LiveChat or Tidio: separate human support from AI conversations

Analysis of official documentation, with no testing on a paid account.

Differences that change choice

Describe three representative requests

Prepare a simple question about schedules, a request that requires order information and a claim. Define which can receive a standard response and which must be repeated by a person. The aim is not to automate all exchanges without distinction.

In Tidio, the Lyro conversations offered to try the service must not be presented as a free allowance renewed every month. Visitors covered by Flows and conversations with agents use different counters. Prepare the cost estimate with these three volumes kept separate.

Identify the purchased product

The LiveChat pages highlight Text for certain functions combining chat, tickets and AI. Check if the quote concerns LiveChat, Text or a combination of services. Historical integration documentation may describe a different cost than the proposed new offer.

For each solution, examine useful languages, schedules, accesses and sources of knowledge. Check that the answer cites the applicable refund policy. Prepare cases where information is missing and ensure that the system is able to transmit the request.

Measure recovery rather than volume alone

Call an agent after an insufficient automatic response. He must find the question, the information already given and the reason for the transfer. Count the correction time and the requests left unaccounted for, not only the exchanges announced as treated.

Choose LiveChat if the way your team works and the selected plan meet the need. Choose Tidio if its automations and counters match your actual volume. In both cases, keep a procedure for contacting a person and assign someone to update the answers; no actual resolution rate is claimed here.

One complaint: what the robot must transmit to the agent

Let us take this case: a customer announces a damaged package, requests a refund and does not have his order number. An answer to the schedule doesn't solve anything. Prepare a transfer rule with the minimum information to be collected, the person responsible and the time announced to the client. Only give the test fictitious commands and identities.

The Lyro documentation states that the introductory quota of 50 conversations does not renew. It counts the first reply in a new thread and says that a new thread can start after fifteen minutes of inactivity. Ten messages therefore do not necessarily equal ten units, and the same visitor can create several threads.

1. Make the complaint without a number, then explicitly request a person.
2. Verify that the officer sees the initial question and the answers already provided, without asking again.
3. Repeat the journey outside working hours: check the waiting message and request assignment.
4. Record requests without an owner, replies needing correction and the time needed to pick up the work. Do not equate a transfer with a resolution.

Which changes the decision

If requests almost always need human checks, prioritise the team's workflow over a large AI quota. If repetitive questions have reliable, stable answers, automation may be justified after this trial. The current LiveChat grid presents Text: make the offer and rights clear, without extrapolating from an old contract. No real resolution rate is claimed.

Reviews to consult

• Tidio
• LiveChat