SaaS Onboarding Chatbots

Route the Trial, Not the Tour, and Price the Bot Before Day 14

Quick answer: The best evidence on what a SaaS onboarding bot should do is a survey, not a vendor page. In January 2026, 200 self-serve software products told ChartMogul, ProductLed and Kyle Poyar how their free users convert. The median free-to-paid rate was 8 percent within six months, and the report's own headline is that "very few products actually have an 8% conversion rate": one in five trial products converts under 2.5 percent, roughly one in four converts above 25 percent. Trials that ask for a credit card convert at 30 percent, more than five times the no-card rate, and sign up fewer people (35 per 1,000 visitors against 45). That spread is the design brief. A bot cannot change your trial model, but the trial model decides what the bot is for: on a no-card trial it exists to get a user to the activation event before day 14; on a card trial it exists to keep the cancellation honest; on freemium it exists to find the few accounts worth a human's time. The second number is activation: Userpilot's 2024 benchmark across 62 B2B companies put the average at 37.5 percent and the median at 37 percent, so most of the people your bot greets never reach the moment the product is for. This guide is the plan for the 14 days between signup and the decision, by day and by surface, with the qualification branch priced against Intercom's outcome meter and a worked bill for a 1,000-signup month.

Onboarding is the one word in SaaS that means two products. The first is the guided first session, the activation moment and progressive disclosure, which our onboarding flow design guide already covers for any bot on any channel and which this page does not restate. The second is the trial as a unit: fourteen days, a user who has not paid, a product that may or may not have shown its value, and a set of messages that arrive in the product, in the inbox and in a chat window. This guide is about the second one, and it treats the bot as the thing that decides which message a trial user gets, when, on which surface, and when to stop.

Step 1. Read the conversion data as a spec, not as a target

The ChartMogul report is worth reading in full because the shape of the data matters more than any single number. Of the 200 products, 57 percent land new users in a free trial and 26 percent in freemium; 7 percent run a reverse trial, temporary premium access on a free plan. The most common trial length is 14 days (62 percent of trial products), then 7 and 30 days at 14 percent each. Among trial products, 20 percent require a card. And 80 percent of trial products add human touchpoints when an enterprise user self-serves in, which is the first thing the data says about a bot: the bot is not the whole onboarding for the accounts that matter most, it is the router that decides which ones a person should call.

The second thing is the distribution. For trial products, 20 percent convert under 2.5 percent, 30 percent land between 2.5 and 7.5 percent, and 23 percent convert above 25 percent. Freemium is similar at the bottom (25 percent under 2.5 percent) and clusters at 10 to 15 percent for another quarter. The report's "good and great" ranges (50th and 75th percentile) are in the step 2 table. If your trial converts at 5 percent you are not at 60 percent of the median, you are in the bottom half of a bimodal distribution, and the first job is to find out which of the two humps you belong to.

The third thing is the report's 1,000-visitor path, which is the argument against copying someone else's model. Freemium products see 90 signups and 5 paying customers; free trials see 45 signups and 3.6 paying customers; ungated freemium (use it before you create an account) sees 70 and 5.6; card-required trials see 35 and 10.5. Per signup, 30 percent is about 3.75 times the rate for trials as a whole (8 percent, a figure that includes the card trials) and more than five times the no-card median of 4 to 6 percent; per visitor the card trial produces about 2.9 times the customers of the average trial (10.5 against 3.6). The bot inherits whichever trade you have made.

Then activation. Userpilot's 2024 benchmark report, built on first-party data from 547 SaaS companies, reports the user activation rate from a 62-company subset measured in its own dashboard: average 37.5 percent, median 37 percent, and Lenny Rachitsky's separate report at 36 percent, which the authors take as confirmation. Product-led companies activated 34.6 percent of new users against 41.6 percent for sales-led ones, and the report's explanation is the one that matters for a bot: a user who has already paid wants the value now, while a trial user "sometimes ha[s] no intention of purchasing the product at all." Time to value was about a day and a half in both models (1 day 11 hours sales-led, 1 day 12 hours product-led); the checklist-completion figure from the same report is in the in-app messaging entry.

Put the two reports together and the spec writes itself. Roughly six in ten signups never reach the activation event. Of the four in ten who do, some fraction converts, and that fraction depends on the trial model more than on anything the bot says. So the bot has two jobs, in order: move the activation rate, then sort the activated users into the ones who will pay on their own, the ones a person should call, and the ones to leave alone.

