Dema
Dema is a commerce intelligence platform that unifies orders, marketing spend, inventory, returns and customer data into one model, then lets AI agents answer questions, monitor risks and propose profit decisions your team approves.
What is Dema?
Dema is a commerce intelligence platform published by Demaai AB, a Swedish company founded in 2022 by Marcus Tagesson, Henrik Hoffman and David Feldell, and now staffed by more than thirty engineers and growth specialists. It presents itself as an agentic operating system for commerce, and it rests on three layers meant to be used together.
The first is a unified commerce data model. Dema pulls orders, advertising spend, returns, costs, inventory and store sales into a single schema covering revenue, acquisition, retention, inventory and site metrics. Every metric carries its source and a mapping status, so a team can see that orders arrive from Shopify and ad spend from Google Ads and Meta, while others stay unmapped until configured. Shared definitions and access controls sit on top, so the same number means the same thing in every department.
The second layer is the agent workspace, organised around four verbs: chat, schedule, build and approve. Anyone can ask a margin, cohort or budget question in plain language. Scheduled jobs watch for margin drift, stockouts and spend anomalies. Named agents such as Profit Pilot, Visual Merchandising, Marketing Optimizer and Anomaly Detector handle standing work, and teams assemble their own reusable skills and no-code nano apps. Agents propose rather than execute: costed moves reach a human approval queue and every action is logged. Users pick a model per question, trading Opus for depth against Sonnet for speed and lower credit cost.
The third layer is unified measurement, placing marketing mix modelling, incrementality testing and attribution in the same view of profit. Incrementality tests use holdouts and geo experiments, typically running four to six weeks including a two-week post-treatment observation window, and the platform decomposes week-over-week shifts in sales, profit, CAC and retention.
Fifty native integrations, a GraphQL API, CSV and sFTP feed the model, and a Slack app carries the assistant into daily conversation. Dema is sold to e-commerce and omnichannel retail brands through a booked introduction, from 2,500 euros per month on a normally annual contract. The vendor itself warns that its agents can make mistakes and that results should be reviewed.
What it does
- Answer business questions in plain language across profit, cohorts, stock and spend, with no SQL
- Unify orders, ad spend, returns, costs, inventory and store sales into one commerce data model
- Measure what really drives growth with marketing mix modelling, incrementality tests and attribution
- Schedule recurring watches for margin drift, stockouts and spend anomalies
- Build reusable skills and no-code nano apps for recurring decisions
- Route budget shifts, markdowns and audience changes through a human approval queue
- Decompose week-over-week moves in sales, profit, CAC and retention down to their actual causes
When to use Dema / When not to
A quick filter to help you decide if Dema is the right fit.
When to use Dema
- E-commerce and omnichannel retail brands already running meaningful volume across several markets
- Growth and performance marketers who arbitrate media budgets by channel, market and campaign
- Buying, merchandising and inventory planning teams tracking stock, margin and sell-through
- Commerce teams with no dedicated analyst, who need answers without writing SQL
- Finance and controlling functions reporting on contribution margin and profitability
When not to use Dema
- Small or early-stage stores that cannot absorb a 2,500 euro monthly floor on an annual contract
- Anyone wanting to sign up and try the product alone, since every path runs through a sales demo
- Businesses outside e-commerce and retail, for which a commerce-specific data model brings little
- Teams needing a mobile app, as Dema ships only as a web app, an API and a Slack app
- Buyers who require a published SOC 2 or ISO 27001 certification before purchasing
How to use Dema
A typical end-to-end flow, from setup to results.
