rule of thumb: do not go live until you have completed Phase 6 (testing). A broken chat widget — one that misses messages, shows the wrong offline text, or pops up on every page load — does more damage to trust than having no chat at all.
Phase 1: Decide What Chat Is Actually For
Live chat is a tool, not a strategy. Before you configure anything, answer these four questions in writing. This takes thirty minutes and prevents months of drifting.
- Pick one primary goal. Support deflection, lead capture, pre-sale questions, onboarding help — choose the one that matters most right now. You can expand later, but a chat that tries to do everything gets configured for nothing in particular.
- Define who it serves. Write one sentence: “Our chat exists for [audience] who need [help] at [moment].” For example: “Our chat exists for pricing-page visitors who need to confirm the plan fits their team before trialing.” This sentence will guide every trigger and greeting you write.
- Decide the human-bot split. Will a person answer every chat, or will a bot handle first contact and escalate? Most small teams do best starting 100% human during business hours with a simple offline form after hours. Add automation only after you have seen fifty real conversations and know which questions repeat.
- Record your baseline. Before launch, note your current numbers: support ticket volume, conversion rate on key pages, average response time on email. You cannot measure improvement without a starting point — our guide to measuring live chat ROI explains which numbers to track and how to compute them.
Phase 2: Staff the Team and Set the Hours
Chat is real-time, which means coverage is a promise you make to visitors. An offline widget that says “we usually reply in minutes” while your inbox fills up for two days is worse than no promise at all.
- Choose your coverage window. Start with the hours your team genuinely works — do not promise 24/7 on day one. Check your analytics for when visitors actually browse; most B2B sites concentrate traffic in a predictable window.
- Assign real people, with backups. Name the primary agents and one backup each. A named owner per shift beats a shared “everyone watches” arrangement, which reliably means nobody watches.
- Set a realistic concurrency limit. New agents should handle no more than two simultaneous chats; experienced ones can manage three or four. Most platforms let you cap chats per agent — set this before launch, not after your first agent gets buried under six conversations.
- Write the offline experience. Decide exactly what visitors see when nobody is online: a contact form, an email address, or a bot that takes a message. Draft that copy now, while you are calm, not at 9pm when you realize the default text says something odd.
Phase 3: Install and Configure the Widget
Now the technical part. On most platforms this is a single JavaScript snippet pasted before the closing body tag of your site — either directly, through a tag manager, or via a plugin. Work through these settings deliberately:
- Install the snippet on a staging page first, if you have one. Load the page, confirm the launcher appears, and send a test chat before touching production.
- Match the branding. Set your brand colors, upload your logo, and choose a launcher style that fits your design. The widget should look like part of your site, not a third-party afterthought.
- Set the position and behavior. Bottom-right is the standard for a reason — it avoids clashing with navigation and cookie banners. Decide whether the widget auto-expands, stays as a bubble, and how it behaves on mobile. See our chat widget placement guide for the full do-and-don’t list.
- Configure the pre-chat form (or skip it). Asking for a name and email before the chat starts improves follow-up but adds friction. A good compromise: no pre-chat form for general questions, a short form for sales routing where the email is genuinely needed.
- Set up departments or routing rules. Even with two people, route sales questions to one and support questions to the other. Simple keyword routing or a quick “what is this about?” chooser works fine at small scale.
- Write your offline form. Title it clearly, ask for email plus message, and set expectations: “We reply within one business day.” Then make sure someone actually monitors the inbox it feeds.
The table below summarizes the settings worth reviewing before you consider installation “done.” Defaults are fine for everything not listed here.
| Setting | What to decide |
|---|---|
| Launcher position | Bottom-right on desktop; bottom-right or bottom-center on mobile, clear of sticky CTAs |
| Auto-open behavior | Off by default; enable only for specific proactive triggers |
| Pre-chat form | Off for support, short form for sales — decide per department |
| Offline behavior | Contact form with clear response-time promise |
| Chat history | Keep visitor history across sessions so repeat visitors are recognized |
| Notifications | Sound + desktop notification for agents; test them during setup |
Phase 4: Write Greetings and Triggers
A live chat that just sits there waiting is a missed opportunity; a chat that assaults every visitor with popups is a liability. The middle ground is two or three carefully chosen proactive triggers with honest, helpful copy.