Step 2. Decide the bot's job from the trial model, because the model already decided

This is the conditional part of the guide, and it is the part most "SaaS chatbot" pages skip. The trial model sets what a bot can move.

If you run a no-card 14-day trial, the bot is an activation engine and everything in step 3 applies. If you require a card, the bot's most valuable message is the reminder before the charge, and ChartMogul names why: Canva tells trial users they will get a reminder seven days before the charge, and the report ties silence at that moment to fewer signups, worse month 1-3 retention and higher refunds. If you are freemium, the bot should fire on limits and account signals, not on the calendar, and the qualification branch in step 4 is the whole product. If you are ungated, the bot's welcome is the account-creation prompt, and the welcome message design guide's first rule, one question and a reason, applies to that prompt rather than to a greeting.

Step 3. The 14-day plan, by day, by surface, by what the bot is allowed to say

The plan below is for the majority case, a 14-day no-card trial, on a product with a website app and a chat widget. Days are counted from signup. "In-app" means a message rendered inside the product (a post, banner, tooltip or checklist, per the in-app messaging entry); "chat" means a message in the widget conversation the bot can continue; "email" is email. The surface matters as much as the copy, and it also has a price, which step 6 works out.

Three rules hold the table together. First, every message after day 0 is conditional on an event, not on a day: the day is when the bot checks the event, not when it speaks. Userpilot's 2026 conversion piece puts the pre-expiry survey at "the 60-70% mark of the trial window" and the behavioral split ("a user who has logged in 12 times, completed the activation checklist, and invited two colleagues is a fundamentally different prospect from someone who logged in once at signup") is exactly the branching a bot is for. Second, the in-app surfaces are for the moment the user is present and the email is for the moment they are not; a post shown to a user who is mid-task is an interruption, which is why the day-2 and day-10 messages sit in chat rather than in a modal. Third, the plan has more stop rules than send rules. A trial user who has ignored two messages has told you something, and the drip campaign entry's rule that a sequence needs an exit for every branch applies here with the trial's own end as the hardest exit.

What the bot should say is shorter than what it should know. Day 0's question is a multi-turn form of length one. Day 2's offer to "do it with you" is a bot that can take an action in the product, which is where an AI agent with a tool is different from a scripted flow; if yours cannot act, say "show you" only. Day 10's plan-fit message uses numbers from the account (seats, records, messages sent) and nothing else; a plan recommendation that ignores the account's own usage is a pricing page in a chat window. And the day-7 survey is the highest-value message in the table because its answer routes everything after it: "approval" means send the one-page summary a manager can forward; "price" means the day-13 message shows the cheapest plan that fits; "missing feature" means a human reads the free text and decides.

Step 4. The qualification branch, and what a qualified user costs

Not every activated trial is worth a call, and calling the wrong ones is the most expensive mistake an onboarding program can make. The branch is: the bot decides, from behavior and answers, whether a trial user is (a) going to convert on their own, (b) worth a person's time, or (c) neither. Userpilot's reading of ProductLed's survey of more than 600 B2B companies gives the size of the prize: free-to-paid averages about 9 percent across models, while companies using product-qualified leads to find high-intent trial users convert those at roughly 25 percent, and only 24 percent of companies use them. We did not read the ProductLed survey itself and report those figures as Userpilot reports them.

A product-qualified lead is a trial user whose behavior in the product, not their form answers, says they are likely to pay: activation reached, a second person invited, a limit approached, a billing page viewed, an integration connected. Our lead scoring entry has the mechanics; the SaaS-specific rule is that the score should be built from three or four product events, weighted by how strongly each correlates with conversion in your own history, and that the bot should act on the score only at two points, day 2 (activated) and day 10 (plan fit). The day-7 survey adds one non-behavioral signal, and it is the one that catches the accounts behavior misses: "approval" from a user who has done everything right is the enterprise account ChartMogul's 80 percent human-touchpoint figure is about.

