Ducklytics Book a call

AI-first data platforms

Get your data ducks in a row.

Ducklytics designs and runs your end-to-end data platform: ingestion, lakehouse, dbt models, dashboards, and an AI layer that lets anyone ask the data a question. Built fast, tested properly, handed over with docs.

Book a call

Marketing performance

Last 8 weeks · sample data
Pipeline
$1.8M
▲ 12%
ROAS
4.1×
▲ 0.3×
Freshness
06:00
▲ on time
030k60k90k W1W2W3W4W5W6W7W8
ASK“Which channel drove W8?” → Paid search, 41% of new pipeline.
412 dbt tests passed

The stack

Build it your way. We handle the rest.

Tap the tools you already use or want. The yellow ones are our default picks. Whatever you choose, we connect every layer so data moves from source to answer without hand-offs.

IngestGet data in
StoreLakehouse & warehouse
TransformModel & test
OrchestrateRun on time
ServeReports people open
AI layerAsk, predict, automate
Ship itInfra & CI/CD
Your stackDatabricks · dbt · Airflow 3 · Claude
Book a call with this stack

Data flow

From raw to report in five hops.

A medallion lakehouse driven by YAML config. Every hop is tested, logged and owned, so a broken source never quietly breaks a dashboard.

SRC

Sources

SaaS apps, ads platforms, product databases, files.

  • CRM & ads APIs
  • Postgres CDC
  • S3 CSV / Parquet
BRZ

Bronze

Raw, as-is, with audit columns.

  • Auto Loader
  • schema drift alerts
  • _rescued_data
SLV

Silver

Clean, deduped, privacy-safe.

  • CDC merge
  • PII → SHA-256 user_id
  • DQ checks
GLD

Gold

Business logic in dbt.

  • facts & dimensions
  • metric definitions
  • 1 model = 1 task
USE

Serve & AI

Where people get answers.

  • BI dashboards
  • AI analyst (Genie / Claude)
  • reverse ETL
Orchestrated byAirflow 3 datasetsdbt + CosmosUnity Catalog lineage

Data models

Modeled like we mean it.

Star schemas with clear grains, keys and tests. Here is the core of a marketing attribution mart: one fact per touchpoint, joined to the dimensions marketing actually filters by.

dim_contacts dim
  • PK contact_idstring
  • account_idstring
  • lifecycle_stagestring
  • first_touch_attimestamp
fct_opportunities fact
  • PK opportunity_idstring
  • FK contact_idstring
  • amount_usddecimal
  • closed_won_attimestamp
fct_touchpoints grain: 1 touch
  • PK touchpoint_idstring
  • FK contact_idstring
  • FK campaign_idstring
  • FK channel_idstring
  • FK date_keyint
  • touch_weightfloat
  • attributed_revenuedecimal
dim_date dim
  • PK date_keyint
  • week_startdate
  • fiscal_quarterstring
dim_campaigns dim
  • PK campaign_idstring
  • campaign_namestring
  • platformstring
  • spend_usddecimal
dim_channels dim
  • PK channel_idstring
  • channel_groupstring
  • is_paidboolean
unique · not_null on every keyrelationships tested for every FKdocs generated and versioned

Reports

Dashboards people actually open.

A few honest numbers, the trend behind them, and an AI analyst for the follow-up question. Every metric traces back to a tested model.

Revenue marketing · Q3

Last 90 daysAll regionsSample data
Pipeline sourced
$2.41M
▲ 14% vs Q2
ROAS
4.1×
▲ 0.3× vs Q2
Win rate
23%
▼ 1 pt vs Q2
Pipeline uptime
99.7%
▲ 0.2 pts

Pipeline sourced by channel

Paid search$820k Organic$640k Events$410k Paid social$330k Partners$210k

MQL → SQL conversion

Weekly, last 12 weeks
10%20%30% 29%

Ask Ducklytics

Why did CAC jump in September?
Paid social spend rose 38% while SQLs stayed flat. Joined fct_touchpoints × dim_campaigns. Want the campaign breakdown?

Sample data for illustration. Your dashboard uses your numbers and your metric definitions.

How we work

Live in weeks, not quarters.

01 · Week 1

Map the pond

Audit your sources, agree on the ten metrics that matter, sketch the model and the stack.

02 · Weeks 2–6

Build the flow

Pipelines, lakehouse layers, dbt models with tests, and the first dashboards in your team's hands.

03 · Week 7 on

Hand over the keys

Docs, training, an AI analyst on your data, and support for as long as you want it.

What AI-first means here: AI drafts models, tests and docs from your schema, and an AI analyst answers questions in plain English. Engineers review every line before it ships.

Domain knowledge

We speak your business, not just SQL.

Years inside these teams means we already know the metrics, the edge cases and the questions your stakeholders will ask next.

Marketing

Know where every dollar lands.

Media-mix modeling, multi-touch attribution, A/B tests and campaign ROI, from one marketing mart.

CACROASMQL → SQL

Sales & revenue

Pipeline you can forecast.

Lead scoring, sales recommendations and clean CRM data across HubSpot and Salesforce.

Win ratePipeline velocityACV

Product & growth

See who stays, and why.

Funnels, retention cohorts, promotions, gamification and loyalty programs that move the numbers.

ActivationRetentionLTV

Ads platforms & trust

Clean inventory, honest clicks.

Advertiser reporting plus early detection of spam, fraud and content that breaks policy.

CTRFill rateFraud rate

Customer service

Faster answers, fewer tickets.

Ticket analytics and automation that cut resolution time by 60% for one support team.

Time to resolveCSATBacklog

Logistics & insurance

Bookings, policies, claims.

Booking behavior and demand by area for on-demand delivery; policy and claims reporting for insurance.

BookingsUtilizationLoss ratio

Ready to get your ducks in a row?

Tell us where your data lives and what you want to know. We'll come back with a plan and a timeline within two days.

Book a call thuy@ducklytics.com