B2B SaaS deep dive

Net new ARR - the single number on the board slide - is New + Expansion - Contraction - Churn. It can look healthy while hiding a fire. A company can post +$110K net while gross churn quietly doubles, as long as new sales sprint to cover it. The bridge forces the gross flows into the open so you cannot hide a retention leak behind a strong sales quarter.

• New - ARR from brand-new logos. Sales and marketing own it.
• Expansion - existing customers buying more (seats, upsell, usage). Customer success and product own it.
• Contraction - existing customers buying less. A downgrade, not a departure. A loss flow, sign negative.
• Churn - customers who leave entirely. The other loss flow, sign negative, and usually the most expensive.

Here is the same bridge as a live tree. The two loss flows carry "sign": -1 so the simulator subtracts them. Edit any flow and watch ending ARR recompute. This is the flagship example of the bridge operator you met in b2.

{
  "unit": "$",
  "root": {
    "label": "Ending ARR", "op": "bridge",
    "children": [
      { "label": "Start ARR",   "value": 1000000, "unit": "$" },
      { "label": "New",         "value": 180000,  "unit": "$" },
      { "label": "Expansion",   "value": 90000,   "unit": "$" },
      { "label": "Contraction", "value": 40000,   "unit": "$", "sign": -1 },
      { "label": "Churn",       "value": 120000,  "unit": "$", "sign": -1 }
    ]
  }
}

Split the New flow into its drivers. Marketing owns leads, sales owns win-rate, packaging and pricing own ACV. Now a "new-business shortfall" stops being vague and becomes one of three named levers.

{
  "unit": "$",
  "root": {
    "label": "New ARR", "op": "x",
    "children": [
      { "label": "Leads",    "value": 600, "unit": "leads" },
      { "label": "Win-rate", "value": 0.10, "pct": true },
      { "label": "ACV",      "value": 3000, "unit": "$" }
    ]
  }
}

Ending ARR and NRR are lagging - they close the book on a period that already happened. To steer, you watch the flows and the inputs to the flows, which lead by weeks or quarters.

• Leading: pipeline coverage (pipeline / target), trial-to-paid rate, product-qualified leads, usage trend inside accounts. These warn early.
• Lagging: ending ARR, net new ARR, NRR, gross churn dollars. These report late but are what the board sees.
• The rule: if pipeline coverage falls below about 3x this quarter, next quarter's New flow is already in trouble - long before ARR shows it.

• Churn spike (a retention problem) -> the Churn flow grows. Owner: customer success. Deeper cause is usually onboarding, value delivery, or a competitor.
• Expansion stalls (a product-value problem) -> the Expansion flow shrinks, and often Contraction grows with it. Owner: product and CS. Customers are not finding more to buy.
• New falls (a pipeline or win-rate problem) -> the New flow shrinks. Owner: marketing (leads) or sales (win-rate). Check pipeline coverage first - it warned you weeks ago.

The seed has a small subscription base. A cancelled subscription (a non-null cancel_date) is a churned account - the raw material of the Churn flow. Run this to see how many are still active versus gone.

SELECT
  COUNT(*) FILTER (WHERE cancel_date IS NULL)     AS active_subs,
  COUNT(*) FILTER (WHERE cancel_date IS NOT NULL) AS churned_subs,
  ROUND(100.0 * COUNT(*) FILTER (WHERE cancel_date IS NOT NULL)
        / COUNT(*), 1)                            AS churn_rate_pct
FROM subscriptions;

On the ARR bridge in Demo 1, edit Expansion from $90K down to $45K. Ending ARR falls by $45K to about $1.065M. More importantly, feel NRR drop: (1.00M + 45K - 40K - 120K) / 1.00M = 88.5%. Expansion is the flow that pushes NRR above 100% - halve it and the installed base leaks faster. Say why a product team, not sales, owns this miss.

Split the churn proxy by plan. Join subscriptions to customers for the plan, then count active versus cancelled in each. Watch where every cancellation lands.

SELECT c.plan,
       COUNT(*) FILTER (WHERE s.cancel_date IS NULL)     AS active_subs,
       COUNT(*) FILTER (WHERE s.cancel_date IS NOT NULL) AS churned_subs
FROM subscriptions s
JOIN customers c ON s.customer_id = c.customer_id
GROUP BY c.plan
ORDER BY churned_subs DESC;

Two SaaS companies both grow net ARR by the same $110K this year. Company A does it with low churn and modest new sales. Company B does it with high churn masked by aggressive new sales. Which is healthier, and why?

• Company A is healthier. Low churn means its NRR is near or above 100% - the existing base compounds, so next year starts from a higher, sturdier floor and every new logo adds on top rather than backfilling a leak.
• Company B is running on a treadmill. High churn means a low NRR; new sales are spent replacing lost revenue, not growing it. The moment new-sales momentum stalls, the leak shows and net growth collapses. It also costs far more - acquiring a new logo is dearer than keeping one.
• The lesson: net ARR growth is a headline; the gross flows are the diagnosis. Low churn compounds, high new sales rents.