Every score on this site is the output of a published rubric applied to published facts. This page shows the whole mechanism, so you can check our work, disagree with a weight, or re-score a product for your own situation.
What a score means
A score out of 10 answers one question: going by public documentation, how good a deal is this product for a small team that wants live chat on its website? It does not tell you whether the agent inbox is pleasant to use or whether the widget is fast on your site. Only a trial on your own site can tell you that, and the guide to widget speed explains how to measure it yourself.
Each product gets a level from 1 to 5 on six criteria. A level is not an opinion: each one has a written rule, printed below, and every product's levels are listed at the bottom of this page. The score is the weighted average of the levels, scaled to 10. Level 1 on everything gives 2.0; level 5 on everything gives 10.
The rubric: criteria, weights and level rules
Half of the score is about money: what the free plan really allows, what the bill becomes when a second, fifth and tenth person joins, and how many people the entry price already covers. That is deliberate. For a small team those questions decide more than any feature list.
What changed in v2.0. On 20 September 2026 the rubric gained a sixth criterion, seats included at the entry price, and the weights were flattened so that no single criterion decides a ranking. Rubric v1.0 measured what a team pays per seat but never asked how many seats the entry price already covers, which is the first wall a growing team hits. No product was re-scored to make room for it: the levels on the original five criteria are unchanged, and the new criterion is read from the same published plan data that the cost arithmetic uses. The effect is that products priced per workspace moved up and per-seat products moved down.
Free-plan usability — 15%
Going by what the vendor documents, could a small team run real support on the free plan?
| Level | What it means |
|---|---|
| 5/5 | A free plan with no documented cap on seats or conversations, where vendor branding can be removed without moving every agent to a paid seat. |
| 4/5 | A free plan with no documented cap on conversations and at least two seats, but with branding that stays or a short chat history. |
| 3/5 | A free plan a very small team could use, with documented caps on seats or conversations and vendor branding that stays on the widget. |
| 2/5 | A free plan exists, but it is documented as a single seat or fewer than 100 conversations a month, so it works as a demo rather than a support channel. |
| 1/5 | No free plan. At most a time-limited trial. |
Cost as you add agents — 15%
What is the cheapest published route for 5 people and 500 conversations a month?
| Level | What it means |
|---|---|
| 5/5 | 50 a month or less. A free plan counts as 0 only where the vendor documents room for both 5 people and 500 conversations; an undocumented cap does not count. |
| 4/5 | More than 50 and up to 125 a month. |
| 3/5 | More than 125 and up to 250 a month. |
| 2/5 | More than 250 and up to 500 a month. |
| 1/5 | More than 500 a month in the vendor's currency, or no published price. |
AI billing clarity — 15%
Can a buyer tell what the AI chatbot will cost before switching it on?
| Level | What it means |
|---|---|
| 5/5 | Price, unit and edge cases are defined in writing, and an allowance is included on the free or entry paid plan, so the AI can be tried before buying usage. |
| 4/5 | Price and unit are public and defined in writing, including the edge cases, but no allowance is included in any plan, or the AI is only sold in a separate product or a higher tier. |
| 3/5 | Price and unit are public, but the written definition leaves common cases open, or the vendor's own pages contradict each other on what is billed. |
| 2/5 | No AI chatbot is offered, or either the price or the billable unit is missing from the vendor's public pages. |
| 1/5 | An AI chatbot is sold, but neither its price nor its billing unit is public. |
Channels and integrations — 15%
Does it cover the places customers write from, and the store or site you already run?
| Level | What it means |
|---|---|
| 5/5 | Everything in level 4 plus mobile SDKs or voice, and a public app marketplace or API. |
| 4/5 | Everything in level 3 plus WhatsApp, Shopify and WooCommerce, and a widget in German and Spanish. |
| 3/5 | Website chat, email and at least one of Messenger, Instagram or WhatsApp, with WordPress and one major ecommerce platform. |
| 2/5 | Website chat plus email or one messaging channel, with a few integrations. |
| 1/5 | Website chat only, with no documented ecommerce or CMS integrations. |
Data location and compliance — 20%
Where do chat transcripts live, and what can the vendor show for it?
| Level | What it means |
|---|---|
| 5/5 | Everything in level 4 on every plan including the free one, with a published sub-processor list. |
| 4/5 | EU hosting for a self-serve customer on every paid plan, plus the vendor's own SOC 2 or ISO 27001. |
| 3/5 | EU hosting exists but only on higher plans or by contract; or EU hosting on every plan without an independent security audit (SOC 2 or ISO 27001) of the vendor itself. |
| 2/5 | A data processing agreement and a documented hosting location, but data is hosted outside the EU only, or the vendor's own statements about location conflict. |
| 1/5 | No data processing agreement and no documented hosting location. |
Seats included at the entry price — 20%
How many people can answer chats before the bill changes?
