Backstory Retiers 141 Accounts in 3-4 Days with Four Connectors
Haya Kamola detailed how Backstory completed account tiering on 141 customers in three to four days using connectors, custom signals, and four iterations.
Backstory completed an account tiering analysis covering 141 accounts in three to four days. Haya Kamola, who leads customer success at Backstory, presented the project at SaaStr AI Day. The same exercise previously required five teams working a full quarter. The board requested the tiering framework with a short turnaround.
Definition Preceded Data Collection
Kamola first gathered input from account teams and senior leadership to describe specific customers that stood out. The resulting definition focused on accounts that treated Backstory as a core part of their tech stack, built systems around the platform, planned five years ahead with it at the center, engaged on the roadmap, and expanded into new use cases. All subsequent scoring measured accounts against this description.
Custom Signals Filled Measurement Gaps
The team identified characteristics not previously tracked repeatably across the full base. These included go-to-market process maturity, tech stack mix, AI maturity on a five-level scale with rationale, partner motions, deployment velocity, executive visibility, total addressable market, and remaining white space. AI maturity was generated by a prompt that pulled CRM fields, public company information on AI-forward launches and investments, and chronological conversation history from emails, meetings, and Slack.
Four Connectors Replaced Cross-Functional Pulls
Amplitude supplied utilization and usage data. Atlassian and Jira provided logged feature requests and gaps. Backstory’s own MCP delivered conversation and engagement history. Slack channels maintained for each customer supplied the earliest account strategy and risk details. The sole manual step was a single CSV export from Salesforce containing account name, executive engagement level, predicted health, AI maturity signal, renewal date, and renewal ACV, according to SaaStr.
Four Iterations Produced Final Scoring Buckets
The workflow executed as a defined sequence in Claude using Cowork. Eight signals were narrowed to four scoring buckets across four rounds of iteration. One signal was corrected after it was found to be scoring backwards. The completed tiering framework was delivered for executive action.
according to SaaStr.