What a qualified user costs depends on who does the qualifying. On Intercom, whose pricing we read on 4 September 2026 and whose Fin AI Agent outcomes article (dated 30 July 2026) is the definitive text, Fin for Sales charges $9.99 for a qualification and $0.99 for a disqualification, and a qualification is counted when Fin "matches a prospect to the qualification criteria you defined and routes them accordingly," with three named routing outcomes: book a call with sales, start a trial, or route to self-serve onboarding, "each billed at $9.99." Read that last clause twice before switching Fin for Sales on for trial users. Routing a prospect to self-serve onboarding is a qualification, at $9.99, which for a trial-onboarding bot means the most common outcome is the priced one. A support resolution is $0.99, charged once per conversation, including an "assumed resolution" where the customer leaves after an answer without asking for more; an escalation Fin makes on its own frustration logic or your workspace rules is not charged; an abandoned conversation after a clarifying question is not charged.

The design consequence: keep the qualification decision in your own scoring, and use the bot's paid qualification only for the two routes that cost money on your side too, a booked call and a sales hand-off. Our Intercom review recorded the $0.99 per-outcome meter and the seat prices in May 2026; the $9.99 qualification outcome is in Intercom's help center as of the 30 July 2026 revision and is not recorded in that review. The lead qualification playbook has the generic funnel; the SaaS difference is that the product is the form.

Step 5. Five hand-off rules to write before the first message

1. Billing questions go to a person, or to a page, never to a guess. A trial user asking what happens to their data at day 14, whether the charge is monthly, or how to cancel gets the exact policy text with a link and, if the knowledge base does not have that text, a human hand-off. An AI answer about billing that turns out wrong is a refund and a review.
2. Cancel means cancel. A user who asks to cancel gets the cancel path in one click, then optionally one question about why. Not a retention sequence, not a discount unless a person offers it. ChartMogul's note on card trials is that the companies that do this well "earn buyer trust by providing transparency about how and when users can cancel," and the ones that do not see worse retention and higher refunds.
3. Frequency has a ceiling and the ceiling is per user, across surfaces. One in-app interruption per session, one email per day, and never both about the same thing on the same day. The Intercom meter counts a post shown on web and on mobile as two "Messages Sent"; the user counts them as two interruptions.
4. The bot does not block the product. No modal a user cannot dismiss, no checklist that hides the workspace, no tour that must be finished. Intercom's availability table is a reminder that tours do not run on mobile web or in native apps at all, so a plan that depends on a tour has no plan for a phone.
5. A frustrated trial user is a hand-off, not a resolution. If the bot detects frustration, or the user asks for a person, the conversation goes to the inbox with the account's trial day, activation state and survey answer attached, the handoff design guide's context rule. On Intercom that escalation is unbilled; on any platform it is the cheapest message in the plan.

Step 6. The bill for a 1,000-signup month, worked on Intercom's public prices

Numbers first, then the assumptions, all of which are ours. Suppose 1,000 new trial signups a month, a web app with a chat widget, the step-3 plan, and Intercom, the one platform in our review catalog with every surface the plan uses (pricing page and help center read 4 September 2026).

Seats. The pricing page lists Essential at $29 per seat per month, Advanced at $85 and Expert at $132 under a toggle set to annual billing; its help-center FAQ says the Essential figure is "on annual billing" and, in a seat-count example that does not name the billing period, prices Essential at "$39 x 3 = $117." Since $29 is the annual figure, we read $39 as the monthly-billed rate, which is the one our pricing methodology uses, so two Essential seats for the people who take the hand-offs are $78 a month. The monthly-billed rates for Advanced and Expert do not appear in the text we read.

In-app surfaces. The pricing page states that "all plans include free, unlimited live chat, support email, in-app chats, banners, and tooltips." Checklists, product tours, posts, surveys and Series need the Proactive Support Plus add-on at $99 a month, which includes 500 "Messages Sent" and then charges per message on a sliding scale ($0.07 for messages 501 to 1,000, $0.06 to 2,000, $0.05 to 3,000, down to $0.0175 above 250,000). The day-0 post to 1,000 users and the day-7 survey to the 600 who log in again are 1,600 Messages Sent: 500 free, 500 at $0.07 ($35), 600 at $0.06 ($36), so $71 in usage, $170 for the add-on with its usage included. Checklists and banners cost nothing extra. (Intercom's Pricing FAQ lists Series among the add-on's included features while the add-on article meters it; we follow the add-on article.)

Email. Outbound email is metered separately as "Bulk Email Sent" on an eleven-tier scale ($0.045 for the first 100, $0.03 to 500, $0.015 to 1,000, $0.0075 to 5,000, and down). The plan's email days (1, 5, 10 or 11, 13, 21 and 45) reach different subsets of the cohort; if they add up to about 4,000 sends a month, the bill is $4.50 + $12 + $7.50 + $22.50, $46.50. Emails sent from the inbox in reply to a user are free. The plan uses no SMS, WhatsApp or phone, each of which has its own meter.