| Level | What it means |
|---|---|
| 5/5 | A workspace price covering ten or more people at the entry plan, or no per-seat charge on any plan. |
| 4/5 | A workspace price covering five to nine people at the entry plan. |
| 3/5 | A workspace price covering three or four people at the entry plan. |
| 2/5 | A workspace price, but the entry plan covers one seat, so a second person means a higher plan. |
| 1/5 | Every agent is billed from the first one: the entry plan is priced per seat, so a second person doubles it. |
The ranking
| Product | Free-plan usability | Cost as you add agents | AI billing clarity | Channels and integrations | Data location and compliance | Seats included at the entry price | Score |
|---|---|---|---|---|---|---|---|
| tawk.to | 5/5 | 5/5 | 5/5 | 4/5 | 2/5 | 5/5 | 8.5 |
| Tidio | 2/5 | 3/5 | 3/5 | 4/5 | 4/5 | 5/5 | 7.2 |
| Crisp | 4/5 | 4/5 | 2/5 | 5/5 | 3/5 | 3/5 | 6.9 |
| Freshchat | 4/5 | 4/5 | 5/5 | 3/5 | 4/5 | 1/5 | 6.8 |
| HubSpot Live Chat | 4/5 | 4/5 | 4/5 | 3/5 | 4/5 | 1/5 | 6.5 |
| Smartsupp | 2/5 | 4/5 | 4/5 | 5/5 | 2/5 | 2/5 | 6.1 |
| Chatwoot | 3/5 | 4/5 | 5/5 | 2/5 | 2/5 | 1/5 | 5.4 |
| LiveChat | 1/5 | 2/5 | 4/5 | 5/5 | 3/5 | 1/5 | 5.2 |
| Lime Connect | 2/5 | 1/5 | 3/5 | 3/5 | 3/5 | 3/5 | 5.1 |
| Intercom | 1/5 | 3/5 | 4/5 | 3/5 | 3/5 | 1/5 | 4.9 |
| Zendesk | 1/5 | 2/5 | 5/5 | 3/5 | 3/5 | 1/5 | 4.9 |
Where the facts come from
- Primary sources first. Prices come from the vendor's pricing page; limits from its help centre or product catalogue; hosting and compliance from its legal and security pages. Each figure links to the page it was read from.
- A vendor is only ever cited about itself. We never use one vendor's blog as evidence about a competitor, and we do not cite affiliate round-ups.
- Stored once. Every price, cap and allowance lives in one product database. Reviews, comparisons, the free-plan audit and the Markdown version of each page are generated from it, so two pages cannot disagree. If a page references a fact that does not exist, the site fails to build.
- Never converted. Prices appear in the currency the vendor publishes. A price in euros stays in euros on the US edition; a price in dollars stays in dollars on the German edition.
- "Not documented" is an answer. When a vendor does not say how long a free plan keeps chat history, the audit says so rather than guessing.
- Contradictions are reported, not resolved. When a vendor's own pages disagree about a price, we show both figures with both sources and use the higher one in cost arithmetic.
How the free-plan audit works
The free-plan audit reads six things from each vendor's documentation: how many people can answer chats, how many conversations a month are allowed, how long chat history is kept, whether the vendor's logo can come off the widget, whether any AI chatbot is included, and whether channels beyond website chat are included. The interactive version then checks your team against those limits and, where the free plan stops fitting, works out the cheapest published route from list prices.
The audit is a reading of documentation. A limit that a vendor enforces but does not write down will not appear in it. If you find one, tell us and we will add it with your evidence.
Why widget speed is not scored yet
How much a chat widget slows a page matters, and we intend to measure it: the same test page, the same connection profile, each widget installed in its default configuration, results and method published in full. Until that has been done, widget speed carries no weight in the score. We would rather leave a criterion out than fill it with vendor claims or guesses. When measurements are published, the rubric will move to v3.0 and the change will appear in the changelog.
How pages are written
Pages are researched and drafted with AI assistance from the sources listed on each page, then checked against the product database. When a named human editor has reviewed a page, the byline says so; when none has, it does not. The site does not invent authors.
How often facts are re-checked
The date on every page under "Prices and limits verified" is the day the figures were last compared with the vendor's pages. The target is a monthly re-check of every price and free-plan limit. Every change is logged in the changelog with the date it was made.
This page was researched and drafted with AI assistance from the sources listed on it. We have not installed or tested these products ourselves. Method: How we review