Most teams don’t have a data shortage. They have a stack held together by exports, one-off connectors, and spreadsheets, and every report carries its own math. Alli for Data Foundation connects your sources, defines each metric once, harmonizes naming across platforms and markets, keeps first-party conversion signals flowing, and watches for what breaks. The foundation is deterministic, so your people can check every number.
Last updated
PMG clients in this page’s stories
Each piece of a typical stack works on its own. The exports run. The connectors mostly connect. The warehouse stores everything. Together they leave small problems that add up: a connector that fails quietly, a KPI calculated three ways, a campaign named five ways in five markets, conversions the platforms never saw. Every report, model, and AI answer built on top inherits them.
The harder problem is trust. When a dashboard and a deck disagree, the first forty minutes go to arguing about the data, and the decision slips to next week. People retreat to spreadsheets and check everything by hand. More tools don’t fix that.
Manual pulls. Analysts spend the week collecting data instead of explaining it
Datasources automates ingestion from ad platforms, analytics tools, commerce, and first-party systems
Quiet connector failures. A platform changes its API and reports fill with incomplete data until someone notices
Datasources maintains the APIs and shows connection status and feed health in one place
Metrics defined per tool. ROAS means one thing in the dashboard and another in the deck
Semantic Layer defines each KPI once for every team, region, and currency
Naming that varies. One campaign carries five names in five markets
Categorizations, part of the Semantic Layer, maps variations into one plain-English taxonomy by rule
Platforms that disagree. A view on TikTok isn’t a view on YouTube
Shared entities, calendars, and categories let channels compare on the same terms
Lost conversion signals. Browser tracking misses events, so platforms optimize on a partial picture
Audience Gateway sends first-party conversion signals server to server to Meta, Google, TikTok, and Pinterest
Problems found late. Nobody knows a feed stopped until a report looks wrong
Alli Alerts flags anomalies, pacing issues, and feed failures in Slack
AI on shaky definitions. An AI answer about revenue conflicts with the dashboard finance uses
Ask Alli and the agents in Alli answer from the same definitions
Alli for Data Foundation is five steps that build on each other. Datasources, the Semantic Layer, Categorizations, and Audience Gateway are deterministic: the same inputs produce the same outputs, there is no generative AI in them, and people write every rule. Alerts sits beside them. It uses PMG’s data science models, and it reads the data without changing it.
Sources flow in automatically from ad platforms, analytics, commerce, and your own first-party systems. API changes are handled, and feed health is visible.
KPIs and business logic are set once, and performance data is modeled against shared entities and calendars.
Rules map platform labels and naming conventions into one taxonomy, by channel, funnel stage, campaign type, region, and product line.
Conversion and audience signals go server to server to each platform’s Conversions API from one managed implementation.
Anomaly detection and custom rules flag outliers, pacing issues, broken URLs, and feed failures.
Five moments from the foundation, each shown by the problem it removes.
Pick a source from the catalog, authorize it, and ingestion starts. Datasources keeps the connection healthy as platforms change their APIs, and it brings first-party systems such as Salesforce in beside media data.
The Semantic Layer holds your KPIs and business logic. Change a definition once and it applies in dashboards, Data Explorer, Ask Alli, alerts, and planning tools, with calendars and currencies handled the same way in every region.
Picture a Summer Sale campaign that five markets name five different ways. A Categorizations rule maps every variation into one Season category, so the global team sees one campaign. Rules replace manual tagging, and they keep working when naming discipline slips.
Audience Gateway captures site events on the server side and sends them to each platform’s Conversions API from one managed implementation. Platform API changes are maintained for you.
Many performance problems turn out to be data problems. When an alert fires, Datasources shows whether the feed is current. If it is, the issue is in the campaign or the platform. If it isn’t, fix the connection first.
Built so your people define the numbers and Alli keeps them consistent.
There is no generative AI in the data layer. A definition changes only when a person changes it on purpose, and every number traces back to its source. Alli does the ingestion, the rule-based mapping, and the watching at a scale no team could do by hand. Your analysts, finance partners, and data leads decide what the numbers mean. We scale expertise, not replace it.
Connect. Ingests from connected sources, maintains API connections, and shows feed health
Choose which sources to connect, and authorize the accounts
Define. Applies each KPI definition the same way for every team, region, and currency
Agree on the definitions across marketing, finance, and operations, and change them on purpose
Harmonize. Applies categorization rules to platform labels and naming
Write and review the rules, and choose categories that match how the business is organized
Capture. Sends first-party signals to each platform and maintains the API connections
Decide which platforms get which signals, and own consent and privacy decisions
Watch. Flags anomalies, broken feeds, and pacing issues
Decide whether it’s a data problem or a campaign problem, and fix it at the source
This page shows how the foundation gets built. Data You Can Trust explains why teams rely on it: a deterministic data layer where mappings and definitions are written down, version-controlled, and traceable, with AI working on top alongside your people. The two pages describe the same products from different angles.