The bot. If 250 of the 1,000 open a conversation with Fin and 175 end in a resolution or a procedure hand-off, that is $173.25 at $0.99. If you switch on Fin for Sales and 40 users are qualified at $9.99, add $399.60.

Totals: with Fin for Sales, about $867 a month ($78 + $170 + $46.50 + $173.25 + $399.60); without it, about $468; with chats, banners and tooltips only, no add-on, plus the bot and the emails, about $298. Every usage line is billed in arrears, so the first invoice is smaller than the second.

OneSignal and Firebase meter the in-app half differently and include no bot; the in-app messaging entry works the same 1,000-user month on all three. What the totals above show is that the conversational column of the plan (day 0's question, day 2's offer, day 10's plan fit, every hand-off) only exists on a product with a bot, and the bot's meter is per outcome: if your trial users mostly never open the chat, the add-on is the cost; if they do, budget the outcomes from your own conversation rate, not from ours.

What our fifteen reviews record about SaaS onboarding

Fourteen of the fifteen platform reviews use the word "SaaS," almost always to describe the vendor's own business model, and one, Intercom, describes in-app messaging as a capability: its proactive engagement dimension scores product tours, in-app notifications, in-app messaging and targeted message sequences at 4 out of 5, its hands-on section captured the setup checklist a fresh trial workspace opens with ("connect Messenger, import contacts, invite teammates, configure Fin AI") and the Outbound content view with per-campaign reach, and it recorded the $0.99 outcome meter and the annual-billed seat prices this guide re-reads. Four reviews mention a product tour or an onboarding checklist; in three (Blip, Tidio, Wati) it is the vendor's own onboarding of the reviewer, and only Intercom's names tours as a feature the buyer can build. Eleven mention a free trial or free tier as a plan feature: Tidio's 7-day no-card trial of paid features on top of a permanent free tier is, in our review's words, a more accommodating entry path than Chatfuel's, and Chatbase's eight-minute setup on its free tier is the fastest measured time-to-first-bot in our Tier 1 batch (Landbot's 8 to 12 minutes is an anchored projection, not a measurement). No review built a trial-conversion flow, measured a trial's activation rate, or tested a qualification hand-off, and the search strings in Sources reproduce that absence.

What transfers from the corpus is the chat half of the plan. The hallucination and citation figures from Scenario D of our methodology say which platforms can be trusted with the billing-policy answer in rule 1 (the ones that cite a source per answer), and the Scenario E handoff scores say which can carry the account context rule 5 needs. For the in-app half (posts, banners, checklists) Intercom is the only reviewed platform with the full set of formats, and the products that specialize in it (Userpilot, Appcues, Pendo) are outside our catalog and are not ranked here.

FAQ

What should a SaaS onboarding chatbot actually do?

Three things, in order: get the trial user to the activation event before the trial ends, sort activated users into those who will pay alone, those worth a call and those to leave alone, and hand billing, cancellation and frustration to a person. What it should not do is tour the product. The activation event is the spec, and it has to be named from your analytics before the first message is written.

What is a good trial-to-paid conversion rate in 2026?

From ChartMogul's January 2026 survey of 200 self-serve products: 4 to 6 percent is good and 10 to 15 percent is great for a no-card free trial; 25 to 35 percent good and 50 to 60 great when a card is required; 3 to 5 good and 8 to 12 great for freemium. The median across everything is 8 percent, and the report's point is that almost no product sits there.

Should the bot message trial users by day or by behavior?

By behavior, checked on a schedule. Every message after the welcome is conditional on whether the activation event has happened, whether the user came back, and what they answered on the day-7 survey. A day-13 "your trial ends tomorrow" email to a user who never set the product up is the message most likely to be marked spam.

Is a chatbot the same as in-app messaging?

No. In-app messaging is the set of formats a product shows inside itself (posts, banners, tooltips, checklists, carousels) and most in-app messaging products send without conversing. A chatbot holds a conversation and can act. The plan in this guide uses both, and prices them separately because the vendors do; the in-app messaging entry has the formats and meters.

How much does Intercom charge to onboard trial users?

