One question across every market: how a group media lead uses AI Insights

Written by Uptimal, the team that builds and operates Upbuild and runs marketing analytics for multinational clients across the globe. Everything below reflects how the product actually works, including the parts that are deliberately boring.

Monday, 08:40. The CMO of a group with brands in many markets sends the media lead a one-line message: which markets are over budget this month, and why? Until recently the honest reply was "by Thursday". The media lead would write to each market's agency, wait for a slide from each, reconcile spend in three currencies and two definitions of budget, and present a deck that was already a week old on the day it was finished.

This is what the same Monday looks like with AI Insights.

The question, typed once

She opens the group's shared chat and types the question as the CMO wrote it:

"Which markets are over budget this month, and why?"

Upbuild AI reads every ad and analytics account the organisation has shared with her, such as Google Ads, Meta or GA4, across every market and brand, and answers in seconds with a table: market, brand, monthly budget, spend to date, pace, and the campaigns behind the difference. Germany is over because the fourth-quarter launch started two weeks early. Spain is over because generic search picked up a competitor's brand term. Italy looks over but is not: the budget in the plan was never updated after the September uplift.

The budget figures come from the plan, which in most groups lives in a spreadsheet the team keeps. Upbuild reads that alongside the platforms, so plan and actual sit in one table without anyone pasting.

She asks one more thing before replying to the CMO: "Show me the campaigns behind Germany, with spend to date and daily pace." The rows arrive under the total, and the reply goes out at 09:05 with the table attached and the German launch named as the reason. The deck would have said the same thing on Thursday.

Then the follow-up, the one that used to be a separate project:

"Compare CPA by platform across EMEA against the same month last year."

The answer comes as a table and a chart, every figure straight from the platform, with the comparison windows labelled. Where a market has no data for last year, because its account was only connected in the spring, the answer says so plainly instead of filling the gap. Making blended figures honest across platforms is a subject of its own: see cross-platform ROAS without the spreadsheet.

The chat sees only what the organisation shares with her

The reason a group can put every market in one chat is that the chat does not see every market by default. It sees what the organisation has shared with the person asking.

Everyone signs in as themselves with Google or Microsoft. An admin decides who joins the organisation and shares accounts one person and one account at a time. The group media lead has been shared every market's accounts, so her question spans all of them. The media manager in Spain has been shared Spain, so the same question from him returns Spain. Access is checked on every single question, and a chat never leaves the organisation.

When a market lead moves to another company, the admin removes them and their access ends that day. The chats, dashboards, connections, exports and alerts they built belong to the organisation and stay. Every question is logged, which was the first thing the group's security team asked about.

A colleague picks up the same thread

The chat is collaborative, which changes what an answer is for. The media lead shares the thread with the group's finance partner, who was going to ask her the same question on Wednesday.

He opens it and reads the whole exchange, question and answer, so he sees how the overspend was found and not just the number. Then he asks his own follow-up in the same thread: "How much of the German overspend is the launch, and where does the year land at this pace?" Upbuild AI answers on the same accounts, and the answer sits under hers, where she sees it that afternoon.

Nothing was exported, forwarded or re-typed. The deck used to be the only way to hand an analysis to a colleague. Now the analysis is the thing you hand over, with its working attached.

The thread is also the record. When the CMO asks in December why Germany was over in September, the question, the follow-ups and the figures as they stood that week are still there, in the order they were found, rather than in a deck attached to an email that left with the person who wrote it.

A good answer becomes a tile, in one prompt

By Tuesday the CPA comparison is clearly not a one-off. The CMO will ask again next month, and the regional leads want it in front of them every week.

"Add this to the EMEA weekly dashboard."

The table becomes its own tile on the dashboard, next to the charts the region already reads, and refreshes on the dashboard's own schedule, every 30 minutes, every hour or once a day. Nothing to rebuild and nothing to copy across. The regional leads open the dashboard through its link, with no Upbuild account and no platform login, and next week they see next week's number. The whole path from a question to a standing tile is in from an answer in the chat to a live dashboard.

The same accounts feed AI Alerts, so the question she asked on Monday can also become a rule that asks itself: "Tell me when any market has spent more than 90% of its monthly budget before the 25th." Next month the overspend arrives in her Teams channel before the CMO's message does.

Versus asking each market's agency for a deck

The agencies are not the problem. Each of them knows its market better than the group ever will, and their judgement about why Spain moved is worth more than any table. The problem is using them as the group's data pipeline.

A deck per market arrives on each agency's timetable, in each agency's definitions and attribution windows, in the local currency, with whatever comparison period the account manager thought was fair. Reconciling the decks is a job in itself, and the result cannot be questioned. If the CMO asks "and by platform?", the answer is another round of emails.

With every market in one chat, the numbers arrive the same way for every market, with the same definitions and windows, in seconds, and every answer can take a follow-up. The agencies keep the part that is theirs: the conversation about what to do next. The group keeps the part that should never have left it: the numbers, in one place, for whoever the organisation chooses to share them with.

For a media lead who prefers to work inside her own assistant, the same accounts are available read-only in Claude, ChatGPT, Copilot Studio or Gemini Enterprise through the Upbuild connector, under the same per-person access. The question is the same. Only the window changes.

Frequently asked questions