AI analytics chat

Ask your click data a question

Type what you want to know and the assistant writes the query, runs it against your own clicks and answers in a sentence. On every plan, including Free.

On every plan, including Free · No credit card required

app.clik.so/ai-analytics
Top 5 links this week?

Your top link this week is clik.is/launch with 4,812 clicks — about 39% of the total.

clik.is/launch4,812
clik.is/pricing2,190
clik.is/demo1,477
Ask Clik AI about your link analytics…
Queried inside a 60-second token scoped to this workspace
Asking Clik AI which links performed best this week

Quick answer

What is an AI analytics chat for link clicks?

An AI analytics chat answers plain-English questions about link clicks by writing the query for you. ClikSo's runs on Gemini 2.5 Flash in a tool-calling loop of at most three steps, writing ClickHouse SQL against your own click data. Every request carries a 60-second token with a row filter, so a query cannot leave your workspace.

Token per request, carrying a row filter the model cannot escape
60-secondToken per request, carrying a row filter the model cannot escape
Messages a day by plan — it is on every tier, including Free
5–unlimitedMessages a day by plan — it is on every tier, including Free
Maximum tool-calling steps behind a single answer
3Maximum tool-calling steps behind a single answer
The furthest back the assistant can query, on any plan
180 daysThe furthest back the assistant can query, on any plan
  • On every plan including Free — 5, 20, 100 or unlimited messages a day
  • Tenant isolation is a per-request 60-second token carrying workspace_id and a timestamp floor
  • Analytics only — it cannot create, edit or delete links, or read anything but your clicks
  • Chat history is saved per user, not shared with the workspace
  • It can see at most 180 days, even on a plan with no retention limit

How a question becomes an answer

Five stages, three of which exist to bound what the model can do.

  1. Ask in plain English

    Type the question the way you would ask a colleague — “top 5 links this week?”, “clicks from USA last month?”. There is no query language to learn and no filter panel to set up first.

  2. It decides whether it needs your data

    A greeting or a how-does-this-work question is answered directly. A question about your links triggers a tool call, in a loop capped at three iterations so an answer can never spiral into an unbounded series of queries.

  3. It writes one read-only query

    Gemini 2.5 Flash writes a single ClickHouse SELECT against your click data. The query is checked before it runs: one statement, no writes, no file, URL, remote or system-table functions.

  4. The query runs inside a 60-second, row-filtered token

    Each request mints a token that lives for sixty seconds and carries a row filter of workspace_id = your workspace AND timestamp >= your plan’s earliest readable date. The filter is applied outside the model’s query, so no SQL it writes can reach past it.

  5. You get an answer, and the rows behind it

    The reply is a short summary in plain language, with the returned rows rendered as a table underneath so you can check the number rather than take it on trust. The conversation is saved to your own history.

What it sounds like

Three questions in a row. The second is a two-word follow-up, which is where a chat beats a filter panel.

  1. Clicks from USA last month?

    The United States sent 3,180 clicks last month — your second-biggest country, behind India on 5,204.

  2. Which cities?

    New York 812, San Francisco 604, Austin 441 and Chicago 388.

  3. Mobile vs desktop split last 7 days

    Mobile 2,104 clicks (68%), desktop 897 (29%) and tablet 93 (3%).

An example conversation. Each answer arrives with the rows it was read from underneath it.

Eight more it can answer

  • Top 5 links this week?
  • Top countries this month
  • Compare my traffic sources for the last 30 days
  • Which country sends me the most clicks?
  • Which cities clicked my launch link?
  • What is my top referrer this month?
  • Which browser and OS do my clicks come from?
  • Which utm_campaign performed best in August?

Nothing here asks about conversions or revenue — those happen on the destination site, and click data does not contain them.

Who ends up using it

Mostly people who know exactly what they want to know, and do not want to build a view to find out.

Marketers who do not write SQL

The question is usually simple and the dashboard route to it is not. Asking “which campaign got the most clicks in August” is faster than setting three filters and a date range, and it gets the same number.

The Monday morning check

One question, one answer, before the week starts. “Top 5 links this week?” is a sentence; the equivalent dashboard state is a handful of clicks you have to remember to make in the right order.

Client and stakeholder answers

Someone asks how the Germany push did while you are in the meeting. Type the question, read the answer off the table, move on — instead of promising to send it later.

Anyone new to the workspace

A new teammate does not know your naming conventions, your tags or where the filters live. They can still ask which links matter and get a straight answer on day one.

Chasing a spike

Clicks jumped and nobody knows why. “Where did clicks come from yesterday?” then “which referrer?” then “which city?” is three sentences, which is faster than three passes through a filter panel.

Exploring, not checking

Fifteen filters stack in the dashboard, which is powerful once you know which two you want. Asking in plain English gets to the same number without learning the panel first — which is the difference between looking something up and going looking.

How your workspace stays separate

A model that writes SQL is only as safe as the boundary around the SQL. Ours is not a prompt instruction — it is a token.

The row filter

Every request mints a fresh token that expires after sixty seconds and carries a row filter with it. The data source applies that filter to whatever query arrives, on top of the query, not inside it:

workspace_id = 'your-workspace' AND timestamp >= 'your-plan-window'

Because the boundary is enforced by the data source rather than described to the model, it holds regardless of what the model writes. A prompt instruction can be argued with; a row filter cannot. The same filter also carries the retention window, so the assistant cannot read further back than your plan allows.