Five meters, read 4 September 2026: seats ($39 a month for Essential billed monthly), the $99 Proactive Support Plus add-on for checklists, tours, posts and surveys, outbound email from $0.045 a send, Fin at $0.99 per outcome, and Fin for Sales at $9.99 per qualification. Step 6 works them through: a 1,000-signup month on this guide's plan runs roughly $298 to $867 depending on which you switch on, before any SMS, WhatsApp or phone usage.

What is a product-qualified lead?

A trial or free user whose behavior in the product (activation, a second seat invited, a limit approached, a billing page viewed) indicates they are likely to pay. It is the SaaS version of lead scoring with the product as the form, and ProductLed's survey, as reported by Userpilot, has companies that use them converting those users at roughly 25 percent against a 9 percent average.

Should a card-required trial use a bot at all?

Yes, for a different job. The activation work is the same, but the most valuable message becomes the reminder before the charge, in plain words, with the cancel path in it. ChartMogul ties silence at that moment to worse retention and more refunds. A bot that hides the cancel path to lift conversion is optimizing the wrong month.

Which platform in your reviews is best for SaaS onboarding?

Our reviews cannot say, and this guide does not rank one. Intercom is the only reviewed platform with the in-app surfaces (posts, banners, checklists, tours) and it is priced for support teams; the specialist in-app products are outside our catalog. For the chat half of the plan, the reviews' Scenario D citation and Scenario E handoff results are the figures to shortlist on.

Related guides

• Chatbot onboarding flow design guide — the first session: the activation moment, one step at a time, minimum data, the human exit. This guide assumes it.
• In-app messaging — the formats a product can show inside itself, who meters them how, and the five numbers to hold them to.
• Chatbot welcome message design — the day-0 question, and why it is one question.
• Lead scoring — the mechanics the product-qualified-lead score in step 4 borrows.
• Lead qualification playbook — the generic inquiry funnel this guide specializes for a trial.
• Drip campaign guide — timed sequences, exits for every branch, and the frequency ceiling.
• Chatbot handoff design guide — what travels with the conversation when rule 5 fires.
• Chatbot metrics guide — how to instrument activation, conversation rate and hand-off rate so the bill in step 6 is yours, not ours.

