Proactive chat fails the same way at most B2B SaaS companies: someone enables "send a greeting after 10 seconds" site-wide, conversion barely moves, the team blames live chat, and the widget gets demoted to a reactive button. The fix isn't more aggressive triggers — it's better-targeted ones. The nine proactive chat rules below tie every fire to a real intent signal: dwell time on the right page, return visits, scroll depth, or exit motion. Most teams ship two or three of these and leave the rest of the conversion lift on the table.
Key takeaways
- Proactive chat rules work when they combine a page-pattern signal (pricing, booking, comparison) with a dwell or behavior threshold — never one without the other.
- A blanket "greet every visitor after 10 seconds" rule typically annoys more visitors than it converts; tighter rules with fewer fires almost always win on net bookings.
- Exit-intent and return-visitor triggers are the highest-yield rules most B2B SaaS teams haven't shipped yet.
- Each rule should fire at most once per visitor session, and the system should respect Do Not Track signals to avoid widget-fatigue and compliance headaches.
- Treat proactive rules as a portfolio: ship 6-9 narrowly-scoped rules instead of 1-2 broad ones, and measure each by accepted-message rate, not just fire count.
What "proactive" actually means in 2026
Proactive chat means the widget initiates the conversation based on visitor behavior, instead of waiting for a click on the launcher. In 2026 the bar is higher than it was three years ago: visitors have seen the "Hi there 👋 need help?" pattern thousands of times and ignore it by default. A proactive message that gets opened in 2026 has to feel context-aware — referencing the page the visitor is on, the action they're about to take, or a piece of friction they're visibly stuck on.
The mechanical building block is simple: a rule with a page-URL condition, a dwell-time threshold, and a message. The judgement call is which combinations of those produce conversations a visitor actually wants. The nine rules below are the ones that consistently earn their fire across B2B SaaS teams running a free trial or a booking-led motion.
1. Pricing-page dwell over 45 seconds
The classic, and still the highest-yield single rule for self-serve SaaS. A visitor on /pricing who has been on the page for 45+ seconds is either comparing tiers, reading the FAQ, or hunting for a price they think is hiding. All three are good moments to offer help — and all three have a chat-accept rate several times higher than a generic homepage greeting.
The specifics matter. Forty-five seconds is roughly the median dwell on a pricing page for someone who's reading rather than bouncing; firing earlier catches scanners who aren't ready. The message should reference the page ("Happy to walk through which plan fits your team size — just reply with how many agents you need"), not generic pleasantries.
2. Second visit to the booking page without booking
A visitor who landed on /book/demo once, left without picking a slot, and came back is signaling intent and friction in equal measure. They want the meeting; something stopped them. Maybe nobody at the right level was available in their timezone. Maybe the form fields scared them off. Maybe they wanted to check with a colleague first.
This rule needs a return-visitor signal — usually a cookie or localStorage flag set on the first booking-page visit. Fire only on the second view, with a message like "Saw you stopped by yesterday — anything blocking the demo? Happy to find a slot manually." Accept rates on this rule are typically 3-5x the pricing-page baseline because the self-selection is so strong.
3. Comparison or alternatives page after 30 seconds
If your site has /vs/competitor or /alternatives pages — and it should — visitors there are mid-evaluation. They've identified your category, they're shortlisting, and they have specific objections. Thirty seconds of dwell is the threshold where they've actually read past the headline.
The message here should be more substantive than a greeting. "Most folks comparing us to [competitor] ask about migration time and pricing for sub-15-agent teams — happy to answer either" outperforms "Have questions?" by a wide margin. You're earning the conversation by naming the objection upfront.
4. Scroll depth + dwell on a long-form pricing or product page
For teams whose pricing page is a single long scroll (feature matrix, FAQ, fine print), pure dwell time misclassifies fast readers and slow readers alike. Combining scroll depth (60%+) with dwell (30+ seconds) catches the visitor who has actually engaged with the content rather than left the tab open in a background window.
Most chat tools require a small custom event to fire scroll-depth signals; it's worth the 20 minutes of setup. The rule should also exclude visitors who've already scrolled to the bottom and started scrolling back up — they're rereading, not stuck.
5. Exit-intent on the booking or checkout page
Exit-intent — the mouse moving toward the browser chrome at speed — is the single highest-intent signal a website visitor produces. Firing a proactive chat on exit from /book/demo or /signup catches the person who was 80% of the way to converting and bailed at the last step.
The message has to be different from the others. This isn't "happy to help" — it's a specific salvage offer: "Before you go — if scheduling didn't work, reply here and we'll find a time async." Some teams pair this with a magic-link CTA so the visitor can leave their email and get a calendar link by email instead of finishing the form. Exit-intent rules don't fire often, but their per-fire conversion rate is the highest in the portfolio.
6. Return visit from a knowledge-base article to the pricing page
The path "help center article → pricing page" is one of the most underrated intent signals in B2B SaaS. The visitor came in to research a specific capability, decided it was worth checking price, and is now evaluating. A rule that fires after 20 seconds on pricing when the previous page was a /help/* URL converts at roughly double the rate of the generic pricing dwell rule.