- Write one default greeting for the widget itself — the short line visitors see before they type. “Hi there! Questions about plans or setup? We reply in under a minute during business hours.” Specific beats generic.
- Build one time-on-page trigger for your highest-intent page (usually pricing or a key product page): after 45–60 seconds, offer help once. Cap it at one display per visitor per day.
- Build one behavior trigger for a hesitation signal you care about — for example, a visitor who reaches the checkout or signup page and idles, or someone who visits the pricing page twice in one session.
- Set frequency caps globally. No visitor should see more than one proactive message per session, and never show a trigger to someone who just closed one. Most platforms support “do not show again after dismiss” rules — turn them on.
For the full playbook of trigger ideas and example messages that do not annoy people, write every message against a specific visitor intent and cap each trigger at one display per session — relevance plus restraint is the whole formula.
Phase 5: Connect Chat to Your Other Tools
Chat works best when it is not an island. At minimum, connect these before launch:
- Your help desk or ticketing system. Offline messages and escalated chats should land where your team already works, not in a separate inbox nobody checks.
- Your CRM. Even a basic integration that attaches chat transcripts to contact records pays off the first time an agent needs context on a returning customer.
- Your analytics. Tag chat-initiated sessions so you can later compare conversion rates of chatters versus non-chatters — this is the backbone of any ROI measurement.
- Email fallback. Confirm that missed chats and offline form submissions reach a monitored address, and send yourself a test to verify the whole path works end to end.
Most established platforms document these integrations in their help centers — for example, the Zendesk support documentation and Intercom help center both cover connecting chat with CRMs and help desks step by step. Follow your vendor’s guide rather than improvising with webhooks on day one.
Phase 6: Test Like a Visitor, Not Like an Admin
You know where everything is; your visitors do not. Test in a clean browser session (logged out, cache cleared) on both desktop and a real phone. Run through every row of this table:
| Test | What “passing” looks like |
|---|---|
| Widget loads on all key pages | Launcher visible and on-brand on homepage, pricing, product, and support pages |
| Online chat, end to end | Visitor message reaches the agent app within seconds; agent reply appears in the widget |
| Offline behavior | Outside coverage hours, visitors see the offline form with correct copy — not a dead chat box |
| Proactive triggers | Each trigger fires exactly when intended, at most once per session, and never on mobile in a way that blocks content |
| File and link sharing | Agents can send links and screenshots; visitors can attach files if you enabled it |
| Notifications | Agents get a sound and desktop notification for every new chat |
| Transcript emails | Visitor receives an accurate transcript when the chat ends, if you enabled them |
| Mobile layout | Widget opens full-screen or near-full-screen on phones and does not cover the checkout button |
Fix everything you find, then test once more. Two full passes is the norm, not a sign of failure.
Phase 7: Soft Launch With Real Traffic
Do not announce chat to the world on day one. Instead, run it quietly for one week:
- Enable chat on half your key pages first — for example, pricing and support pages but not the homepage. This keeps volume manageable while agents build confidence.
- Review every transcript daily. For the first week, read all of them. You will spot greeting copy that confuses people, triggers that fire at the wrong moment, and questions nobody prepared for.
- Hold a 15-minute daily debrief with whoever answered chats. Ask: what confused visitors today, what took too long, what should become a canned response or a trigger?
- Fix fast, in small batches. Tweak one trigger’s timing, rewrite one greeting, add one canned response — then watch for another day. Resist redesigning everything at once.
Phase 8: Measure and Iterate
After two weeks of live operation, pull your first real report and compare against the baseline from Phase 1. Look at chat volume by page, first response time, resolution rate, customer satisfaction, and — if chat touches sales — conversion differences between chatters and non-chatters.
Then pick exactly one improvement for the next two weeks: a new trigger, extended hours, or better routing. Chat operations improve through iteration, not through a perfect launch. And if the numbers suggest your current plan tier is wrong for your volume — too limited or needlessly expensive — revisit the math in our live chat pricing guide before your renewal comes up.
Conclusion
That is the full checklist: define the purpose, staff the hours, install and configure the widget, write your greetings and triggers, connect your tools, test like a visitor, soft-launch for a week, then measure and iterate. None of these steps requires special expertise — just the discipline to do them in order instead of skipping to the fun part. Your first real conversation is closer than it looks: most teams that follow this list are chatting with customers within ten days of signup, and chatting well within a month.
