Magnum RadiusEst. Ranchi · MMXXV
10 · Analytics & Measurement

Measurement that changes decisions, not just dashboards.

GA4, server-side tracking, self-attribution, warehouse (BigQuery / Snowflake / ClickHouse), and dashboards designed for the person who has to sign the ad budget.

Live · 10 · Analytics & Measurement
Positioning

Why this matters now.

Most Indian mid-market brands have a GA4 install, a dashboard someone made in Data Studio in 2022, and no idea if any of the numbers reconcile. Meta reports one number, Google reports another, the CRM reports a third, the finance team runs a fourth in Excel, and the CEO trusts none of them. This is the standard state, and it is the wrong state.

We fix it in three moves. First, a real tracking spine: GA4 rebuild if needed, server-side event architecture (GTM SS or a custom Cloudflare Worker), offline conversion imports, and self-attribution survey on your thank-you page. Second, a warehouse (BigQuery or Snowflake — pick one, we'll help) where every source lands and gets reconciled. Third, an executive dashboard that reads like a one-page memo, not a 40-panel Data Studio wall.

For D2C brands, this typically pays back inside quarter one — we regularly find 8–15% of revenue that platforms had been mis-attributing, which means real ad spend gets reallocated. For B2B and SaaS, the payoff is longer but larger: real closed-won attribution instead of the perennial 'sales says one thing, marketing says another' loop.

We are honest about the trade-offs. Server-side tracking costs real money and adds real complexity. If your business is under ₹1cr/month in revenue, you probably don't need it yet — you need better product-market fit. We will tell you if we think you're over-buying data infrastructure.

What you actually get

Deliverables, spelled out.

06 · outputs
01

Server-side tracking spine

Google Tag Manager Server-Side or a Cloudflare Worker collecting events. Sent to Meta CAPI, Google Ads, GA4, and your warehouse in one place.

02

Self-attribution survey

Post-purchase / thank-you page survey. Reconciled against platform-reported data monthly. The one attribution signal that platforms can't game.

03

Data warehouse setup

BigQuery / Snowflake / ClickHouse. Fivetran or Airbyte for ingestion. dbt for transformation. Modelled tables your BI tool can query.

04

Executive dashboard

One-page dashboard for the founder / CFO / board. Not 40 charts. 6 numbers that matter. Refresh cadence set to weekly.

05

Operator dashboard

Deeper dashboard for the marketing operator. Channel-level, campaign-level, creative-level. Still designed, not a raw pivot table.

06

Attribution model

Multi-touch attribution — data-driven where you have volume, position-based where you don't. Reconciled against self-report survey monthly.

Capability map

The whole surface area.

Every sub-capability lives on the map. Click one and we'll build a working prototype in a week — we don't handwave scope.

TrackGA4GTM SSCAPIOfflineWarehouseBigQuerySnowflakeClickHouseFivetranModeldbtAttributionLTVCohortsVisualiseLookerMetabasePresetSupersetAttributionData-drivenPositionSelf-reportMMMGovernConsentDPAPIIRetentionData
Micro-services

4 deep specialisations. Each its own page.

Click any tile to open a full engagement brief — deliverables, process, KPIs, and honest FAQs.

04 · clickable
Solution architecture

From input to outcome.

Signal-in / Impact-out
01 · Signal inAd platform eventsWebsite eventsCRMPayment / order dataSurvey responses02 · CaptureGTM Server-SideCloudflare WorkerOffline importsFivetran / Airbyte03 · StoreBigQuery / SnowflakeRaw layerStaging (dbt)Marts (dbt)04 · ModelAttributionLTVCohortForecast05 · DecideExec dashboardOperator viewWeekly memoBoard deck

Foundational engagements: 6–10 weeks. Ongoing warehousing: retainer.

Engagement process

6-step engagement, no filler.

  1. 01Measurement auditTwo-week audit of your current tracking, warehouse, and attribution setup. Delivered as a written memo with prioritised fix list.
  2. 02Tracking spine buildWeeks 3–5: GA4 rebuild if needed, GTM Server-Side or Worker deployment, Meta CAPI, Google offline conversion imports, self-attribution survey deployed.
  3. 03Warehouse setupWeeks 4–8: BigQuery / Snowflake, Fivetran/Airbyte ingestion of ad platforms, CRM, order data. dbt models for staging and marts.
  4. 04Attribution modelWeeks 6–8: multi-touch attribution model calibrated against self-report survey. Confidence intervals published, not hidden.
  5. 05Dashboard layerWeeks 8–10: exec dashboard (6 numbers), operator dashboards (channel, campaign, creative), weekly ops memo template.
  6. 06Ongoing reconciliationMonthly: platform-reported vs. self-report vs. warehouse reconciliation. Discrepancies investigated, models re-calibrated.
Benchmarks we hold ourselves to

The numbers we ship.

0
Weeks · foundation build
0
Numbers · exec dashboard
0
Attribution sources reconciled
0%
Server-side coverage · new events
Radical honesty

What we won't report.

FAQ

Questions smart clients ask.

Only if you're doing enough revenue that reconciling platform data actually pays back. For most D2C brands that's ₹1cr+/month in revenue, and for most B2B/SaaS it's ₹40L+/month ARR. Below that, a well-configured GA4 + spreadsheet is fine.

BigQuery for most cases (Google Cloud consolidation, generous free tier). Snowflake if you have a data team already fluent in it. ClickHouse for event-heavy workloads at scale. We'll advise honestly per project.

No — we build the foundation and hand over cleanly. We're not trying to be your permanent data function. If you already have a data team, we work alongside them and often accelerate their roadmap by two quarters.

GDPR-compliant consent management (Cookiebot / OneTrust integration), PII redaction at ingestion, DPA agreements with every vendor. Compliance is built in, not bolted on.

Foundation build: ₹12–35L one-off. Ongoing warehousing retainer: ₹2–6L/month depending on ingestion volume and dbt model complexity.

Next step

Kick off with a 30-minute strategy call.

We diagnose live, on the call. No decks. If we're the wrong fit we'll say so — and point you to who isn't.

Book the call →