Paid media
Move budget where profit is actually incremental.
Reported ROAS says scale everything. Ask which spend causes profit after all costs, get next week's moves prepared with the evidence attached, and approve them before anything changes in an ad account.
Agent examples
Agents that move budget on evidence.
One weighs reported ROAS against incrementality before proposing a move; one watches every category for profit anomalies. Neither touches an ad account without a named approval.
Profit Pilot
Runs Fridays 16:00
Prepare next week's budget moves where profit is incremental, and send them for approval.
Reasoning
“Reported ROAS says scale everything — but brand search is soaking up credit. Check incrementality before moving budget.”
Queried modeled data
Spend, contribution margin, and returns by channel and market · last 28 days
Asked the MMM
Diminishing-returns curves per channel · profit-optimal split for next week
Used skill: budget-move-guardrails
Max ±20% per move, never against low stock cover
Modeled next week's allocation
3 moves · +€19.5k projected weekly profit
3 budget moves prepared
Awaiting approvalMeta · Google Ads
Plan posted to #growth with the numbers behind each move
App examples
The queue where budget moves get approved.
Every prepared move lands in one place with the numbers behind it — approve it, and it executes to Meta or Google Ads.
Budget Review Queue
Prepared spend moves · nothing executes without a named approval
Moves queued
6
Net impact
+€48K CM3/mo
Platforms
3
Approved this week
4
Auto-executed
0
Meta Prospecting DE → Google PMax DE
€12,000/mo · MMM + holdout test · +€9.8K CM3/mo
Brand Search SE → Meta Retargeting SE
€6,500/mo · Incrementality test · +€4.1K CM3/mo
Meta ASC US — budget cap +15%
€9,200/mo · Scale headroom model · +€6.3K CM3/mo
Approved moves execute to Meta & Google AdsQuestion examples
Questions that settle the budget debate.
Which spend is actually incremental, where the next euro should go — asked ad hoc, answered on the same measurement the agents act on.
Getting started
Live before the end of the week.
No implementation project, no code, no analyst backlog — connect once, describe the workflow, and the whole team has it.
01
Connect your data once
Store, ad platforms, and logistics connect in about a day and stay maintained — nobody on your team owns a pipeline.
Pause any campaign whose cost per new customer runs 25% above its 30-day average for 3 days.
02
Describe this workflow
Type what you want the way you'd brief a colleague. Every example on this page is a real prompt — that sentence is the whole setup.
Runs on schedule
LiveShared with the team
YLEMTAdjusted in chat
~40s
03
It's live for the whole team
Scheduled, shared, and permissioned from day one. Teammates use it without setting anything up, and changes happen in chat.
See this on your own data
Leave your email and pick a time — we'll walk through this workflow live on a 20-minute call.
The moves are the easy part.
What makes them safe to approve is everything underneath: one profit figure after all costs, definitions the AI cannot misread, measurement that proves cause, and an approval step your team controls.
01
The number is CM3, and everyone gets the same one.
Orders, spend, COGS, fulfilment and returns are reconciled into one profit figure. Marketing, finance and merchandising stop arguing about which export is right.
Profit waterfall
Revenue to CM3 breakdown for the current period. Each row shows the contribution or cost deducted.
02
The AI never has to guess what a metric means.
Your metrics, markets and relationships are modeled before AI touches them, so it reads the definition your team set instead of inferring one from raw tables.
“Dema's Agent is truly mind-blowing. We struggled to get AI to understand our data structures and metrics, but Dema worked perfectly right out of the box. It's the first tool that truly understands our data from the very first prompt!”
Metrics
Select which metrics to include in this analysis. The agent will track and report on all selected metrics.
Available metrics
Revenue
Contribution Margin 3
ROAS
New Customer Rate
Impressions
CPA
Sessions
AOV
4 of 8 selected
03
Know which spend actually caused the growth.
Marketing mix modeling, incrementality tests and attribution sit inside the same profit model. A recommendation to shift budget arrives with the evidence behind it.
“Since we started using Dema's incrementality testing, we've gained a much clearer view of what truly drives incremental demand in our paid social. Those insights now guide a significant part of our global spend.”