The message should reference what they were reading: "Saw you were reading about SLA policies — they're on Growth and up. Want me to walk through which plan fits?" This requires the chat widget to pass the previous-page URL as context, which most tools support.
7. Trial signup page dwell over 60 seconds without submitting
A visitor who's been on /signup for a full minute without clicking submit is hesitating. Usually it's one of three things: card-required friction (which you should fix separately), uncertainty about which plan to pick, or a missing piece of information about data residency, security, or contracts.
The rule should fire with a message naming the most common blocker for your business: "Quick note — no card needed for the 14-day trial, and you can change plans later." If the most common objection is something else, name that instead. The point is that the proactive message answers a question before the visitor asks it.
8. Logged-in customer on the pricing or billing page
If you have a customer identification mechanism — a JWT, a cookie, an analytics identify call — you can detect when an existing customer hits the pricing page. They're almost always evaluating an upgrade, a downgrade, or a billing question. None of those should route to the same flow as a prospect.
The rule should fire faster (15-20 seconds), the message should acknowledge they're a customer, and the conversation should route to the customer-success or billing group rather than sales. This is where a chat tool that supports identify-API calls earns its keep: with a plan: 'Starter' trait visible to the agent, the conversation starts already-contextualized.
9. Stuck-state on a form: dwell + no field changes
The subtlest rule in the portfolio. If a visitor opens a form (signup, demo request, contact) and there's no field-change activity for 30+ seconds, they're stuck. Maybe a required field is unclear. Maybe a validation error is confusing. Maybe they're waiting for permission from someone on Slack.
This rule needs custom form-tracking events but produces conversations with extremely high conversion intent. Message phrasing should be diagnostic, not promotional: "Any field giving you trouble? Happy to fill in the blanks on a call instead."
How the nine rules compare
| Rule | Trigger signal | Relative fire volume | Relative accept rate |
|---|---|---|---|
| Pricing dwell 45s | High | Highest | Medium |
| Booking page 2nd visit | Medium | Low | Very high |
| Comparison page 30s | Medium | Medium | High |
| Scroll + dwell | High | High | Medium |
| Exit-intent on booking | Very high | Low | Highest |
| KB → pricing path | High | Medium | High |
| Signup dwell 60s | High | Low | High |
| Logged-in on pricing | High | Low | High |
| Form stuck-state | Very high | Very low | Very high |
The pattern: low-volume, high-accept rules are where the conversion lift hides. A team running only "pricing dwell" is leaving most of the table.
How Helptal fits in
Helptal's proactive chat rules are configured under /settings/proactive-rules using pageUrlContains and dwellSecondsMin conditions — the two primitives every rule on this list is built from. Each rule fires once per visitor row, respects Do Not Track, and lands the conversation in the same unified inbox as your email and web-form tickets. For the rules that depend on customer identity (#8) or custom traits, the Helptal.identify() API attaches plan, MRR, and any custom fields to the live-visitor row so the conversation starts pre-contextualized. The live-visitor panel shows current page, dwell, and country before a chat opens, so agents can also fire manual proactive chats on rows that don't match an automated rule yet.
Frequently asked questions
What's the best dwell time threshold for a proactive chat rule?
It depends on the page. For pricing pages, 45 seconds catches engaged readers without firing on scanners. For booking and signup pages, 60 seconds works because forms take longer to read. For comparison or alternatives pages, 30 seconds is enough. Test these against your own analytics — the right threshold is roughly the median dwell among visitors who eventually convert.
How many proactive chat rules should a B2B SaaS team run at once?
Six to nine narrowly-scoped rules outperform one or two broad ones in almost every case we've seen. Each rule should target a specific page pattern and intent signal, fire at most once per visitor session, and be measured independently by accept rate. Broad rules generate noise; narrow rules generate qualified conversations.
Does proactive chat hurt conversion if it fires too aggressively?
Yes — and measurably. Visitors who dismiss a proactive message are less likely to open the widget later in the same session. The fix isn't to disable proactive chat; it's to tighten the rules so each fire correlates with a real intent signal. A rule that fires on 100 visitors and gets 2 accepts is worse than a rule that fires on 20 visitors and gets 8 accepts.
How do exit-intent triggers work technically?
Exit-intent fires when the visitor's mouse moves at speed toward the top of the browser viewport — toward the address bar, tab bar, or close button. It's a heuristic, not a guaranteed signal, and it doesn't work on mobile. For mobile equivalents, teams typically use a combination of scroll-up motion and dwell time on a conversion page as a proxy.
Should proactive chat rules use AI or just static messages?
Static, context-aware messages outperform AI-generated greetings for the first proactive message. AI is more valuable once the visitor replies — answering their question, qualifying intent, and handing off to a human if needed. The proactive trigger itself is a rule engine job; the response can be agent-assisted or bot-handled depending on volume.
This week, audit your current proactive chat rules and count how many of the nine above you've actually shipped. Most teams find they have one or two — usually the pricing-page dwell rule — and the rest are unbuilt. Pick the two highest-leverage gaps (exit-intent on booking and the second-visit rule are usually the right starting pair) and ship them before adding anything else. If you're evaluating tooling, Helptal's free trial includes the proactive rules engine and live-visitor tracking on every paid plan.