Beyond Algorithms: Reshaping a Startup's Data Science Strategy for Real-World Impact

Client Background:

A Series B startup in the manufacturing industry was building a proprietary process for inspecting product quality. They had an in-house data science team in place, but needed an external assessment of whether their current approach was optimal and aligned with industry best practices. The company was at a critical juncture in their growth and wanted to ensure that their data science efforts were effectively supporting their business objectives and providing a competitive advantage in their rapidly evolving market.

The Business Challenge

Komodo Technologies was engaged to:

1. Evaluate the modeling techniques of the data science team;

2. Assess their overall scientific approach to product quality assessment;

3. Explore alternative methods and data sources for more efficient problem-solving.

Crucially, time was of the essence for the client. A protracted discovery process culminating in a findings report would not work - they needed Komodo to get up to speed and start delivering results immediately. They asked us to join their sprint process and update them weekly on our findings.

We were given a sample of their data and asked to:

1. gauge the data’s usefulness and effectiveness in solving the product quality issue;

2. propose enhancements, which might include:

   1. different data modeling methods,

   2. analytical process adjustments for superior data inputs;

   3. organizational process suggestions to enhance the value of the existing data science team to the company.

The Komodo Solution

Within 3 days of receiving the data & source code, we were up to speed, participating in meetings, and contributing new ideas and approaches based on our findings. Simultaneously, we began conducting interviews with their stakeholders and subject matter experts to build a more in-depth understanding of their data science approach. We convened weekly with stakeholders and visited the client's offices for further investigation.

We assessed the code and data to learn whether the techniques and technologies applied to the problem were well-suited to solve the challenges that had been posed to the data scientists. In addition, we analyzed the business context to learn whether the data scientists were really working on the areas of R&D that made the best use of their time. Critically, we investigated the impact of past data science work on prior R&D efforts and the contributions of data science to the organization’s shared understanding of the problem at hand.

“In data-intensive projects, it’s crucial to align data scientists’ efforts with organizational goals. ”

In data-intensive projects, it's crucial to align data scientists' efforts with organizational goals. Many companies struggle to obtain actionable insights or tangible results from their data science teams. It's also common for data scientists to become overly focused on specific technical challenges, losing sight of the broader business objectives. The success of the business hinges on data scientists focusing on core business problems, not isolated technical issues.

The Results

Within one month, we had highlighted significant issues with the scientific approach and data quality being employed in their efforts. We identified gaps in the client’s data collection process that were hindering their ability to accurately measure their progress towards a solution. We delivered alternate approaches they might employ for generating better truth data.

We also found ways the data science team could be reorganized and deployed more effectively towards the company’s business goals.

We produced a detailed due diligence report and delivered it in person to talk through our findings. Instead of simply pointing out issues, some of which confirmed suspicions held by the leadership team, we also suggested concrete changes and improvements, and strategies for rolling out and communicating those changes to the organization.

In all, we amplified and accelerated the client’s data science efforts by:

• describing ways to improve their training data;

• recommending new approaches, areas of study, and analysis techniques to utilize in order to further refine their approach to the business-level problem;

• designing a new project management structure for the data science team to improve engagement, transparency, and efficacy of their efforts;

• identifying organizational changes necessary to better promote and utilize good data science across all the organization’s initiatives, ensuring that data-driven insights more directly inform business decisions;

• building an actionable roadmap for implementing these improvements, with suggestions for how to roll out any changes with a view towards promoting improved transparency and better communication of scientific results throughout the organization.

“Organizations often assume hiring more data scientists will accelerate progress.”

Organizations often assume hiring more data scientists will accelerate progress. However, the real solution typically lies in reassessing the overall strategy and questioning whether data science is the appropriate approach for their goals.

Komodo Technologies specializes in practical data science solutions. We think about data science in situ, rather than as a mathematical exercise. If you’re struggling with your data science projects or unsure if you’re addressing the right issues, contact us for a free consultation to discuss how we can help you achieve your goals.