What it cannot do

  • Reach another workspace’s clicks — the row filter is applied outside its query
  • Read past your plan’s retention window, or past 180 days on any plan
  • Write, update or delete anything — the query is a single read-only SELECT
  • Read your links, billing, team or settings — the only table it can query is your clicks

AI analytics assistants across link shorteners

Three shorteners publish one: ClikSo, Bitly and Dub.co. Bitly Assist does more than analytics — it creates links too, which ours deliberately does not — and Dub.co, like us, includes its assistant on the free plan.

AI analytics assistants across link shorteners
CapabilityClikSoBitlyDub.coShort.io
Natural-language chat over your own click data
Available on the free plan
Cheapest plan that includes itFree$10/moFreeNot offered
Publishes a per-day message allowance5 / 20 / 100 / unlimitedNot publishedNot publishedNot applicable
Publishes how workspace data is isolated from the model60-second row-filtered tokenNot publishedNot publishedNot applicable
Can also create and edit links for you

Competitor details taken from each provider's own published pricing and documentation, last verified 16 September 2026. Short.io publishes AI slug generation but no natural-language chat over click analytics; Rebrandly offers conversational access only through an MCP server driven from an outside assistant, not a chat inside its own dashboard.

Good to know

  • The assistant is capped at 180 days of history, and the cap is hard-coded rather than tied to your plan. On Business, whose analytics window is a full year, the dashboard will still answer questions about older clicks that the chat cannot.
  • It reads and does not act. It cannot create, edit, archive or delete a link, change a setting or generate a QR code, and the only data it can query is your click data — not billing, team or link settings.
  • Quotas are per workspace per day: 5 on Free, 20 on Starter, 100 on Pro, unlimited on Business, with a short-term ceiling of 15 messages a minute. When the day’s allowance is gone the chat stops until tomorrow.
  • Chat history is scoped to you rather than to the workspace, so conversations are not a shared record your team can search later.
  • It writes a query and summarises what comes back, and like any language model it can misread a question. The numbers come from your own data, but a surprising answer is worth checking against the dashboard.

AI analytics chat by plan

On every plan, including Free. Among link shorteners that publish an AI analytics assistant, only Dub.co also includes one at no cost. It answers from the same data as click analytics, which has no 180-day ceiling of its own.

Free5 messages a day
Starter20 messages a day
Pro100 messages a day
BusinessUnlimited messages
EnterpriseCustom allowance

Frequently asked questions

Three publish one, as of September 2026: ClikSo, Bitly and Dub.co. ClikSo includes its AI analytics chat on every plan including Free, at 5 messages a day rising to unlimited on Business. Dub.co’s Ask AI turns a question into dashboard filters, and Bitly Assist starts on Bitly’s Core plan at $10 a month. Short.io, Replug and TinyURL publish no natural-language chat over click analytics, and Rebrandly offers one only through an MCP server driven from an outside assistant.

ClikSo’s chat is built for one job, answering questions about your click data, while Bitly Assist also creates and manages links. ClikSo’s is on the free plan rather than from $10 a month, and ClikSo publishes both its per-day allowance and the mechanism that keeps one workspace’s data out of another’s answers, neither of which Bitly publishes.

No. Every request mints a token that lasts sixty seconds and carries a row filter — workspace_id equal to your workspace, and a timestamp floor set by your plan. The filter is applied to the data source outside the query the model writes, so a query cannot reach past it even if the model tried.

A maximum of 180 days, on every plan. If your plan’s retention is shorter than that — 30 days on Free, 90 on Starter — the assistant is held to the shorter window. On Business, where the analytics dashboard looks back a full year, the assistant is still capped at 180 days.

5 on Free, 20 on Starter, 100 on Pro and unlimited on Business. The allowance is per workspace per day and resets daily, with a short-term ceiling of 15 messages a minute. When the daily allowance is used up the chat stops until the next day rather than billing you for more.

No. Conversations are saved per user, not per workspace. A teammate in the same workspace sees their own chat history and not yours, even though you are both asking about the same links.

Any of them, including Free. The Free plan gets 5 messages a day, which is enough to decide whether it is useful before paying for anything. Among shorteners that publish an AI analytics assistant, only Dub.co also includes one on its free plan; Bitly charges from $10 a month for Bitly Assist.

Not yet. Claude, ChatGPT and other AI apps can connect to ClikSo through its MCP server to create and manage links, but the click-statistics tools for AI apps are not switched on. Questions about clicks belong in this AI analytics chat, inside the ClikSo dashboard, or in the analytics endpoints of the REST API.

It works alongside it. The chat is faster for a single question and for comparisons that are awkward to build from filters, while the dashboard is the place for browsing, for the world map, and for date ranges longer than the chat’s 180 days.

Last reviewed 16 September 2026 against the live product. Competitor details come from each provider's own published pricing. See the full feature list or ClikSo vs Bitly.

Ready to get started?

Join the marketers and developers building the future of link sharing.

Free forever plan. No credit card required.