Backstory's AI-Driven Customer Tiering: Quarter to 3 Days
Backstory reduced its customer base re-tiering process from a quarter to three days by leveraging AI to define 'golden customers,' build custom signals like AI maturity, and automate data collection via connectors. This allowed for rapid, iterative analysis and strategic decision-making across customer segments.
Key Takeaways
- Qualitative definition of ideal customers should precede data analysis for effective tiering.
- AI can create sophisticated custom signals (e.g., AI maturity score) by analyzing diverse data sources, including unstructured text.
- Automating data collection with connectors significantly reduces manual effort and improves data freshness.
- Iterative refinement of AI models is crucial to validate signal weights and uncover non-obvious insights.
- Understanding all customer tiers, including 'Tier D,' is vital for comprehensive go-to-market strategy.
Backstory, a company specializing in customer success, dramatically streamlined its customer tiering process, reducing a quarter-long, five-team effort to just three days. This transformation, spearheaded by Customer Success lead Haya Kamola and presented at SaaStr AI Day, involved a strategic blend of qualitative insights, custom AI-driven signals, and automated data integration.
Redefining Customer Value with AI
The mandate was clear: categorize 141 accounts to identify the most valuable customers and redistribute team focus accordingly. Previously, this required extensive cross-functional collaboration and manual data aggregation. Kamola's approach began not with data, but with a qualitative definition of a "golden customer." Account teams and senior leadership collaboratively identified key traits: customers who embedded Backstory into their core tech stack, built systems around it, planned long-term with the platform, actively contributed to the roadmap, and consistently discovered new use cases. This foundational definition ensured that subsequent data analysis aligned with strategic business value, rather than merely relying on available CRM fields.
Crafting Custom AI-Powered Signals
A critical step involved developing signals that didn't exist in standard CRM data. Backstory identified characteristics such as go-to-market process maturity, tech stack mix, AI maturity (and its velocity), partner motions, deployment speed, executive visibility, and total addressable market within an account. Many of these were siloed or unquantified. A standout example was the five-level AI maturity score. Instead of manual categorization, Backstory built an automated signal: a prompt executed systematically against each account. This prompt analyzed CRM fields, public company information (like AI product launches), and the account's full chronological conversation history (emails, meetings, Slack). The output provided both a maturity level and its rationale, solving a previously "grueling" manual task.
Automating Data Integration
The traditional, quarter-long data pull involving product, BI, finance, and other teams was replaced by just four connectors and a Salesforce export. These connectors automated the collection of vital information:
- Amplitude: For utilization and usage data.
- Atlassian and Jira: For years of logged customer feature requests and identified gaps.
- Backstory's own MCP: For the complete conversation and engagement history.
- Slack: For internal account team discussions, which often provide the earliest and most candid insights into customer sentiment, strategy, and risks.
According to SaaStr, the Slack integration is particularly noteworthy, as internal dialogue rarely makes it into structured systems but offers invaluable, real-time customer insights.
Iteration and Strategic Outcomes
The analysis workflow was meticulously defined as a sequence in Claude, utilizing Cowork to ensure steps executed in order. This included data ingestion and normalization, processing conversation history, utilization data, growth potential, and Jira insights, followed by reconciliation and scoring. The process took about 20 minutes per run.
Kamola emphasized the importance of iteration, undergoing three to four rounds of revalidating signal weights. Initially, eight signals were considered, eventually narrowing down to four core scoring buckets: growth potential, AI maturity/velocity, engagement level, and current account health. A key discovery during this phase was that a high volume of feature requests, initially perceived negatively, actually correlated strongly with high adoption and AI-forward engagement, proving to be a positive signal. This highlights how AI-driven analysis can uncover counter-intuitive insights that manual processes might miss.
The final output consisted of four tiers:
- Tier A: High-growth potential (10x in 2-3 years).
- Tier B: Long-term 10x potential with a different playbook.
- Tier C: Accounts at a critical juncture.
- Tier D: Customers potentially not aligning with Backstory's future direction. Identifying Tier D was crucial for strategic decisions, prompting companies to either cut losses or adapt service models.
This case study from Backstory demonstrates how AI, when thoughtfully applied to data strategy and workflow automation, can revolutionize core go-to-market operations, turning a lengthy, resource-intensive task into a rapid, insight-driven process.
Why This Matters for AI Product Managers
For AI Product Managers, Backstory's success story underscores the power of integrating AI into core business processes. This case highlights several critical areas:
Data Strategy & Agent Design: The creation of the AI maturity score demonstrates how AI agents can synthesize complex, multi-source data (CRM, public info, conversation history) into actionable, structured insights. AI PMs should consider how to design similar internal agents to automate data analysis for product usage, customer sentiment, or market fit, reducing reliance on manual data pulls.
Roadmap & GTM Alignment: The 'golden customer' definition directly informs product strategy by identifying users who value deep integration and roadmap input. For AI PMs, understanding which customer segments truly drive product evolution is key for feature prioritization and ensuring new AI capabilities resonate with high-value users. The tiered output directly supports GTM teams in tailoring sales and success playbooks.
UX of Internal Tools & Analytics: The use of Cowork for sequential AI execution, and the integration of Slack data, points to the need for intuitive internal AI tools. AI PMs should focus on building user-friendly interfaces for data ingestion, model iteration, and outcome visualization, ensuring that AI-driven insights are accessible and actionable for GTM teams, not just data scientists.
Related Articles
Clockwork: The Calendar for Scheduling & Orchestrating AI Agents
Clockwork, a new tool highlighted on Product Hunt, introduces a novel calendar interface designed for scheduling and orchestrating AI agents, providing visibility into their tasks and availability. This innovation addresses the growing complexity of managing autonomous AI workflows in business operations.
Grove: A Unified Terminal for Human-AI Agent Collaboration
Product Hunt featured 'Grove,' a new concept proposing a unified terminal for human-AI agent interaction, aiming to streamline collaboration and workflow management.
AI Renewal Agents: Personalized Decks for Every Customer, Not Just Big Ones
SaaStr reports on a new AI renewal agent capable of creating personalized, high-quality renewal decks for every customer account, overcoming the traditional challenge of prioritizing only large clients due to time constraints. This innovation promises to boost efficiency and improve customer retention by ensuring all customers receive tailored engagement.