Everyone's revenue number is the same number.
Sightline holds one versioned definition of each metric and serves it to every dashboard, notebook and spreadsheet you already use. The meeting stops being about whose query is right.
Free for 14 days · No card required · 11-minute setup
sightline.app/metrics
Metric catalogue
1 rebuildingCertified
84.1%
Freshness breaches
0.4%
1 of 240, upstream
Median runtime
1:12
Down from 4:06 in June
Data teams already defining metrics in Sightline
- Northwind
- Everline
- Blackford
- Quanta
- Riverstone
- Palladio
The gap
You do not have a dashboard problem. You have an agreement problem.
Four people, four revenue figures, none of them wrong. Each answered a slightly different question and nobody wrote the questions down.
Where it breaks today
The definition lives in whichever query you opened
Every dashboard carries its own filters and its own idea of what counts. Reconciling two of them means reading two SQL files line by line and hoping you spot the WHERE clause that differs.
What Sightline does instead
The definition is the artefact
One metric, one definition, versioned like code. Dashboards, notebooks and exports all resolve to it, so they cannot quietly disagree.
Where it breaks today
Nobody knows what a change will break
Renaming a column is a twenty-minute job and a two-day investigation, because the only way to find what depends on it is to ship it and wait for the complaints.
What Sightline does instead
Lineage down to the column
Before a change merges, Sightline lists every metric, dashboard and export that reads the column — by name, with an owner beside each one.
Where it breaks today
Stale data looks exactly like fresh data
A pipeline fails at 3am and the dashboard renders a perfectly convincing chart of yesterday. The first person to notice is usually in the meeting.
What Sightline does instead
Freshness is a contract
Each metric declares how old it may be. Past the floor it is marked stale everywhere it appears, and the owner hears about it before the meeting does.
The catalogue
Open any number and read where it came from.
The sources it was built from, the definition applied, the checks it passed and everywhere it is published — on one record.
sightline.app/metrics/MET-2214
Net revenue · daily
MET-2214Fresh · 1:12- Read sources0:22
3 tables · 48.2M rows scanned
- Build model0:31
fct_revenue_daily
- Apply definition0:04
net_revenue@v11
- Row-count check0:02
within 0.6% of the 7-day median
- Freshness check0:01
source 14 min behind · floor 60
- Certify0:12
all four checks passed — publish
- Definition
- net_revenue@v11
- Sources
- 3
- Rows
- 48.2M
- Owner
- Finance
01
Definition history
Every version of a metric, who changed it, and what the number did the day it changed.
02
Diff before you merge
See what a proposed definition would have returned for the last ninety days, against what it did return.
03
Query cost per metric
Warehouse spend attributed to the definition that caused it, not to the month.
Capabilities
Built for the team that owns the numbers.
Everything below ships on every plan. The Enterprise tier adds controls, not features.
Certification policy
Write what a metric must satisfy before anyone may build on it, in plain language. Sightline compiles it into checks that run on every merge.
Compiled to
- Named owner
- Freshness floor
- Row-count bounds
- Required reviewers
Review before certified
A definition change that affects a certified metric waits for the owner. They see the diff, the affected dashboards and the ninety-day impact on the number itself.
Column-level lineage
Trace any published figure back through models to the source column, and forward to every dashboard and export that reads it.
Warehouse-native compute
Queries execute in your warehouse against your data. Sightline holds the definitions and never a copy of the rows.
Runs on
- Snowflake
- BigQuery
- Databricks
- Postgres
Sixty native connectors
Read from the warehouse and sources you already have; publish to the BI tools, notebooks and sheets your team already opens. Anything missing takes an afternoon over the SDK.
Reads and publishes
- Harbor
- Beacon
- Kettle
- +57 more
Enterprise controls
SAML SSO, SCIM provisioning, regional data residency, customer-managed keys and a signed DPA.
Getting started
One agreed metric this week, not a six-month modelling project.
No warehouse migration. No rewriting the models you already trust.
- 1
Point it at the warehouse
Read-only credentials and one schema. Sightline maps what is there — it does not ask you to move or reshape anything.
- 2
Import the models you have
Existing dbt or SQL models come across as they are. Sightline reads them for lineage on the first sync.
- 3
Define one metric properly
Pick the number people argue about most. Write it once, certify it, and let the arguments resolve against it for a week.
- 4
Point the dashboards at it
Repoint one dashboard to the certified definition. The rest follow when their owners are ready — there is no cutover.
Connected
It reads from your warehouse and publishes where people already look.
Sixty native connectors, a typed SDK for everything else, and no per-connector pricing.
In production
The numbers our customers report back.
240
Certified metrics, median
Per mature workspace
94%
Dashboards on a definition
Median, first 90 days
0.4%
Freshness breaches
Trailing twelve months
99.98%
Query-layer uptime
Trailing twelve months
Proof
Teams that stopped reconciling in the meeting.
Our Monday review used to open with fifteen minutes of working out whose number was whose. It opens with the number now. Nobody made a decision to stop arguing — there was just nothing left to argue about.
15 minutes back, weekly
Column-level lineage is what let us finally rename things. We had columns nobody dared touch because the blast radius was unknowable. We can see the radius now, so we touch them.
Renames stopped being scary
The freshness contract caught a silently failing pipeline four hours before the board pack went out. That single morning paid for the year, and it has happened twice since.
Caught before the board saw it
Pricing
Priced on metrics, not on the people who read them.
Viewer seats are free on every plan. A number nobody is allowed to look at is not worth defining.
Starter
For a first team agreeing on its core numbers.
You save $120 a year
Start free- Unlimited viewers and editors
- 25 certified metrics
- 10 native connectors
- Column-level lineage and full definition history
- Community support
Growth
Most teams start hereFor data teams serving the whole company.
You save $480 a year
Start free- Everything in Starter
- 500 certified metrics
- All 60 connectors, plus the SDK
- Certification policy and review queue
- Freshness contracts and alerting
- Priority support, 4-hour response
Enterprise
For regulated teams with their own security bar.
- Everything in Growth
- Unlimited metrics, committed pricing
- SAML SSO and SCIM provisioning
- Regional residency and customer-managed keys
- Private deployment in your VPC
- Named engineer and 99.9% SLA
All prices in USD. Warehouse compute is billed by your warehouse, not by us — we never mark it up.
Questions
The things people ask before signing up.
Still unsure? A support engineer answers, not a bot.
Talk to a humanNo. Sightline holds definitions, lineage and metadata; queries compile to SQL and execute in your warehouse under credentials you control. We cache result sets only if you switch caching on, and that cache is regional and expires on your schedule.
Define the number people argue about.
Pick the one metric that starts every disagreement, write it down once, and let a week of meetings resolve against it. That is the whole evaluation.
Free for 14 days · No card required