Five teams. One quarter. That's what Backstory's account tiering exercise used to cost: a spreadsheet ranking 141 accounts by growth potential, engagement depth, and retention risk. The kind of analysis every B2B SaaS company with an ABM motion needs, and the kind most avoid redoing because the coordination tax is brutal.

Then Haya Kamal, working with four data connectors and a structured Claude workflow, compressed the whole thing into three days. Same 141 accounts. Same tier outputs. A fraction of the people and none of the cross-functional scheduling gymnastics.

Why the Old Process Broke Down

Account tiering in most B2B orgs follows a familiar pattern: marketing pulls firmographic data, CS pulls usage data, sales adds gut-feel notes, product flags feature requests, and RevOps tries to reconcile it all in a spreadsheet nobody fully trusts. That takes 10–12 weeks because the bottleneck isn't analysis. It's coordination.

Each team guards its own data source. Definitions drift. "Engagement" means something different to CS than to sales. By the time the final tier list ships, the intent signals are already stale.

What Kamal Actually Built

Before touching any data, Kamal's team defined the "golden customer" with account teams and senior leadership. The traits: accounts that had integrated Backstory into their tech stack, engaged in product roadmap discussions, explored new use cases, and planned long-term with Backstory at the center. This step matters more than the tooling because tiering without a clear business question produces segments nobody acts on.

With the definition locked, Kamal identified signals that hadn't been consistently measured: GTM process maturity, AI maturity and progression, deployment velocity, executive visibility, and TAM white space. She developed a five-level AI maturity score per account. The scoring model used three dimensions: fit, intent/timing, and value/revenue potential.

Four Connectors, 20 Minutes

The data pull that used to require five teams became four automated connectors:

The only manual step: exporting the customer list from Salesforce. The Claude workflow ingested and normalized CSVs, then analyzed conversation history, utilization data, growth potential, and feature gaps. Full run time: about 20 minutes.

One counterintuitive finding during four iterations of weight validation: high feature request volume correlated with customer engagement, not dissatisfaction. The initial assumption (more requests = more friction) was wrong. That's exactly the kind of scoring bias that goes unchallenged when tiering happens once a quarter in a spreadsheet nobody revisits.

The Tier Outputs

Final tiers across 141 accounts:

Tier sizing is capacity-driven, which is the right call. Tier 1 should stay small enough that teams can deliver high-touch plays at the required depth. Reported win-rate ranges in tiered ABM programs back this up: Tier 1 accounts see 25–40% win rates versus 12–22% for Tier 2. Concentration pays.

Post-tiering, Backstory made real operational changes: enhanced QBR cadence for Tier A, a new executive sponsor program, growth targets linked to tier, and prescriptive playbooks per tier. The plan is to rerun quarterly to track account movement.

The Trade-Off You're Accepting

Speed solves the coordination bottleneck, but introduces a different risk. When tiering was a 12-week cross-functional exercise, it forced conversations that surfaced context no data connector captures: the rep who knows a champion just left, the CSM who heard about a budget freeze last Tuesday.

Kamal partially addresses this by pulling Slack discussions into the analysis. But automated tiering still requires a validation step where humans with account context can flag exceptions. Without that, you'll occasionally promote a Tier D account to Tier B because usage data looks healthy while the relationship is dying. Store tiers in CRM with explicit reason codes so reps can understand in under a minute why an account landed where it did. That transparency drives adoption.

The other risk: static tiers go stale. Quarterly re-scoring is the minimum. Weekly signal refreshes (intent data, usage spikes, champion job changes) feeding into the next full re-tier keep the system honest.

Backstory's exercise didn't require new tools. It required a clear definition of what a good account looks like, four existing data sources wired together, and a willingness to let the scoring model challenge assumptions. The 20-minute run time gets the attention. The five-level maturity score and capacity-based tier sizing are what made the output actionable instead of decorative. Most tiering projects fail not because the analysis is wrong, but because nobody changes what they do on Monday morning. Backstory changed Monday morning.