Walk into reviews where every team’s numbers agree, and spend the time deciding.
Work from the same definitions as marketing, so budget conversations start from agreement.
Stop assembling data and rebuilding mapping logic in spreadsheets and SQL for every report.
Compare markets fairly, with consistent definitions across regions and currencies.
Hand off API maintenance and Conversions API builds, and keep clear control over where signals go.
Product lines buried in technical campaign names across many accounts, grouped by rule into one view.
Tie media activity to pipeline by bringing CRM data in beside platform data.
Most companies piece their data together with spreadsheets, one-off connectors, and a warehouse that only engineers can use. Each part works on its own. Alli doesn’t replace your warehouse. It adds the connective layer around it.
| Spreadsheets and manual pulls | Warehouse plus custom connectors | Alli for Data Foundation | |
|---|---|---|---|
| Connecting sources | Exports by hand | Engineers build and fix each connector | Managed ingestion, with API changes handled for you |
| Definitions | Different in every file | Defined in code, hard for business teams to see | One set of definitions every team and tool uses |
| Source naming | Renamed by hand | Custom mapping per source | Rules map every source into one taxonomy |
| Monitoring | Someone notices | Separate tooling, if any | Anomaly detection and custom alerts in Slack |
| Lost conversion signals | Not addressed | Separate tracking projects | Managed server-side tracking and Conversions APIs |
| Role of people | Do the reconciling | Wait on engineering | Define the numbers, with the plumbing handled |
Start with the problem costing the most time or trust. Each entry point connects to the rest of the foundation later.
| If your biggest problem is... | Start with | First step |
|---|---|---|
| Manual pulls and a backlog of connector work | Datasources | Connect the platforms behind your weekly report, and retire the manual pull |
| Reports that don’t match | Semantic Layer | Pick the KPI that causes the most arguments, agree one definition, and set it |
| Naming that varies by market or platform | Categorizations | Map one dimension first, such as season or product line |
| Conversions missing in ad platforms | Audience Gateway | Route the platform where you spend most and trust the data least, and compare before and after |
| Problems found days late | Alli Alerts | Turn on no-spend and feed alerts for your highest-spend campaigns |
How Alli connects, defines, and watches your data, with your experts in charge.
Alli for Data Foundation is a set of Alli products that connect your sources, define each metric once, harmonize naming across platforms and markets, keep first-party conversion signals flowing, and watch the data for problems. It combines Datasources, Semantic Layer, Categorizations, Audience Gateway, and Alerts, so every team and tool works from the same numbers.
Each piece can work while the whole disagrees. Exports arrive on different schedules, connectors fail quietly, and every tool defines ROAS its own way. When two reports disagree, meetings turn into debates about the data. People go back to spreadsheets, and everything built on the stack, including AI, inherits the doubt.
No. Alli works alongside the systems you already use, including warehouses like Snowflake and BigQuery. A warehouse stores data. Alli adds ingestion, shared definitions, and monitoring on top, so the stored data means the same thing to every team and is ready for reporting, planning, and action.
A semantic layer is a shared set of business definitions that sits between raw data and the people and tools that use it. It decides once what ROAS or conversion means. Marketing needs one because every platform reports differently and every market names things differently. Without shared definitions, comparisons turn into reconciliation.
Categorizations is now part of Alli Semantic Layer. It still maps fragmented platform labels and naming conventions into one plain-English taxonomy, using rules instead of manual tagging. Bringing it into the Semantic Layer puts KPI definitions and category mapping in one place, so a campaign is classified the same way everywhere it appears.
No. Datasources, the Semantic Layer, Categorizations, and Audience Gateway are deterministic: the same inputs give the same outputs, and people write every rule. Alli Alerts sits beside them, uses PMG’s data science models to flag anomalies, and reads the data without changing it.
Datasources manages API updates for connected platforms as part of the service. When a platform renames fields or changes its reporting structure, the connection is maintained so data keeps flowing, and connection status shows where each feed stands. Your engineers don’t rebuild pipelines every time a platform ships an update.
Browser tracking misses events, so platforms optimize on a partial picture. Audience Gateway sends first-party conversion and audience signals server to server to Meta, Google, TikTok, and Pinterest through their Conversions APIs. It doesn’t replace consent management or legal review. It gives privacy teams a more controlled path for the signals.
Yes. Ask Alli and the agents inside Alli answer from the Semantic Layer, so an answer about revenue matches the dashboard finance already uses. The Alli MCP Server extends the same definitions to AI assistants outside Alli, within each user’s permissions. Shared definitions are what keep AI answers consistent with each other.
Alli is available to PMG clients. Marketing, finance, analytics, and data teams work in the same platform PMG’s experts use, so both sides start from the same sources and definitions. Book a walkthrough and we’ll map your current sources, definitions, and trouble spots to the Alli products that address them.
Book a walkthrough and bring the KPI your teams argue about most. We’ll show you how it’s defined once, how your sources connect, and how the same number shows up everywhere. Alli is available to PMG clients.