Try Oscar Chat — free popups & forms, AI chat from €60/mo
What roundups get right
Feature coverage is the strong suit of the genre. If a tool trains on your website, hands off to a human, or connects to your helpdesk, the roundup will tell you. Pricing snapshots are usually accurate on the day they’re captured — the base plan price you see is the base plan price you’ll pay.
Roundups also do a fair job of sorting tools by personality: the support-suite incumbents like Intercom and HubSpot, the SMB-friendly all-rounders like Tidio, the sales-leaning tools like Drift, the train-on-your-docs specialists like Chatbase. For a head-to-head look at that specialist model, see our Oscar Chat vs Chatbase comparison. If you’re starting from a blank page, that map is worth having.
The problem isn’t what’s in the lists. It’s what the scoring models leave out.
Miss #1: The price on the page is not the price at scale
Most AI chatbots are not priced like software. They’re priced like utilities. Three models dominate:
- Per-seat pricing — you pay for each human agent with a login. Predictable, but you pay even in quiet months.
- Per-resolution pricing — you pay each time the AI closes a conversation. Feels aligned, until a product launch or a shipping delay triples your “resolutions.”
- Credit-based pricing — you buy message credits and burn them as the bot talks. Cheap at 200 conversations a month. Very different math at 5,000.
Roundup tables almost always show the entry number. None of them run the scenario that matters: your busiest month, not your average one. A store doing €40k in a normal month might do €150k in November. Under a credit or per-resolution model, your chatbot bill scales with your success. Under flat pricing, it doesn’t.
Step 1: Price your worst month, not your first month
Before you trial anything, multiply your peak-season conversation volume by the tool’s per-resolution or credit cost. If that number makes you flinch, the “cheap” plan isn’t cheap.
Step 2: Ask what happens when credits run out mid-month
Does the bot stop answering? Do you get auto-charged? Does a human get paged at 3 a.m.? Roundups never test this. Support docs will tell you — read them before you buy.
Miss #2: Conversion tooling scores zero
Here is the strangest habit in chatbot roundups: they score chatbots only as support tools. Response quality, ticket deflection, CSAT — the whole rubric is about answering questions cheaper.
But on a real store, the chat widget sits next to two other conversion levers: exit-intent popups and lead-capture forms. When those live in separate tools, you pay three subscriptions, your visitor gets three disconnected experiences, and your data lives in three dashboards. When they’re one toolkit, a visitor who abandons a cart sees a popup, leaves an email in a form, and asks a question in chat — and it’s all one journey you can actually measure. For more on timing those moments, see our Popup Display Rules Tutorial: Show the Right Form at the Right Moment.
Roundups treat “does it also do popups and forms?” as a nice-to-have checkbox, if they mention it at all. For a store doing under €1M a year, I’d argue it’s half the value. Support deflection saves you money; lead capture makes you money. A scoring model that only counts the first half will reliably rank expensive support suites above tools built for selling. It also helps to track the right numbers — see 12 Chatbot Metrics That Actually Matter (and 5 Vanity Metrics to Ignore).
Miss #3: “Multilingual” is a checkbox, not a depth test
Almost every tool in every roundup claims multilingual support. What varies wildly is what that means:
- Does the bot detect the visitor’s language automatically, or must the visitor find a settings toggle?
- Are answers generated natively in that language, or machine-translated on the fly (you can tell — your customers definitely can)?
- Does it handle the messy reality of a Spanish question about a product described in English in your catalog?
If you sell cross-border — and most European stores do, whether they planned to or not — this is the difference between a chatbot that opens new markets and one that produces apologetic reviews. A checkbox in a feature table can’t tell you. Ten minutes of testing in your second-biggest market language can. Related reading: Multilingual AI Chat: How to Sell in 50+ Languages Without Hiring Translators.
Miss #4: What “free plan” actually means
The free tier is where roundups are most generous and least precise. “Free plan available: ✅” hides four very different realities:
- Conversation caps — free until 50 or 100 chats a month, then the widget quietly stops.
