How Dialpad uses Secoda as a core layer in its AI infrastructure

Introduction

Dialpad is a leading business communications platform that powers modern collaboration with voice, messaging, meetings, and AI tools. With over 15 years in the industry, Dialpad serves thousands of organizations and is a direct competitor to platforms like Slack and Zoom.

Dialpad has long been recognized as a leader in AI-powered communications. As early as 2017, the organization was among the first to launch auto-generated conversation summaries using in-house models, well before generative AI entered the mainstream. With a team of 25+ PhD-level data scientists, AI is foundational to Dialpad’s product strategy, powering everything from real-time voice intelligence to smart meeting recaps.

As Dialpad’s external AI products advanced, so too did the vision for its internal data organization. Jesu Joseph, Head of Data, chose Secoda to help support smarter, AI driven data workflows across 12 departments. He saw the opportunity to apply the same level of rigor to Dialpad’s internal data stack, treating metadata and governance as critical components of the company’s AI infrastructure. 

"AI is foundational to Dialpad’s product, and I wanted the same AI-first approach internally. But to do that, you need a lineage layer, a documentation layer, and strong governance. Secoda gives us all the building blocks to do that in one tool.”

The results

Within an implementation and adoption period of less than two months, Dialpad has already built a foundation for AI-first governance and self-serve analytics, bringing trusted insights to every team:

• Surfaced predictive sales insights through Secoda AI, allowing managers to ask questions like “Where should I focus?” and get instant, actionable recommendations
• Enabled opportunity loss analysis in seconds using natural language prompts to analyze pipeline breakdowns by competitor and revenue, without writing SQL
• Accelerated self-serve analytics across the business; more than 200 employees engaged with Secoda AI after an all-hands mention from the CEO
• Cut down analyst time by replacing repetitive manual pulls with AI-driven exploration, allowing questions that previously took hours to be answered in seconds
• Introduced field-level, end-to-end lineage across Fivetran, dbt, BigQuery, and Tableau, enabling teams to understand and trust the full flow of their data
• Centralized documentation and definitions into one trusted platform, replacing fragmented Confluence pages and tribal knowledge with a consistent source of truth

The goal

Jesu’s team set out to do more than just centralize Dialpad’s data stack. Their aim was to build an AI-first data environment that empowers business users to explore, trust, and act on data. To do that, they focused on four key outcomes:

1. AI-powered self-service: Enable every team to explore predictive insights and complex analyses through natural language AI, without writing SQL or relying on the data team
2. End-to-end data trust: Establish full lineage and transparent documentation across the stack, giving users confidence in AI outputs and key business metrics
3. Scalable knowledge management: Consolidate tribal knowledge, model outputs, and definitions into one unified platform, making it easy to govern and evolve over time
4. Company-wide AI adoption: Drive a culture shift toward AI-first analytics by training business users to ask more sophisticated questions through Secoda AI

The solution

To bring this AI-first vision to life, Jesu’s team followed a phased approach: first building a trusted foundation of data and governance, then layering AI-driven workflows on top to empower business users across the organization.

Step 1: Building an AI-ready foundation

Before the team could scale AI-powered use cases, they first needed to rebuild trust in the underlying data. When Jesu joined Dialpad, the environment was fragmented: over 9,500 dashboards lived in Domo, definitions were scattered across Confluence, and even basic questions like “What was our revenue last month?” required an analyst to spend an hour stitching together results from 16 different tables.

To change this, Jesu’s team focused first on simplifying and centralizing the data experience:

1. They migrated key documentation out of Confluence and into Secoda, eliminating the need to send users across multiple tools to understand how data was defined.
2. Using Secoda, they organized Dialpad’s newly rebuilt 190+ Tableau dashboards with clear context and certified definitions, giving business users confidence in what they were seeing.
3. They built a living Glossary of metrics and KPIs, tying definitions directly to 10 to 15 certified datasets the team had rebuilt from the ground up.
4. They connected end-to-end lineage across Fivetran, dbt, BigQuery, and Tableau, providing field-level visibility for the first time. In the past, Dialpad’s lineage was limited to dbt models alone. Now, users could trace a Tableau dashboard all the way back to the raw source, with full transparency.

With this foundation in place, Jesu’s team was ready to start layering AI on top of a trusted, governed data environment.

Step 2: Delivering predictive sales insights through AI-powered exploration

One of the team’s most advanced AI use cases involves surfacing predictive sales insights through Secoda AI to help sales managers coach their teams more effectively.

Dialpad’s data science team built an internal model to predict whether each sales rep was likely to meet or exceed quota in a given month. The model used features such as pipeline value, MQL volume, and company size targeting, and output a predicted attainment score for each rep. The challenge was making this data usable and actionable for frontline managers.

To solve this, Jesu’s team:

• Published the model outputs into a clean table, with each feature clearly documented in Secoda so business users could understand the drivers of the predictions.
• Enabled sales managers to ask questions like: “Summarize what *manager name* needs to focus on in order for his team to reach their goals.” and “Which reps are at risk of missing quota?”
• Secoda AI then generated actionable summaries highlighting top performers, underperformers, and areas for coaching, directly within the Secoda interface.

