Guides

Answer Traffic Questions With a Grok Bot Template for Web Analytics

Taras Shynkarenko
Taras Shynkarenko
7 min read
A Grok bot template for web analytics answering questions in chatA Grok bot template for web analytics answering questions in chat

TL;DR, Quick Answer

7 min read

Install the Flowsery template at x.ai/bot/tOP05p0n0XVUcpJDfPH0k, paste a workspace token, and ask questions about traffic, revenue and the issues Flowsery found in session recordings. Reads are safe to automate. PATCH /issues is the one call that needs a human.

Wiring up a Grok bot template for web analytics is less interesting than what it exposes. Traffic numbers in a chat are a party trick. The thing worth putting behind an agent is the issue list Flowsery builds from session recordings, because that is the data nobody opens often enough.

The short version: install the Flowsery template from x.ai/bot/tOP05p0n0XVUcpJDfPH0k, paste a workspace token into the credential field, then ask "what are the open critical issues and how many sessions did each hit?"

The Flowsery Grok bot template share page on x.ai with the Add to Grok Bot button
The Flowsery template's share page. The disclaimer under the description is xAI's.

What is a Grok bot template?

A Grok bot template is a share link that copies one person's configured Grok Bot into your own app. SpaceXAI turned the feature on with a single sentence from the Grok Bot account on 28 August 2026: "You can now share templates of your Bots with others." The link is x.ai/bot/ plus a 21-character id, and the "Add to Grok Bot" button is a deep link into the desktop or iOS app, grokbot://app/v1/bot-template?id=<id>.

Grok Bot launched on 11 August 2026 in early beta. Each Bot runs on its own cloud computer, which is why xAI can say it works "across apps, tools, and websites, including platforms with no clean API or MCP." Access needs an eligible paid plan: SuperGrok, SuperGrok Plus or SuperGrok Heavy, or Cursor Pro, Pro+ or Ultra, or Cursor Teams Standard or Premium. Entry pricing on x.ai/bot is $30 a month for SuperGrok and $20 a month for Cursor.

Why put analytics behind a chat interface?

A chat interface wins when the question is exploratory and the follow-up matters more than the first answer. Dashboards are good at questions you already knew you had. Every Flowsery breakdown, by page, referrer, channel, campaign, country, device, browser, is a chart somebody has to think to open.

The follow-up is where a bot earns its place. "Traffic is down 18% week on week" is one call. "Which channel, and did the drop land on one page?" is three more, and a bot chains them without you deciding the order in advance.

One question, four calls
GET /overview
GET /channels
GET /pages
GET /issues
"Why is traffic down?" is not one lookup. Chaining the follow-ups is the part worth automating.

What is in the Flowsery Grok bot template?

The Flowsery template contains a bot profile, a skill describing Flowsery's public REST API, and one weekly routine. The skill is written as prose so that anyone reviewing a third-party template can read exactly what the bot is allowed to touch.

Purpose: query analytics and AI-detected issues through the Flowsery REST API.
 
Auth: every request needs the header "Authorization: Bearer <token>". The owner
creates a workspace token at https://flowsery.com/api-tokens (it starts with
flow_ws_). Keep it in the connector credential field, never in the conversation.
Workspace tokens span several websites, so pass websiteId or domain on every
call. GET /websites lists them.
 
Base URL: https://analytics.flowsery.com/analytics/api/v1
Full spec: GET /openapi.json on the same base URL, no auth needed.
 
Workflow:
1. GET /websites once to learn website IDs and domains.
2. Issues: GET /issues (filters: status open | in_progress | resolved |
   suspended, severity low | medium | high | critical, sort severity | recency)
   for the ranked list with sessions-affected counts. GET /issues/{issueId} for
   occurrences, affected sessions and steps to replicate. PATCH /issues/{issueId}
   with {"status": "resolved"} only when the owner says an issue is fixed.
3. Traffic: GET /overview and GET /timeseries for visitors, sessions, revenue
   and conversion over a startAt/endAt range. GET /realtime for who is on the
   site now.
4. Breakdowns: GET /pages, /referrers, /campaigns, /channels, /countries,
   /devices, /browsers to find where a drop-off concentrates.
5. Revenue: GET /goals for completions. POST /goals and POST /payments record
   server-side conversions.
 
Validation: confirm with the owner before any PATCH or POST. Reads need no
confirmation.
 
Failure: a 404 means the websiteId or domain does not belong to this workspace.
List /websites and ask, do not guess.

The workspace-token rule is the one that trips people up. A workspace token can read several websites, so every call needs websiteId or domain to say which. Website API keys are already scoped to one site and can skip the selector. Get this wrong and you get a 404 that looks like a permissions problem but is really a missing parameter.

What can the bot read, and what can it change?

Flowsery's public API has 26 operations, and 21 of them only read. Five can change data, and the skill uses three of those. That ratio is the reason this template is comfortable to automate.

CallWhat it returnsChanges data
GET /issuesAI-detected issues ranked by severity, with sessions-affected countsNo
GET /issues/{issueId}One issue with occurrences, affected sessions and steps to replicateNo
GET /overviewVisitors, sessions, revenue and conversion for a date rangeNo
GET /timeseriesThe same metrics bucketed over timeNo
GET /realtimeVisitors active right nowNo
GET /pages, /referrers, /channels, /campaignsBreakdowns for locating a drop-offNo
GET /visitors/{visitorId}One visitor's profile and activity timelineNo
PATCH /issues/{issueId}Moves an issue to open, in progress, resolved or suspendedYes
POST /goalsRecords a goal completionYes
POST /paymentsRecords a payment, renewal or refundYes

PATCH /issues/{issueId} is the one to guard. Marking an issue resolved removes it from the default list, and the default list is what everybody looks at. An agent that resolves issues on its own judgment quietly shrinks your bug backlog without fixing anything.