- Branding requirements — “Powered by” badge on your store, on the free tier, forever.
- Data limits — the bot can only train on a handful of pages or documents, so it answers confidently and wrongly.
- Feature locks — the integration you actually need (your CRM, your email platform) sits behind the first paid tier.
None of this makes a free plan bad. It makes it a trial with extra steps. The honest question is never “is there a free plan?” — it’s “what, exactly, turns off, and when?” We break down a classic free option in our Oscar Chat vs Tawk.to comparison, including where the free tier stops working.
This is also why we made a specific choice at Oscar Chat: the popup builder and the form builder are completely free — not a capped trial, not a teaser. The AI chat is the paid product (flat €60/mo per seat). If a store only needs popups and forms to capture leads, they should cost €0. We’d rather earn the upgrade when the chat volume justifies it.
So how should you actually shortlist? By buyer type
Forget “best overall.” Here’s the frame I’d use, based on who you are:
If you’re cutting support costs: You live in tickets-per-day math. Per-resolution pricing can work if your volume is stable. Prioritize helpdesk integrations and handoff quality; test deflection on your 20 most common questions before committing.
If you’re a conversion-focused store: Weight lead capture as heavily as chat. Ask: can this tool run my exit-intent offer, capture the email, and answer the product question in one flow? Are popups and forms free or a second subscription? Is pricing flat, so a great Black Friday doesn’t come with a great invoice?
If you’re an agency or run multiple stores: Look for workspaces — separate sites, teams, and widget sets under one roof — and per-seat pricing you can pass through to clients without a spreadsheet. Most roundups barely test this; it’s the whole game for you.
Our own honest placement: Oscar Chat is the flat-price + conversion-toolkit pick. AI chat at a flat €60/mo per seat in 50+ languages, popups and forms free, integrations with Shopify, Klaviyo, Salesforce, and Zoho, and workspaces built for multi-store teams. If you need a deep enterprise helpdesk suite, the incumbents are ahead of us and the roundups are right to say so. If you need a store to sell while it supports, that’s the gap we built for.
Frequently asked questions
Why do roundups keep recommending the same big tools?
Partly quality, partly mechanics. Large vendors have partner programs, review pipelines, and brand recognition that make them easy to include. Smaller tools often enter lists only after someone pitches them — which, full disclosure, is partly why articles like this exist.
Is per-message (credit-based) pricing ever a good deal?
Yes — at low, stable volume. If you handle under a few hundred AI conversations a month and your traffic doesn’t spike seasonally, credits can be the cheapest option. The trap is success: the model taxes growth. Run your peak-month numbers before deciding.
Do popups and forms really belong in a chatbot comparison?
If the comparison is “best support chatbot,” arguably not. If it’s “best AI tool for a store’s website” — which is what most merchants are actually shopping for — yes. The widget, the exit-intent popup, and the lead form are one visitor experience. Scoring them in isolation misses how stores actually convert.
How can I test multilingual quality before buying?
Write your 10 most common customer questions in your second-largest market’s language — including one with a typo and one mixing two languages, because that’s how real customers write. Run them against the bot trained on your site. If the answers read like they were written by a person who speaks the language, you’re fine. If they read like subtitles, keep looking.
What does Oscar Chat cost, exactly?
The popup builder and form builder are free. The AI chat is a flat €60/mo per seat — no per-message credits, no per-resolution charges. One seat can serve customers in 50+ languages, and workspaces let agencies separate multiple stores and teams.
I’m an editor maintaining a roundup. How do tools get added?
Usually you pitch them — like we do. The good lists re-test hands-on, so give editors something testable: a live store demo, real pricing math at different volumes, and answers in more than one language. If you’re writing one of these roundups and want to test Oscar Chat, our door is open.
The short version
Roundups are a good map and a bad compass. Use them to build your longlist, then ignore their scores and test the three things they can’t measure for you: your peak-month price, your second language, and your full visitor journey from popup to form to chat. The right tool is the one that’s still the right price in your best month — not just your first one.
— Kseniia Shevchenko, Founder, Oscar Chat