- Request an introduction through the booking form, giving your name, work email, company and team size
- Take the 30-minute walkthrough with the Dema team to scope your data sources and objectives
- Agree a plan: the Foundation suite, plus the Unified Measurement or Omnichannel add-ons if needed
- Work through setup with the dedicated onboarding manager included in every plan
- Connect your sources through the native connectors: Shopify, ad platforms, warehouses, ERP
- Review the commerce data model and manually configure any metric still flagged as unmapped
- Feed anything the connectors do not cover through the GraphQL API, CSV upload or sFTP
- Start asking questions in plain language from the workspace, or from Slack via the app
- Schedule recurring jobs for margin drift, stockouts and spend anomalies
- Build skills and nano apps for recurring decisions, and set who approves the actions agents propose
Pros & Cons
Pros
- Published, itemised pricing, which is rare for B2B software sold by demo
- Marketing mix modelling, incrementality and attribution in one view, where many tools stop at attribution
- Unusually detailed subprocessor disclosure: 21 entries with purpose, data processed and hosting region
- Agents propose but do not execute, with human approval required and every action logged
- A dedicated onboarding manager on every plan, with most customers live within two weeks
- Unlimited team members and API access on all plans, so no per-seat billing
- Explicit commitment not to train generalised models that benefit other customers
Cons
- High entry point: 2,500 euros per month for the base suite, 4,500 with both add-ons
- No free plan and no free trial; the one month free is an annual-billing discount
- No self-serve signup, so evaluating the product requires going through sales
- No public terms of service, so the contract cannot be read before negotiation begins
- No postal address published anywhere on the site
- No security certification claimed, neither SOC 2 nor ISO 27001
- Variable cost beyond the included credits, with the monthly cap left to the customer to set
Pricing & Plans
There is no free plan and no free trial. The lowest paid entry point is the Commerce Analytics Suite, called Foundation, from 2,500 euros per month. The Unified Measurement add-on starts at a further 1,500 euros per month and Omnichannel Analytics at a further 500 euros per month, so a complete configuration begins around 4,500 euros per month. Every plan includes 15,000 AI credits monthly; beyond those, credit packs run from 50 to 3,000 euros per month and any further usage is billed on demand, between 0.0052 and 0.0031 euros per credit depending on the largest active pack. Contracts are annual by default, with quarterly options available, and enterprise pricing is quoted on request.
- real-time commerce dashboard
- unified marketing and e-commerce reporting
- product
- variant and collection tracking
- AOV
- CVR
- bounce rate and revenue insights
- marketing spend and CAC across channels
- full integration of MMM
- MTA and incrementality testing
- lift tests
- holdouts and geo experiments
- scenario modelling and budget planning
- POS integration with Shopify POS
- Lightspeed and Sitoo
- cross-channel customer journey tracking
- store-by-store dashboards
- offline sales attribution for digital campaigns
- custom location tags and geo-based segmentation
- dedicated onboarding manager
- all core dashboards
- custom dashboards and saved reports
- unlimited team members
- API access
- priority support
- 50 euros for 12
- 000 credits
- 100 for 25
- 000
- 250 for 65
- 000
- 500 for 140
- 000
- custom pricing on request for multi-market operations or dedicated support
Data, GDPR & hosting
A consolidated view of how Dema handles your data.
GDPR overview
GDPR implementation is documented rather than merely claimed. Demaai AB publishes a full Data Processing Addendum built on Regulation (EU) 2016/679: the customer is controller, Dema is processor, processing follows documented instructions, and personal data breaches are notified without undue delay. Transfers outside the EEA rely on prior general authorisation together with the European Commission's standard contractual clauses. Customers hold an audit right on twenty business days' notice, and supervisory authorities are granted direct, unrestricted access. Articles 82 and 83 govern damages and fines. Subprocessor changes carry at least fourteen days' notice with a right to object. Access, rectification and deletion requests go to a published address and are answered within the one-month GDPR window. Established in Sweden, Dema designates no Article 27 representative, which is expected. Neither document displays an effective date.
Who owns the data?
Under Dema's published Data Processing Addendum, the customer remains the controller of any personal data placed in the platform, and Demaai AB acts solely as a processor on its behalf. Dema processes that data only on the customer's documented instructions, and only data the customer uploads or inputs is processed at all. Special categories of personal data are not permitted. When the agreement ends, Dema stops processing and returns the data together with all copies in its possession, at the customer's expense; if the customer does not answer that offer within one month, Dema may delete it. Dema is controller only for its own website cookies and integrations.
Reuse rights
Dema states it uses customer data only to deliver, improve and optimise the service, on documented instructions. Data drawn from the Google and Meta APIs serves solely to produce attribution analytics, profitability reports and recommendations for the customer who connected that account, and Dema says it derives no insight about that customer's own ad audiences. Nothing is sold, rented or shared with third parties for their own marketing; sharing is limited to the service providers listed in the addendum, under confidentiality obligations. Dema explicitly states it does not use Meta or Slack data to train generalised models that benefit other customers. Handling of Google API data follows the Google API Services User Data Policy, Limited Use requirements included. As a processor, Dema cannot repurpose customer data on its own initiative.