Flowsery API documentation for the list issues endpoint showing severity and sessions affected
The issues endpoint the skill leans on, documented at flowsery.com/docs.

How do I install and test it?

Installing the Flowsery Grok bot template takes four steps: open the share link in Grok Bot, create a flow_ws_ workspace token, paste it into the credential field, and ask it to list your websites as a read-only test.

From share link to first answer
1
Open the link on a device with Grok Bot installed. Press "Add to Grok Bot" on x.ai/bot/tOP05p0n0XVUcpJDfPH0k. The deep link needs the app; a browser tab alone does nothing.
2
Create a workspace token at flowsery.com/api-tokens. It starts with flow_ws_.
3
Paste it into the credential field, not the chat. Anything typed into a message stays in the conversation.
4
Ask: "list my websites, then show open critical and high issues for the first one." That exercises GET /websites and GET /issues and writes nothing.
If step 4 returns a 404, the bot skipped the websiteId. Tell it to call /websites first.

What should the weekly routine actually say?

A good routine names the comparison, the sort order and what to do on a quiet week. The template ships with this one:

Name:      Site health report
Schedule:  weekly, Monday 09:00, owner's timezone
What:      Pull last week's /overview against the week before, the top pages and
           referrers, and all open issues sorted by severity. If a new critical
           or high issue appeared since the last run, lead the report with it,
           including sessions affected and steps to replicate. One message,
           numbers first, no fluff.
If empty:  send the report anyway. Traffic of zero is itself a finding.

That last line is the opposite of the advice I would give for a lead-alert routine, and deliberately so. A quiet week in a mention inbox is normal. A quiet week in site traffic means either nothing happened or the tracking script broke, and you want to know which.

xAI's usual path to a routine is demonstration: "Ask a Bot to follow along as you complete a workflow once. It saves it as a routine and runs it on its own next time." For an API-driven report, writing it out is better. You get to specify "sorted by severity" instead of hoping the Bot inferred it from watching you scroll.

What does the template not carry?

A template carries the profile, the skill text and the routines, and nothing else. It does not carry the creator's workspace token, their websites, their session recordings, their issue list, or any conversation their Bot had. Your copy starts empty and reads only the sites your own token can see.

Flowsery
Flowsery

Start FREE Trial

Real-time dashboard

Goal tracking

Cookie-free tracking

What crosses the share link
Copied to your Bot
  • Name, title, description, avatar
  • The skill text
  • The routine definition
Never copied
  • Workspace or website API tokens
  • Session recordings and issues
  • The creator's chat history
  • Anything their Bot remembered
The skill text is plain English. Read it before installing any third-party template, including this one.

xAI puts a line on every share page: a third-party bot "may act on your behalf." That is accurate. A skill is instructions, instructions can name any endpoint, and the only defence is that you can read them.

Frequently asked questions

What data can the Grok bot see?

The bot sees whatever the token allows, which for a workspace token is every website in that workspace. Flowsery is cookie-free and stores no personal data, so the bot reads aggregate traffic, revenue, goal completions and the issues found in session recordings.

Can the bot close bugs on its own?

Not with the shipped skill. PATCH /issues/{issueId} is the only call that moves an issue's status, and the skill requires the owner's confirmation before any PATCH or POST. Reads run without confirmation.

Why does the bot get a 404 on a website it can see?

A workspace token spans several websites, so every call needs websiteId or domain to say which one. A missing selector returns 404, not 403, which reads like a permissions error. The skill tells the bot to call GET /websites and ask rather than guess.

Does the Grok bot replace the Flowsery dashboard?

No. The bot answers questions and writes reports; it does not render funnels, replay sessions or show heatmaps. It is faster for "which page lost traffic last week" and useless for watching a recording of somebody rage-clicking a broken button.

How is this different from connecting Flowsery over MCP?

The endpoints and permissions are identical. MCP suits clients that speak it, such as Claude, Cursor and Codex. The Grok bot template exists because Grok Bot reads skills written in prose rather than MCP tool definitions.

Do I need a paid Grok plan for this?

Yes. Grok Bot requires an eligible subscription, and the template is free on top of it. Entry pricing on x.ai/bot is $30 a month for SuperGrok and $20 a month for Cursor. xAI notes that Grok Bot "comes with its own usage, separate from your Grok and Cursor plans."

Install the Flowsery Grok bot template and ask your analytics questions in a chat, or read the API docs first.

Sources: x.ai/bot, Introducing Grok Bot (11 August 2026), @bot on X (28 August 2026), and the Flowsery OpenAPI spec. Checked 30 August 2026.

Was This Article Helpful?

Let us know what you think!

See us more often in Google

One click marks Flowsery as a preferred source, so our articles sit higher in your Top Stories, AI Mode, and AI Overviews.

Before you go...

Flowsery

Flowsery

Revenue-first analytics for your website

Track every visitor, source, and conversion in real time. Simple, powerful, and cookie-free.

Real-time dashboard

Goal tracking

Cookie-free tracking

Related Articles