Answer anything the property can answer
Standard, pivot, batched, and realtime reports, composed from your question and returned as visual reports in the chat.
Everyone agrees GA4 holds the answers. Almost nobody enjoys going to get them. The screen buries simple questions under reports, the metrics changed names, and the person who really knew your site left two reorgs ago. So the data everyone trusts becomes the data nobody opens.
The GA4 connector separates the answers from the screen. It links the AI you already use, like Claude, ChatGPT, or Copilot, to your GA4 properties through Mad Fish Elements. Your AI runs real reports against your live property: sessions, users, events, conversions, revenue, any dimension you would build an exploration for, and returns the answer in plain language with a visual report in the chat. The tracking stack you already pay for finally talks.
Everything the GA4 Data API can report, your AI can ask for. Standard reports with any mix of dimensions and metrics over any date range: traffic by channel, conversions by landing page, revenue by campaign, events by device. Pivot reports when the question is a crosstab, like conversions by channel by device, without denormalizing rows by hand. Batched reports when one question needs five views of the same property, answered in a single pass. Realtime reports when you need to know what is happening on the site in the last 30 minutes, like launch day, or the minutes after the email went out.
It also reads the property itself: which audiences exist, how the property is configured, and even data-access reports for when governance asks who has been looking at what.
The plain-language layer is the point. "Where did our conversions come from last month?" becomes the right dimensions and metrics without you knowing what they are called this year. And the answer arrives with the numbers visible, so you can verify any of it in GA4 directly.
GA4 is your system of record, so the connector treats it with respect: analysis is unlimited, and the write surface is on purpose narrow.
Standard, pivot, batched, and realtime reports, composed from your question and returned as visual reports in the chat.
Period over period, channel against channel, page against page, with the driver of a change named and checkable.
The one write that matters: your AI can create GA4 audiences, with the membership window and filter rules shown to you before anything is saved. The segment you described in a sentence becomes an audience your ad platforms can use.
GA4 results land beside Google Ads, Meta, and Search Console in the same chat, which is where tracking earns its keep: did the spend move the numbers that matter?
Nothing here edits your events, streams, or property settings. Your tracking setup stays exactly as your team built it.
In practiceThe working session: your CMO asks why leads dipped in July. Old way: an afternoon in reports. Here, you ask: "Compare July to June: sessions, conversions, and conversion rate by channel." Your AI returns the table, and the story is visible: organic held, paid sessions dropped a third. You ask: "Paid by campaign, same check." One campaign went dark mid-month. You ask Google Ads, same chat: "What happened to that campaign in July?" Budget hit its cap on the 12th. That is the answer: not "leads dipped," but "one campaign capped on the 12th, here is the number." Twelve minutes, three questions, no reports.
Questions that meant reports and exports become sentences. The answer arrives before the meeting instead of after it.
Everyone quotes the same live site instead of their own screenshot. Arguments get shorter and better.
Launch days and send days get a live read in the chat, not a nervous refresh loop.
Describe the audience; review the rules; done. Your remarketing lists stop waiting on whoever knows the audience builder.
GA4 beside your ad platforms in one answer, so "did it work" gets settled with evidence.
"Show me sessions, conversions, and conversion rate by channel for the last 30 days."Google Analytics 4
The core traffic report, live from your site, as a visual report in the chat.
"Which landing pages drove the most conversions this quarter, and how did each convert?"Google Analytics 4
A ranked landing-page answer that used to be an exploration build.
"Break out conversions by channel and device as a crosstab for last month."Google Analytics 4
A pivot report from one sentence, no manual denormalizing.
"What is happening on the site right now? The email just went out."Google Analytics 4
A realtime read of the last 30 minutes, on demand.
"Create an audience of visitors who viewed pricing twice in 30 days but did not convert."Google Analytics 4
A real GA4 audience, rules shown for sign-off before it saves.
"Did last week's Google Ads changes show up in GA4 conversions?"Google Analytics 4 + Google Ads
Spend-to-outcome attribution in one pass, both sides live.
The property your team built finally pays rent. Anyone with a Read seat can ask real questions and get checkable answers, so the analytics bottleneck stops being a person and starts being an access level.
For marketing teamsClient tracking questions stop costing analyst hours. Read every client property through one connection with per-client access walls, answer "what did our work move" with live numbers on the client call, and build the audiences your media team needs in the same chat.
For agenciesAsk in chat and share the dashboard without building exports or taking screenshots.
Sessions, conversions, and rates by channel over any range. Built for the weekly review.
Conversions and conversion rate by page, ranked; built for the CRO chat.
GA4 outcomes beside the spend that drove them. Built for proving the work moved the numbers.
The charts do the talking: line for trend, bar for channel and page check, funnel for stage drop-off, heatmap for conversions by day and hour. And when your leaders or clients want these anytime, the Data Visualization add-on turns your reports and analysis into always-current dashboards they open from a link. Ask in chat; share the dashboard. No exports; no screenshots.
Access is set per person, so Read seats see everything and change nothing, while Write seats create and edit with your approval flow in front of each change; Delete permissions go only to the people you trust to clean up. Many teams start read-only, check answers against the platform for two weeks, and then widen access as the audit trail earns trust.
You choose the level of human review and can change it at any time, while every action records what changed, when it changed, and who approved it. Per-user OAuth means Elements never stores your Google Analytics 4 password, and your data is never harvested or resold.
It will not touch your tracking setup. Events, streams, conversion definitions, property settings: all read-only, all yours. It will not invent data GA4 does not have. Sampling limits and thresholds are real, and your AI names them when they apply. It will not replace your analysts. It answers the questions that were stealing their afternoons, so they can do the work that needs a human. And the one write it has, creating audiences, shows you the full rules before anything saves. Your property stays exactly as your team built it. The answers just stop hiding inside it.
Plans, account limits, and the multi-client Agency plan are on the pricing page .
No. The connector reports on your property and can create audiences with your sign-off. It does not touch events, data streams, conversions setup, or property settings. Your tracking stays as your team built it.
No; ask in plain language and your AI composes the right report. The returned answer shows what was queried, so your analysts can verify the translation any time.
Yes. Connect the properties you manage and ask questions per property, with per-person access rules deciding who can see which. Agencies get per-client walls as usual.
No. Your AI acts only with the access rules you set, changes go through the sign-off level you choose, and Read-only seats cannot change anything at all because the sign-off flow is the product working as designed.
Your team can use Claude, ChatGPT, and Microsoft Copilot, while other AI tools can likely connect to Elements when they support MCP, the open standard for linking AI assistants to outside tools.
No. Your data stays yours, so nothing is gathered and nothing is resold.
The teams winning with AI in marketing are not handing their accounts to a black box; they keep the strategy and delegate the hands-on-keyboard work, while approvals and an audit trail make it safe to move fast. Connect the account you already have through the Gateway and ask the question you would normally build a report to answer.
Your first 14 days are free. Add your card to start, and cancel before the trial ends to pay nothing.