Northwind Software: A Blended Score That Hides Two Different Businesses

The blended score is a mixture artifact; the real question is what changed inside each segment.

Decide by disaggregation and by driver analysis, not by narrative:

1. Check the mixture first. Did the score fall because segment experience collapsed, or because the response mix shifted toward small accounts after the support consolidation prompted more complaints? Compare score by segment, response rate by segment, and account-level movement for the same accounts across quarters.
2. Correlate at account level. Test small-business score against effective price change from the usage model, first-response time, and usage intensity. If the accounts with the largest effective price increase and the longest waits are the ones churning, both mechanisms are operating and they interact.
3. Check cost-to-serve. Margin per account by segment and support contacts per account tells you whether small business was profitable before the change or merely quiet.

On the evidence given, enterprise is demonstrably healthy (+55, renewals holding), so this is not a company-wide product failure. The plausible reading is that usage pricing raised effective price for low-usage small accounts while tiering removed the human help they depended on - a segment-specific value proposition break, not an NPS drift.

'One strategy' should mean one logic for allocating service against value - not one treatment for all segments.

For small business, the options are fix or exit, and the data should choose:

• Fix as low-touch. Remove bespoke support from this tier, invest in onboarding, self-service, and community support, and price to reflect cost-to-serve. Test in one cohort: change price and service together, then measure gross margin per account, retention, and contact volume over two quarters.
• Managed exit. If cost-to-serve exceeds gross margin even at corrected pricing, migrate accounts to a partner or a stripped product, with notice and a retention offer for the top decile of small accounts.

Evidence needed: fully loaded cost-to-serve per account, gross margin by decile, churn reasons coded from interviews, and the share of support volume generated by accounts under 50 employees. Do not let enterprise margin silently subsidize a segment the board believes is strategic without saying so explicitly.

Also challenge the 80% target: state which mix shift produces it and what it costs in revenue and retention, so the board is choosing a trade-off rather than approving arithmetic.

Bespoke commitments transfer unpriced obligations from sales to operations, so govern the commitment, not the salesperson.

Mechanisms:

• A priced menu. Publish standard service tiers with defined response and resolution targets; anything outside the menu is a non-standard deal requiring operations sign-off.
• A deal desk with authority limits. Thresholds by contract value and by deviation from standard terms; above the threshold, operations and finance must approve before signature.
• Cost of non-standard terms. Attach an internal cost estimate to each deviation (extra coverage, penalty exposure, engineering work) so margin erosion is visible at the point of decision, not discovered post-signature.
• Post-signature tracking. Report margin and service performance for non-standard accounts quarterly; if the exception book grows, the menu is mispriced.

Rationale: in a quarter where the blended score is falling and churn is concentrated in small accounts, ad hoc commitments buy short-term bookings and create exactly the support load that caused the decline. The governance answer is to make exceptions deliberate and priced, not to ban them - sales needs flexibility for genuinely strategic accounts, and a blanket ban will be routed around.