Sources

• ChartMogul, Growth Unhinged and ProductLed, The SaaS Conversion Report: A new look at free-to-paid conversion (chartmogul.com/reports/saas-conversion-report), by Kyle Poyar, survey conducted January 2026 with 200 responses, read in full on 4 September 2026. Quoted or reported: the six key stats (57 percent free trial, 26 percent freemium, 7 percent reverse trial; 14-day trials at 62 percent, 7 and 30 days at 14 percent each; 20 percent of trials require a card; median free-to-paid 8 percent and "very few products actually have an 8% conversion rate"; card trials at 30 percent, "more than 5x"); the 80 percent human-touchpoint figure for trial products; the distribution passages (trial: 20 percent below 2.5 percent, 30 percent between 2.5 and 7.5, 23 percent above 25; freemium: 25 percent below 2.5, 29 percent between 2.5 and 7.5, 25 percent between 10 and 15); the good-and-great table (freemium 3-5/8-12, ungated freemium 7-9/8-12, no-card trial 4-6/10-15, reverse trial 4-6/8-12 "not statistically significant," card trial 25-35/50-60, SaaS 5-7/12-16, AI 6-8/15-20, B2C or hybrid 5-7/8-12, B2B 6-10/15-20); the 1,000-visitor path (freemium 90 signups and 5 paying, all trial products 45 and 3.6, ungated 70 and 5.6, card trial 35 and 10.5); the Canva seven-day reminder example and the sentence on companies that "earn buyer trust by providing transparency about how and when users can cancel"; the three closing questions; and the methodology (Typeform survey; conversion defined as "the percentage of leads or free signups that convert to become a paying customer within six months"; "a typical respondent is between $1 to $10 million ARR").
• Userpilot, Product Metrics Benchmark Report 2024: 6 Key Insights for Product Teams (userpilot.com/blog/product-metrics-benchmark-report, published 16 June 2024, last updated 7 April 2026) and User Activation Rate Benchmark Report 2024 (userpilot.com/blog/user-activation-rate-benchmark-report-2024), both read 4 September 2026. Reported: the 547-company dataset; the activation figures from "62 B2B companies collected with Userpilot's New User Activation dashboard" (average 37.5 percent, median 37 percent; Lenny Rachitsky's 36 percent cited as consistent); PLG 34.6 percent against SLG 41.6 percent and the explanation quoted in step 1; time to value 1 day 11 hours against 1 day 12 hours; average checklist completion 19.2 percent; the 80 percent multimedia finding among companies above 50 percent activation; 1-month retention 48.4 percent PLG against 39.1 percent SLG. Userpilot sells onboarding software and the reports are marketing content built on its customers' data; we report them as the vendor's figures.
• Userpilot, Are the "SaaS Average Conversion Rates Benchmarks" Real in The Post-AI Economy? (userpilot.com/blog/saas-average-conversion-rate, published 10 June 2026, last updated 19 August 2026), read 4 September 2026: the survey timing passage ("typically at the 60-70% mark of the trial window") and the behavioral-segment sentence quoted in step 3; and its report of ProductLed's survey of 600+ B2B companies (free-to-paid "averages 9% across models," PQL-driven conversion "around 25%," 30 to 39 percent at $1k to $5k ACV, 24 percent of companies using PQLs), which we did not read at source and attribute as reported.
• Intercom, Pricing (intercom.com/pricing), read 4 September 2026: Essential $29, Advanced $85 and Expert $132 per seat per month as displayed, Fin AI Agent "From $0.99 per Fin outcome," the sentence "All plans include free, unlimited live chat, support email, in-app chats, banners, and tooltips. Pay-as-you-go for email campaigns, SMS, WhatsApp, and Phone," the Proactive Support Plus add-on card ("Posts, Checklists, Product Tours, Surveys, and Series campaign builder," $99/mo), and the billed-annually/billed-monthly toggle. Intercom Help, Pricing FAQs (article 8344190, "Updated over a week ago" on the read date), read 4 September 2026: "the Essential plan starts at $29/seat/month on annual billing," the Essential seat-count example "$39 x 3 = $117," and the statement that you "can also choose to pay monthly on the Essential and Advanced plans." Intercom Help, Usage-based channels (article 9061703, "Updated yesterday" on the read date), read 4 September 2026: the free channels (Chats, Banners, Tooltips, Inbound WhatsApp Conversations), the "Bulk Email Sent" definition and eleven-tier table ($0.045 to 100, $0.03 to 500, $0.015 to 1,000, $0.0075 to 5,000, $0.005 to 10,000, $0.003 to 50,000, and down to $0.00025), the $39-for-3,000-emails example, the sentence "Inbound emails or emails sent from the Inbox are free to use at no additional cost," and the "billed ahead for your seats and add-ons, and in arrears for usage-based charges" rule. Intercom Help, Fin AI Agent outcomes (article 8205718, dated 30 July 2026): the outcome table ($0.99 resolution, $0.99 procedure handoff, $0.99 disqualification, $9.99 qualification), "one outcome per conversation," the assumed-resolution definition, the three qualification routing outcomes and the note that "routing a prospect to self-serve counts as a qualification," the unbilled escalation and abandonment rows, and the 14-day Fin trial. Intercom Help, Proactive Support Plus add-on (article 9061648, dated 21 May 2026): the "$99/month (includes 500 'Messages Sent')" base fee, the included features (Checklists, News, A/B testing, webhooks, versioning, event-based messaging), the metered features (Posts, Mobile Push, Product Tours, Mobile Carousels, Surveys, Series), the Messages Sent definitions including the web-and-mobile double count, the nine-tier price table ($0.07, $0.06, $0.05, $0.04, $0.03, $0.025, $0.02, $0.0175) and the $145 carousel example. Intercom Help, Outbound explained (article 3292835, dated 17 April 2026): the message-type list and the omnichannel availability table (product tour desktop web only; banner, checklist, news item and tooltip not supported in iOS/Android apps; post and survey supported everywhere). The step-6 arithmetic is ours: 2 × $39 = $78; 1,600 Messages Sent = 500 × $0 + 500 × $0.07 + 600 × $0.06 = $71; 4,000 bulk emails = 100 × $0.045 + 400 × $0.03 + 500 × $0.015 + 3,000 × $0.0075 = $46.50; 175 × $0.99 = $173.25; 40 × $9.99 = $399.60; totals $867.35, $467.75 and $297.75, rounded in the body.
• OneSignal, Pricing (onesignal.com/pricing) and Firebase, Firebase In-App Messaging (firebase.google.com/docs/in-app-messaging, "Last updated 2026-09-01 UTC"), both read 4 September 2026: the Free-plan in-app allowance "for organizations with up to 1,000 monthly active users," the Growth plan's "$19/mo" and "$0.012 per monthly active user," and the absence of a price on the Firebase page. The full read of both, with quotations, is in the in-app messaging entry's Sources.
• Chatbotscape review corpus (the fifteen platform reviews listed at /reviews), searched 4 September 2026 from the repository root. Denominator: ls sample-reviews/*-review.md | wc -l returns 15. grep -liE '\bsaas\b' sample-reviews/*-review.md returns 14 (all but blip); grep -liE 'in-app messag|in app messag' sample-reviews/*-review.md returns 1 (intercom); grep -liE 'product tour|onboarding checklist|setup checklist' sample-reviews/*-review.md returns 4 (blip, intercom, tidio, wati; a looser 'product tour|checklist' returns 9 because two files match only on an internal "verification checklist" note); grep -liE 'free trial' sample-reviews/*-review.md returns 11; grep -liE 'trial-to-paid|trial to paid|product.qualified|activation rate' sample-reviews/*-review.md returns 0. Passages cited: intercom-review.md lines 198-199 (proactive engagement, 4/5), lines 538-539 (the trial workspace checklist caption), lines 612-625 (Outbound content view and Series editor captions), line 97 (seat prices and the $0.99 outcome meter); tidio-review.md line 320 (7-day no-card trial on top of the Free tier); chatbase-review.md line 412 (setup speed, 8 minutes). The statement that no review tested a trial-conversion flow is an argument from absence across the fifteen files and is stated that way in the body.
• Ahrefs Keywords Explorer, US overview and volume-by-country, queried 4 September 2026 — the demand, difficulty, CPC, global-volume and parent-topic figures in this page's keyword note.
• Chatbotscape evaluation methodology, including the six-scenario protocol under which the Scenario D citation and Scenario E handoff figures referenced in the corpus section were recorded. /methodology (continuously updated).

