Product portfolio debates stalled — Segment, SAP & Zendesk feed Snowflake for cited memos

Overview

A device maker’s product portfolio reviews were contentious because evidence for usage, margins, and support burden lived in different systems and formats. Teams brought competing charts to leadership, and decisions stalled or looped back for rework. Intelligex created a governed dataset that blended product telemetry from Segment, gross margin extracts from SAP, and support tickets from Zendesk, then introduced a decision memo template in Notion that required cited charts and links back to the governed source. Leadership meetings shifted to evidence-backed choices with clearer trade-offs, fewer circular debates, and a consistent record of why decisions were made.

Client Profile

• Industry: Consumer and enterprise devices
• Company size (range): Mid-market to enterprise with multiple product lines
• Stage: Scaling product portfolio and regional variants
• Department owner: Strategy, Analytics & Executive Leadership (Portfolio & Corporate Strategy)
• Other stakeholders: Product Management, Hardware Engineering, Support/Customer Success, Finance/FP&A, Sales Operations, IT/Data, Legal & Compliance

The Challenge

Portfolio rationalization required a shared view of product usage patterns, unit economics, and support load. In practice, usage telemetry echoed different client schemas, margins came from finance extracts with SKU and region nuances, and support burden sat in ticketing systems with inconsistent categories. During reviews, Product and Finance presented different cuts of the same questions. Meeting time went to reconciling definitions and screenshots instead of evaluating options such as retire, refresh, or invest.

Tooling existed but was siloed. Segment collected device events with varying traits, SAP held the cost and margin picture by SKU and region, and Zendesk tracked ticket volumes and themes. Analysts exported data into spreadsheets, re-keyed joins, and introduced small mismatches in SKU codes, firmware versions, or time windows. Leadership lacked a single, governed dataset and a standard memo format that forced citation to the same source of truth.

Why It Was Happening

Identity and taxonomies were fragmented. Telemetry used device IDs and product slugs, SAP applied SKU hierarchies and region price books, and Zendesk relied on agent-entered categories. Without a mastered product map and harmonized time windows, evidence stitched together manually produced defensible but incompatible views.

Governance arrived late. There was no requirement to cite metrics back to a governed dataset, no common definitions for “active device” or “ticket rate,” and no approval step that bound recommendations to agreed inputs. Notion hosted narratives, but templates did not enforce links to validated charts or require review by Finance and Support before executive sessions.

The Solution

We built a governed analytics layer that blended telemetry, financials, and support signals into a conformed model and embedded it in the decision process. Segment events landed in the warehouse with standardized traits; SAP margin tables and SKU master data joined on a mastered product map; Zendesk tickets were categorized and linked by device family and firmware version. dbt encoded definitions and validations, and Power BI rendered certified views. Notion memos pulled in these charts and required inline citations back to the dataset. A light approval gate captured Finance and Support sign-off for key assumptions before executive discussion. Existing systems remained in place; the orchestration unified data, definitions, and decisions.

• Telemetry ingestion from Segment with standardized event traits (product family, firmware, region) and late-arriving data handling
• Margin and SKU hierarchy from SAP with regional price lists and cost elements joined via mastered product IDs (SAP Help Portal)
• Support ticket data from Zendesk including categories, device metadata, and resolution codes
• Conformed warehouse model in Snowflake with product and region dimensions, harmonized calendars, and time-stamped snapshots (Snowflake)
• dbt transformations and tests encoding definitions for active devices, ticket rate, and margin by SKU and cohort (dbt)
• Certified Power BI views for usage cohorts, margin ladders, and support intensity with drill-through to source records (Power BI)
• Master product map with stable IDs and alias history to link telemetry slugs, SAP SKUs, and support tags
• Notion decision memo template with required chart embeds, citations to certified views, and approval tasks (Notion Help)
• Human-in-the-loop review: Finance validates margin breaks; Support validates ticket categorization and outliers before memos publish
• Role-based permissions, data minimization for device identifiers, and audit logs for memo approvals and dataset versions