Data retention & training
Hosting summary
Dema publishes hosting by region rather than by country. AWS, its storage and hosting subprocessor, is declared in the EU region, and most of the platform sits alongside it: Intercom, Sentry, Mixpanel, Pydantic, HubSpot, Alva Labs, Teamtailor, Google Workspace, Contrast, Trumpet and TLDV are all listed as EU. A US group remains, covering Clerk for identity, Resend for transactional email, Linear, Slack and Zapier, plus the optional AI inference providers OpenAI and Anthropic. Google Cloud's Vertex AI is declared US and EU. Transfers outside the EEA are contractually foreseen under prior general authorisation with the European Commission's standard contractual clauses. Data is encrypted at rest and in transit, with strict access controls limiting access to staff having a legitimate business need. No individual hosting country is ever named, which is why the country-level field stays empty. One point deserves care: the server country resolved from DNS is the United States, but that is the anycast address of the web front end and says nothing about where customer data lives.
Things to keep in mind
Risks and trade-offs to weigh before adopting Dema.
- Beyond the included credits, spend is variable: a single eight-second video costs about 2,000 credits, the entire per-prompt cap
- Contracts are annual by default, so a poor fit is expensive to exit
- No public terms of service, so the contract only becomes readable once negotiation has started
- If a user links their personal Slack account, the assistant can read channel and private message history they can access, and act in their name
- AI inference subprocessors sit in the US, including OpenAI, Anthropic and Google Cloud, even though they are marked optional
- Subprocessor changes carry only fourteen days' notice, and audit costs fall on the customer unless a material breach is proven
- Dema warns that its own agents can be wrong: approving a recommendation without reading it quietly hands judgement to the machine
Setup & Integrations
Technical difficulty
Low to moderate, and deliberately assisted. Native connectors cover the main sources, and metrics are available as soon as a source is connected, with no modelling sprint required. Every plan includes a dedicated onboarding manager who connects the sources, configures dashboards and trains the team, and most customers are live within two weeks. Some residual work remains: the data model flags metrics as mapped or unmapped, and unmapped ones need manual configuration. Skills and nano apps are built without code. Advanced paths exist through the GraphQL API, CSV and sFTP.
Deployment
Integrations
Behind Dema
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
Is there a free plan or a free trial?
How much does Dema cost in full?
What are credits, and what happens when they run out?
What can the AI agents actually do?
Which systems does Dema connect to?
Is my data used to train AI models?
How long does setup take?
Where is data hosted, and is a DPA available?
Should you pick Dema?
Dema occupies a specific and increasingly crowded space, commerce analytics, but it does so with more depth than most. Putting a unified data model, an agent workspace and causal measurement on the same foundation is a real architectural choice, and it pays off: marketing mix modelling, incrementality testing and attribution sitting beside the same profit data is something many competitors stop short of. The approval queue is a genuine safeguard rather than a marketing line, since agents propose costed moves and a human signs them off, with everything logged.
Two things stand out on transparency. Pricing is published and itemised, down to the credit cost of a single question, which is rare in B2B software sold by demo. And the subprocessor disclosure is unusually thorough: twenty-one entries, each with its purpose, the data it processes and its hosting region. The Data Processing Addendum is public and properly built on the GDPR.
The reservations are mostly commercial and documentary. The floor is 2,500 euros per month on a standard annual contract, with no free plan, no trial and no way to test the product without talking to sales. No terms of service are published, so the contract only becomes readable once negotiation begins. No postal address appears anywhere on the site, and no security certification is claimed, neither SOC 2 nor ISO 27001, which enterprise procurement will notice. Costs beyond the included credits are variable, and the monthly cap is yours to set.
For an established e-commerce or omnichannel retail brand with real data volume and a budget for analytics, Dema is a serious, well-documented option. For a smaller store, or any business outside commerce, the entry ticket and the specialisation both argue against it.
- Choosing a selection results in a full page refresh.
- Opens in a new window.