About this guide

Chatbotscape launched in 2026 as an independent review site for chatbot platforms. This guide is part of our SMB chatbot Academy and is written for the founder, product lead or growth owner of a self-serve software product deciding what a bot and its outbound messages should do during a trial. It names no best platform for SaaS onboarding, because our reviews have not tested one, and it reads the two conversion reports and the three vendors' pricing pages it is built on rather than summaries of them.

Methodology

The ChartMogul report, the three Userpilot articles and the Intercom, OneSignal and Firebase pages were read in full on the day of drafting, and every figure carries the page, table or article it came from so that a reader can check it. The corpus figures are our own published measurements and anchored assessments, with the evidence class of each stated in Sources. Every corpus search was executed in the form the page states. No SaaS trial was run for this guide and no conversion figure on it is a measurement of ours.

The editorial judgments on this page, listed rather than flagged line by line:

1. Reading a survey of 200 self-serve products, where a typical respondent was $1M to $10M ARR, as the design spec for a bot on any SaaS trial. The distribution and the model-level ranges are what we transfer; no individual product's rate is.
2. Treating Userpilot's activation and checklist figures, which come from that vendor's own customer base and are published as marketing, as the best available benchmark, and attributing the ProductLed PQL figures through Userpilot's report of them rather than reading the survey. Lenny Rachitsky's independent 36 percent is the only cross-check we have.
3. Building the 14-day table for a no-card trial and describing the other four models by difference. The day assignments and the stop rules are ours; the day-7 survey timing follows Userpilot's "60-70% mark."
4. Recommending that qualification be decided in the buyer's own scoring rather than by a metered vendor outcome, and making cancellation a one-click hand-off rather than a retention sequence. The first is a cost judgment drawn from Intercom's help text, not a finding about Fin's accuracy; the second is an editorial rule supported by the ChartMogul retention-and-refund observation.
5. Pricing Essential seats at the monthly-billed $39 from a worked example in Intercom's FAQ rather than the $29 annual figure the pricing page displays, and stating that the monthly figures for Advanced and Expert were not in the text we read.
6. Publishing the corpus gap (fourteen mentions of SaaS as a business model, one in-app capability, zero trial tests) instead of a shortlist.

See our methodology for how platform facts are verified.

Last updated

5 September 2026.