Implementation

• Discovery: Cataloged telemetry schemas and key events from Segment, SAP tables for margin and SKU master, and Zendesk fields for ticket type and device metadata. Mapped product hierarchy variants, regional price lists, and common date cut-offs. Reviewed recent portfolio debates to identify conflicting definitions and repeated pain points.
• Design: Defined the mastered product map and alias strategy, conformed dimensions for product and region, and a shared calendar. Authored dbt models and tests for device activity, ticket rate, and margin. Designed Power BI certified views and a Notion memo template with required chart embeds and citation fields. Specified Finance and Support review steps and ownership.
• Build: Implemented Segment and Zendesk connectors into Snowflake; loaded SAP extracts; built dbt models, tests, and documentation; created the master product map with stewardship. Published certified Power BI datasets and reports. Built the Notion template with properties for decision type, affected SKUs, links to charts, and a lightweight approval workflow.
• Testing and QA: Replayed recent quarters to reconcile device activity, margin, and ticket rates against known baselines. Validated product joins and alias coverage. Tuned dbt tests to flag common edge cases such as retired SKUs and firmware forks. Verified Power BI drill-through and Notion embed behavior. Dry-ran memo approvals with Finance and Support.
• Rollout: Launched read-only dashboards alongside legacy spreadsheets; introduced the memo template for a pilot set of portfolio decisions. After validation, required cited charts and pre-review for memos heading into executive forums. Expanded product coverage and retired redundant spreadsheet workflows.
• Training and hand-off: Delivered short guides for Product on interpreting cohorts and ladder charts, for Finance on margin definitions and approval steps, and for Support on ticket categorization and outlier reviews. Assigned stewardship for the product map, dbt definitions, and the memo template with a change-control cadence.

Results

Portfolio reviews centered on the same charts and definitions. Product, Finance, and Support referenced a single, governed dataset with drill-through to telemetry cohorts, margin ladders, and ticket intensity. Notion memos embedded certified views and carried citations and approval stamps, so debate shifted from whose spreadsheet was right to which option best matched strategy and constraints.

Decisions moved faster because evidence traveled with the recommendation. Leaders compared SKUs and variants on a like-for-like basis, and contentious topics—such as region-specific support spikes or firmware-related usage dips—were grounded in data with accountable owners. Follow-ups referenced the same snapshot through links in the memo, reducing rework and ensuring subsequent iterations started from shared facts.

What Changed for the Team

• Before: Teams brought competing charts and definitions. After: Certified views and dbt-encoded definitions became the single source of truth.
• Before: Telemetry slugs, SKUs, and support tags did not align. After: A master product map harmonized identity across systems.
• Before: Memos were narrative-only. After: Notion templates required embedded, cited charts and pre-review by Finance and Support.
• Before: Edge cases were discovered during executive meetings. After: Tests and human-in-the-loop reviews flagged anomalies ahead of time.
• Before: Decisions were hard to revisit. After: Memos linked to the governed snapshot and approvals, creating a durable record.

Key Takeaways

• Unify telemetry, financials, and support under a conformed model; rationalization requires consistent identity and time windows.
• Encode definitions in transformations and certify views; governance in code and BI beats spreadsheet debates.
• Require citations in decision memos and pre-review by Finance and Support; context and controls improve speed and quality.
• Keep Segment, SAP, Zendesk, and Notion; orchestrate them with a master map, tests, and light approvals rather than replatforming.
• Preserve auditability with snapshots and approvals so choices and their evidence can be revisited without rework.

FAQ

What tools did this integrate with?
We ingested device events from Segment, margin and SKU master data from SAP, and support tickets from Zendesk into Snowflake. Definitions and tests ran in dbt, certified charts were delivered in Power BI, and decisions were captured in a Notion template with required citations.

How did you handle quality control and governance?
We created a master product map to align telemetry slugs, SAP SKUs, and support tags. dbt encoded definitions and tests for active devices, ticket rate, and margin, with failures routed to owners. Certified datasets in Power BI enforced consistent visuals. Notion memos required links to certified charts and went through Finance and Support review before executive forums. Snapshots and approvals created an audit trail.

How did you roll this out without disruption?
We ran dashboards in parallel with existing spreadsheets and introduced the decision memo template in a pilot. After validating joins, definitions, and review steps, we required cited charts for memos bound for leadership. Existing tools stayed in place; the new layer harmonized data and standardized the decision process around them.

How were telemetry and ticket data unified with financials?
A conformed model aligned product and region dimensions, and a master product map linked device IDs and slugs to SAP SKUs and Zendesk tags. Calendars were harmonized to shared cut-offs. Cohorts and ticket rates were computed in dbt and joined to margin ladders at the SKU and region level, with drill-through to source records for validation.

How did you address privacy and sensitive identifiers?
Telemetry was minimized to necessary traits, and device identifiers were hashed or abstracted at the cohort level for decisioning. Role-based permissions limited access to detailed records, and certified views exposed only aggregated metrics needed for strategy. Access and approvals were logged to meet compliance and audit expectations.