A semantic layer so revenue means the same thing on every dashboard, in every report and in every AI answer.
Relate your tables once, describe what the columns mean, define margin, churn and DSO in one place, and mark which tables are trusted. Every chart, formula and question to Klayara AI then starts from the same definitions.
Relationships set onceOne metric catalogCertified tables and ownersSynonyms for business termsReused by Klayara AI
What is a semantic layer in BI?
A semantic layer is the shared set of business definitions that sits between raw tables and the people using them: how tables relate, what each column means, and how metrics such as revenue or margin are calculated. In Klayara you relate tables once, describe columns, define metrics in one workspace catalog and certify trusted tables, and every dashboard and AI answer reuses them.
Why a semantic layer gives you a single source of truth
A single source of truth means two people asking the same question get the same number. Without a semantic layer, each dashboard author joins the tables and writes the margin formula in their own way, and the finance pack and the sales dashboard disagree by a few percent. The meeting then spends its time on whose number is right rather than on what to do.
Klayara keeps the definitions in one place. Relationships, column meanings and metrics are set once for the workspace and reused by every chart, formula, scheduled report and AI answer. When a definition changes, it changes everywhere, and nobody has to hunt down the copies.
Data modeling for BI: table relationships set once
Data modeling for BI starts with relationships: telling the system that orders belong to customers, invoices belong to orders and branches belong to regions. In Klayara you relate customers to orders to invoices once, and every chart and formula can then use columns from both sides without anyone writing a join.
Lookup columns pull a value such as a customer's segment or an item's category from a related table, so a sales chart can be split by segment even though segment lives in the customer table. When you open a row, its related rows from other tables appear beside it, such as a customer's open invoices next to their details.
- Relationships: customers to orders to invoices, set once
- Lookup columns: bring segment or category across without a join
- Related rows: see a customer's invoices beside the customer
- Friendly names: show "Acme Ltd" instead of customer number 10427
Business definitions: what each column means
Business definitions tell Klayara what a column is and how it should behave, so authors do not set the same options on every chart. You can say a column is a country, a phone number or a postcode, how to format it, how to total it by default, and that months sort January to December rather than alphabetically.
Clock times and durations can be totalled and averaged like numbers, which matters for shift hours or ticket handling time. A cell holding several values, such as "red, blue, green", can be filtered by any or all of them and split so each value gets its own bar. Column profiling shows gaps, blanks, top values and spread for every column in one click, so you can see what you are describing before you describe it.
A metrics layer: governed metrics defined once
A metrics layer is the part of the semantic layer that holds calculations. In Klayara it is one metric catalog for the whole workspace: define gross margin, churn, days sales outstanding or average order value once, and reuse it on every dashboard so every chart agrees.
Metrics are written with the same 138 formula functions used everywhere else in Klayara, which read the way a spreadsheet does. Year-to-date and quarter-to-date follow your own financial year, such as April to March. A KPI card can also show a metric defined on another dashboard, so the receivables figure on the CEO's overview is the same one finance maintains.
- Formulas are checked as you type, with errors explained in plain words
- Authors cannot use a column they are not allowed to see, and get no hint it exists
- Table settings keep a version history you can restore from
Certified tables, owners and synonyms
Certification tells readers which tables to trust. In Klayara you can mark a table as certified, name the business owner responsible for it, and add synonyms so that someone searching for "turnover" finds revenue and someone typing "debtors" finds receivables.
Plain-English descriptions help the next person, too. On the Growth plan and above, one click drafts descriptions of a table and its columns for a person to review and approve, which makes describing a large ERP schema a matter of checking rather than writing from scratch. Projects group related tables and dashboards, and search works across everything.
A semantic layer for AI: answers that match the dashboards
An AI assistant is only as consistent as the definitions it uses. Klayara AI, available on the Growth plan and above, reuses your saved metrics, synonyms and certified tables, so when someone asks for gross margin by region last quarter, the answer uses the same margin definition as the finance dashboard.
Because the semantic layer carries meanings and relationships, the assistant can tell that "debtors" means receivables and that region comes from the branch table. It returns a real chart built from your data, never an invented number, and it sees only what the person asking may see. Admins choose which tables the AI may use and can give it company-wide instructions.
From first click to everyday use.
Relate the tables
Connect customers to orders to invoices once, and add lookup columns for values such as segment or category.