For example, a manager could instantly see that 3 out of 5 reps were exceeding quota, that pipeline and MQL engagement patterns varied across the team, and that specific reps required targeted interventions. 

Beyond surfacing team-level coaching insights, Jesu’s team has also begun experimenting with individual opportunity analysis through Secoda AI.

In one proof of concept, Jesu demonstrated how Secoda AI could help answer highly targeted questions about specific deals:

• He asked Secoda AI: “What is the latest opportunity created?” Secoda returned the latest opportunity ID.
• He then followed up: “What is the probability of this particular opportunity closing?”

Although the team had intentionally provided only a small subset of fields (10 out of 900+ available), Secoda AI was able to analyze the data and generate an insightful prediction. The output included factors such as rep performance history, opportunity attributes, and relevant trends, allowing Jesu to quickly assess deal risk.

“It did a mind-blowing thing. Even with a limited set of fields, it figured out who the rep is, what their historical trend is, what other factors matter, and came back with a very insightful answer,” said Jesu.

This example demonstrated the potential for using Secoda AI as an exploratory Q&A layer on top of predictive models, allowing business users to go beyond static reports and interact directly with model-driven insights in real time.

Step 3: Analyzing pipeline loss drivers

Another high-impact use case came from enabling competitive pipeline analysis through Secoda AI.

Previously, analyzing why deals were lost and which competitors were involved required time-consuming manual work. Opportunity cancel reasons were logged in various formats and scattered across systems. Compiling this into a usable view required SQL expertise and hours of analyst time.

Jesu’s team streamlined the process:

1. They fed all opportunity cancel reasons into a well-documented opportunity table in Secoda, providing rich context on reasons, competitor names, and associated pipeline and revenue impact.
2. Using this structured and documented data, business users could ask natural language questions such as: “Who are the primary competitors we lost opportunities to last month?” and “What is the revenue impact by X competitor?”
3. Secoda AI returned instant, clear summaries breaking down opportunity loss patterns. As one example, the team was able to quickly generate a detailed Competitor Loss Analysis, identifying their top 5 competitors, along with common loss reasons and strategic recommendations.
4. Users could also follow up by pivoting the view from number of lost opportunities to revenue impact, or ask additional clarifying questions - all within Secoda AI.

This workflow helped Dialpad’s sales, product, and strategy teams gain faster insight into why deals were lost and where to focus competitive efforts, without waiting days for an analyst-generated report.

Step 4: Driving broad AI analytics across the business

As the team saw early success with AI-driven sales insights and opportunity analysis, they shifted their focus toward scaling AI adoption across the organization.

To spark interest, Jesu partnered with Dialpad’s executive team. During an all-hands meeting attended by 800 employees, the CEO directly promoted Secoda AI as a tool for self-serve analytics. The response was immediate.

Within days, 200 employees engaged with Secoda AI to start asking their own questions. Jesu actively encouraged a culture shift: basic questions still belonged in dashboards, but more sophisticated questions, such as model-based predictions or deep analyses, should flow through Secoda AI.

“We’re training the company to rely on dashboards for basic things. But for anything more sophisticated, use Secoda AI,” said Jesu.

This clear guidance helped accelerate adoption, particularly across sales, customer success, and executive teams, who quickly saw the value of accessing insights without relying on analysts or learning SQL.

Step 5: Embedding AI governance and automation

To ensure that AI-driven insights remained trustworthy and explainable, Jesu’s team invested in embedding governance best practices and automation directly within Secoda.

They began by integrating Jira with Secoda. This allowed users to trace discussions, decisions, and Q&A history tied to specific resources, eliminating the need to search across multiple tools and making it easier to reference prior conversations.

Next, the team explored Secoda’s data quality monitors. They started with high-level alerts to minimize noise and focus on critical issues, with plans to expand usage across additional teams. Monitors now alert them to anomalies in key datasets, particularly those with numerous downstream dependencies.

Finally, the team leveraged custom Automations to accelerate documentation and streamline organization. They automated the tagging of assets and the curation of Tableau dashboards into Teams and Collections, reducing manual effort and improving consistency across the catalog.

Looking ahead

As Dialpad evolves their AI stack, Secoda is emerging as a critical layer of their AI infrastructure, helping embed governance, automate documentation, and manage metadata at scale.

Their next steps include:

• Expanding the use of Secoda Monitors and Automations to scale governance and data quality
• Exploring additional use cases for Secoda AI and building more predictive models to support them
• Training more business users to integrate Secoda AI into their daily workflows
• Further consolidating tribal knowledge and AI model outputs into a single unified platform

By treating metadata and governance as foundational elements of their AI infrastructure, Dialpad is leading the way in building internal data systems as advanced and intentional as their external AI products.

If you're interested in exploring how Secoda can support your data and AI initiatives, book a demo today to learn more.