Observability initiative across Product/Tech/Exec layers. Discovery phase to define leadership problem statements. MVP + portfolio/programme metrics surfaced via Looker Studio. Proposed centralised org-wide Observability and maturity model. Specified Tech Hub requirements and leadership visibility requirements.
Impact
Context
Challenge
My role
Approach
Key activities
Purpose
- Align leadership across Tech, Science, and ELT on the scope, priorities and value of Observability investments
- Establish a cross-functional foundation to define clear goals, deliverables, and success metrics for Observability in 2025 and beyond…….
Why now
- Current lack of unified direction risks fragmented tooling, duplication of effort and missed opportunities for insight-driven operations
- Rising complexity across distributed systems and data infrastructure demands scaleable, proactive monitoring solutions
Benefits
- Strategic alignment: Make sure all the hubs work towards and benefit from the same Observability goals
- Cross-hub collaboration: Build a shared language and understanding across Engineering, Product, and ELT
- Accelerated delivery: Discovery resolves ambiguity early on, reduces rework and aligns teams and enables faster, more predictable delivery
Top 5 questions raised by all LT members:
Pipeline progression, status and flow | A fundamental and consistently raised need among all stakeholders was clear visibility into the status, flow and progress of programs or projects through the pipeline stages. This includes metrics like the number of programs in each stage, (breadth view), and understanding the major stage gates and sub-steps. The goal is to get an early signal of where programs are and have a shared understanding across leadership of the current state. |
Attrition, failure points and the why | All leaders expressed a strong interest in understanding where programs or projects were failing or stalling, and crucially, the reasons behind this attrition. This includes metrics on failure rates, and being able to drill down and understand why deviations and failures occurred. Identifying bottlenecks and failure points is key to being proactive |
Cost, resource allocation, and efficiency | Understanding the cost associated with different aspects of the pipeline and OS Hub operations was a universal concern. This includes metrics like cost per program or per stage gate, resource burn per project, how resources are allocated, (e.g. Compute, FTEs) and linking metrics to Financial and budget planning. The objective is often framed around achieving more with the same amount of spend, or identifying the biggest levers for accelerating drug discovery and their associated investment. |
Speed, velocity and turnaround time | The speed and TAT at which programs or experiments move through Workflows and Stage gates was highlighted as a critical operational metric. Leaders want to understand how fast things are moving, (Flux view), track turnaround times for different phases or processes, and see if projects are moving faster or if the system is tracking against objectives related to speed. |
Strategic alignment and business impact | Leaders consistently emphasized the need for Observability to demonstrate the impact of the OS Hub and IWs on higher level company goals, business objectives and the overall pipeline. This includes linking metrics to being on track with company goals, using data for data driven decision making, and understanding the efficiency with which projects are being delivered and seeing how the technology platforms contribute to specific programmes and the broader pipeline story. |
OS Hub 2025 Annual Platform goal
Implement Recursion OS Observability to drive pipeline efficiency without sacrificing scale or quality
Establish unified, longtitudinal metrics (TAT, Flux, Cost, Pipeline status) across industrialized workflows |
Connect leadership views to granular Hit and Hit to lead analytics and demonstrate impact on IW improvement and pipeline status |
Build a durable data and dashboard foundation for cross-hub decision making |
2025 Goals: What we said we would do
Create foundational data products and portfolio sources of truth with clear ownership |
Deliver OSHLT Portfolio dashboards with drill down into Stage gates (Hit, Hit to Lead, LO) and program status |
Enable analytics on pipeline progression, attrition and why, and cost-resource efficiency |
Define a longer term Observability product vision and operating model (roles, cadence, governance) |
2025 Goals: What we achieved - Foundations and leadership visibility
Centralized Portfolio SOT modeled into reliable data products backing OSHLT dashboards |
OSHLT MVP - V1 Leadership dashboard widely adopted for portfolio progression and Stage gate status |
Semantic layer in Looker plus conversational analytics agent to answer portfolio questions self-serve |
Core H2 2025 objective (Portfolio analytics via SOT + data products + dahsboard) substantially completed |
2025 Goals: What we achieved - Stage gate and workflow insights
Hit, Hit to Lead, and LO deep dive analytics initiated with partial delivery (Time in stage, failure modes, merit metrics) |
Early cost tracking threaded into H2L / LO views in partnership with finance |
Clearer program status and progression views across discovery portfolios |
Foundation laid to double down on foundational data + prioritized analytics as a strategic capability |
2025 Goals: Where we fell short - Depth of impact
Did not fully land a clear, repeatable story linking OS tools to program and portfolio outcomes |
Platform differentiation and industrialized workflows vs traditional pharma metrics remained shallow |
Attrition / why did this die analytics were only partially implemented and not yet systemic |
Cost, resource allocation and efficiency analytics across the full portfolio stayed in progress |
2025 Goals: Where we fell short - Operating model and ownership
Centralized ‘OS Observability component’ model proved hard to scale across hubs and teams |
Company wide observability process (shared language, responsibilities, cadences) not fully established |
Reliance on a small core team limited distribution ownership of definitions, metics and dashboards |
Some planned simulations / forecasting and Exscientia aligned work did not reach completed production use |
Net assessment - 2025 outcomes
Core portfolio observability (data products + OSHLT dashboard + semantic layer) is a clear win |
Leadership now has a much stronger, automated view of pipeline progression and spend |
The ‘next layer’ (tool impact, differentiation, attrition / why, cost, efficiency, shared process) remains incomplete |
2025 ends with a strong analytics foundation, but an only partially realized company-wide observability system |
Implications for 2026
Shift from centralized Observability component to distributed ownership with clear standards and support |
Deepen analytics on attrition, cost / efficiency, and tool-programme impact to tell a stronger platform story |
Embed Observability into leadership rhythms (reviews, decision logs, investment calls) rather than one off decks |
Use the 2025 foundation as the data backbone for Goal 9: Platform and tool impact observability |
Made with Bullet