Describe the columns
Set friendly names, column meanings, formats, default totals and sort orders, with column profiling to check the data.
Define the metrics
Add margin, churn, DSO and other shared metrics to the workspace catalog, following your financial year.
Certify and reuse
Mark trusted tables, name owners and add synonyms; dashboards, reports and Klayara AI reuse it all.
What goes in the semantic layer, and who benefits
| What you define | Example | Who benefits |
|---|---|---|
| Relationships | Customers to orders to invoices | Dashboard authors, who stop writing the same joins |
| Lookup columns | Customer segment on every sales line | Sales and marketing, who split results by segment |
| Friendly names | "Acme Ltd" instead of customer number | Every reader of every chart |
| Column meanings and formats | Country, currency format, months sorted January to December | Authors, who stop fixing the same settings per chart |
| Shared metrics | Gross margin, churn, DSO defined once | Finance and leadership, who see one number everywhere |
| Certified tables and owners | "Sales ledger, certified, owner: Finance" | Analysts choosing which table to trust |
| Synonyms | "Turnover" and "debtors" map to revenue and receivables | Search users and Klayara AI |
Everything included: model and describe.
Relate tables once, describe what the columns mean, and every chart, formula and AI answer reuses it. See every feature.
Relationships, set once
Relate customers to orders to invoices once; every chart and formula can then use fields from both sides.
Lookup columns
Pull a value such as a customer's segment or an item's category from a related table without building a join.
Related rows
Open a record and see its related records from other tables beside it.
Friendly names
Show "Acme Ltd" instead of the customer number stored in the table.
Column meanings and formats
Tell Klayara a column is a country, phone number or postcode, how to format it, how to total it by default, and that months sort January to December.
Times and durations that add up
Clock times and durations can be totalled and averaged like numbers.
Multi-value columns
A cell holding "red, blue, green" can be filtered by any or all of its values, and split so each value gets its own bar.
Certified tables, owners and synonyms
Mark trusted tables, name a business owner, and add synonyms so "turnover" finds revenue.
AI-drafted descriptions
One click drafts plain-English descriptions of a table and its columns for a person to review and approve.
Growth and EnterpriseColumn profiling
One click shows gaps, blanks, top values and spread for every column; click any part to filter.
Version history for tables
Table settings keep a version history you can restore from, just like dashboards.
Projects and search
Group tables and dashboards into projects, search across everything, and pick up recent work from the home page.
Semantic layer questions, answered.
Something else on your mind? Ask us directly — a person answers.
Do I need a data engineer to set up a semantic layer in Klayara?
No. Relationships, column meanings, friendly names and certified tables are set in the product without code, and metrics are written with spreadsheet-style formulas that are checked as you type. Errors are explained in plain words before you save.
What is the difference between a semantic layer and a metrics layer?
A metrics layer is one part of a semantic layer. The semantic layer covers relationships between tables, column meanings, friendly names and certification as well as metrics. The metrics layer is the shared catalog of calculations, such as margin or churn, which in Klayara is defined once per workspace.
Can different departments use different definitions?
They can, but the point of the catalog is to agree once. Shared metrics live in one workspace catalog so every dashboard uses the same margin or churn. Where a team genuinely needs a variant, define it as a separately named metric so readers can see the difference.
Does the semantic layer respect row-level security?
Yes. Definitions describe how to calculate a number, not who may see it. Row-level security and column masking, on the Growth plan and above, are applied to every chart, download, report and AI answer that uses those definitions, and formula authors cannot reference columns they are not allowed to see.
Which plan includes the semantic layer?
Relationships, lookup columns, friendly names, column meanings, certified tables, synonyms, column profiling and the metric catalog are on every plan, including Starter. AI-drafted descriptions and Klayara AI, which reuses the definitions in its answers, are on Growth and Enterprise.
Does it work with live connections as well as imported tables?
Yes. Relationships, column meanings and metrics apply to a table whether it is queried live from your database or imported into Klayara, so a dashboard can mix both and still use one set of definitions.
How does a semantic layer help AI answers?
It gives the AI the same definitions people use. Klayara AI reuses your saved metrics, synonyms and certified tables, so it knows that "debtors" means receivables and calculates margin the way finance does. Its answers are real charts from your data that match the dashboards.
See it on your own data.
Bring one report your team relies on. We'll rebuild it live in Klayara, with your permissions, so you can judge it on real numbers.
Prefer to see plans first? See pricing