Kousha Torabi
Co-founder, Ninepine
Recommended shift
+€9.8k profitMeta Prospecting
−€6,000
Google Search
+€6,000
Backed by incrementality tests, not platform attribution.
04
It already knows how your company works.
Organisation-wide instructions set the rules every agent follows, and each person's own preferences sit on top. New joiners inherit the house standard instead of learning it.
Prioritize profit over revenue in all analyses. Always report CM3, not CM2. Focus on Nordic markets first.
Show me weekly trends by default. Include new customer contribution in all reports. Highlight anomalies with a brief explanation.
05
A good prompt lives on one laptop. This runs for the team.
Share it, set who can edit, and have it delivered into Slack on a schedule. When that person leaves, the workflow stays.
Share
Share this analysis with team members or generate a public link for stakeholders.
People with access
You
you@company.com
Lisa Eriksson
lisa@company.com
Marketing Team
team channel
Link settings
06
The tool your team wanted, live this afternoon.
Build the app around how the team actually works instead of waiting for a roadmap. Anyone who can describe the change can make it.
“With Dema Agent, what took two people three to four months now runs in a single command. We built 20 skills in three weeks without writing a line of code.”
07
Teams like yours already shipped this.
More than a million AI actions run inside Dema every month, at brands like Acne Studios, Axel Arigato and Represent. These are workflows their own commercial teams built, not pilots a vendor set up.
AI actions run in Dema every month
1M+
Anyone can get an answer once.
The question is whether the recommendation survives an incrementality test, whether the moves reach the ad accounts, and whether anyone signed off in between. That is the difference between a prompt and a system.
An AI assistant on your exports
Where the numbers come from
An AI assistant on your exports
The ad platform's own reporting, exported and pasted in — each platform grading its own homework.
Dema
Spend, orders, returns and costs reconciled into one profit figure before anyone asks.
What a metric means
An AI assistant on your exports
ROAS, CAC and margin re-explained in the prompt, slightly differently by each person.
Dema
Defined once by your team, then reused by every answer, report and budget move.
Whether the recommendation holds up
An AI assistant on your exports
Reads platform attribution back to you, confidently.
Dema
Checked against MMM and incrementality tests before it suggests moving a euro.
Who gets it, and when
An AI assistant on your exports
Whoever remembers to re-run the prompt before the budget meeting.
Dema
Prepared moves land for approval every Friday, with the tests attached.
What happens when the data is wrong
An AI assistant on your exports
It answers anyway, in exactly the same confident tone.
Dema
Broken feeds and missing spend get flagged instead of quietly averaged in.
What it is allowed to change
An AI assistant on your exports
Nothing. You carry the numbers into each ad platform yourself.
Dema
Approved moves execute in Meta, Google, and the other connected channels.
Where it runs
An AI assistant on your exports
The provider's default region, on the provider's terms.
Dema
EU or US inference, your choice, and never used to train a model.
A one-off idea, right now
An AI assistant on your exports
Genuinely better. No setup, no connection, no context — just ask it.
Dema
Built for budget decisions that repeat weekly, which is more than a throwaway question needs.
Profit, in real time
“Since we started using Dema, we can optimize our budgets for the most profitable growth in real time.”
Questions
Questions about optimizing spend
Every platform grades its own homework — its recommendations maximize spend on that platform, measured by its own attribution. Dema measures each channel against holdouts and MMM, on contribution profit after returns and fees, so the recommendation can also be to spend less.
No. Reported numbers stay visible next to the modeled ones — the point is seeing both. Recommendations are checked against incrementality before they are made, so you always know which of the two a suggested move relies on.
Only within the permissions you set. The default workflow prepares moves with the evidence attached and waits for a named person to approve. Approved moves then execute in Meta, Google, and the other connected channels.
Marketing mix modeling, incrementality tests, and causal attribution sit inside the same profit model as your orders and costs. A recommendation to shift budget arrives with the test or model behind it, not just a number.
Yes. CAC targets, breakeven levels, and guardrails are defined once per market by your team, and every answer, monitor, and prepared move reuses them — nobody re-explains the rules in a prompt.
Bring last month's spend