"Net new customers: plus twelve." It's a good headline until you ask how it happened. Forty acquired, thirty churned, two came back. Now it's a churn problem with a good quarter of sales on top of it. Revenue reporting is full of headlines like that: pipeline totals that are mostly dead quotes, promotions that discounted revenue that was coming anyway, recurring revenue that nobody tagged as recurring.

Part one covered thirteen customer analytics prompts new in the Sonar edition of my NetSuite AI Prompt Library. This is the second thirteen, and they share a theme: the mechanism under the headline. They're grouped by the question each one answers.

If you're new to this, Sonar AI is an AI agent that runs inside NetSuite. Every prompt in the library is a playbook that I engineered and tested against live NetSuite data, and you run it inside your own account, against your own records.

How Customers Move

Customer Status Flow renders the quarter-over-quarter customer flow as a Sankey: retained, churned, new, and reactivated, all at once. It's the picture that turns "net new" back into its parts, and the prompt flags any quarter where churn passes 15% or acquisition drops more than a fifth. Customer Segment Migration does the same for revenue and health segments. A snapshot tells you how many Whales you have today. The migration table tells you which accounts became Whales, which slid to Mid, and which went quiet, and it states the revenue thresholds that define each segment so nobody has to guess. Whale-to-inactive moves go to a person.

Customer Segmentation Blueprint is the bigger tool. It produces three to seven operational segments from RFM plus a behavioral overlay, with a hard ceiling on the count because twelve segments create paralysis where four drive action. Each segment has to hold at least 5% of the base, each gets an explicit lifetime value formula rather than an assertion that LTV is high, and each comes with a prioritized strategy: channels, costs, expected lift, and confidence. It also defines the migration logic, so the segments can be measured again next quarter instead of rebuilt.

What Is Really in the Pipeline

Four prompts work the front of the funnel, and they're honest about what NetSuite exposes there. The sales rep on an estimate header, for instance, isn't always reachable through a query, and the prompts say how they got it.

Pipeline Velocity reduces the funnel to one number per period from deal count, win rate, average deal size, and cycle length, and then watches the trend, because the direction of the number is the signal. It walks the stages from opportunity to estimate to sales order to invoice through the created-from links and flags a velocity drop of more than 20% month over month. Quote-to-Cash Funnel measures conversion and drop-off at every stage from estimate to cash receipt, in count and in dollars, with the median elapsed time per stage, so you can see where value is preserved and where it leaks.

The Forgotten Estimate Pipeline is the cleanup prompt. It surfaces open estimates and opportunities by age, owner, and customer, the ones that inflate the pipeline headline with no real chance of closing. It never invents an amount, and any recommendation to close a deal over $10,000 as lost, or to remove more than a fifth of the pipeline's face value, is a human decision. Quote Aging and Follow-up Prioritizer is its active counterpart: probability-weighted follow-up, because a 90-day quote isn't worth the same as a 7-day quote at the same dollar value. It flags overloaded reps and large quotes headed for lost status for review.

Where the Margin Is

Per-Customer Margin Profile is a bubble chart of revenue, gross margin percentage, and order count for every customer. Its purpose is to expose the unprofitable share of the book that a revenue report can't see, and it cross-checks against the native income statement in base currency. Recommendations to reprice, deprioritize, or exit a customer are flagged for a person.

Promotion and Discount Effectiveness compares the margin and volume of orders carrying a promotion code against the baseline, to separate incremental revenue from revenue that would have occurred anyway at full price. Discontinuing a promotion or changing discount policy is a human call. Hidden Recurring Revenue finds the subscriptions nobody set up as subscriptions: customer and item pairs that transact on a predictable monthly or quarterly cadence. It's a behavioral proxy, and the prompt says so; if the account uses NetSuite's subscription records, it prefers those.

What Sells Together, and When

Market Basket Co-Occurrence applies association-rule mining to sales order lines to find the items bought together, for bundling, cross-sell, and merchandising. Every pairing and lift value traces to a query, and any change to list prices or product strategy goes to a person. It pairs naturally with the cross-sell map from part one: one finds what's bought together, the other finds what should be.

Top Revenue Spike Days ranks the highest-revenue calendar days and takes each one apart by customer, item, and channel, then asks whether the spike was a campaign or an anomaly. Attributing a spike to a campaign needs corroborating evidence, such as a promo code, and without it the cause is labeled possible rather than asserted. Lead Source and Marketing Campaign ROI attributes closed revenue to lead source and campaign and computes realized return per channel. It never invents campaign costs or conversion rates, and any conclusion that would move budget is flagged for review.

Using Them Together

These prompts were written to be run in sequence, not in isolation. A quarterly review might start with Customer Status Flow to see the mechanism behind the customer count, then Customer Segment Migration to see which named accounts moved, then Per-Customer Margin Profile to check whether the accounts that grew are the ones worth growing. On the sales side, Pipeline Velocity gives the trend, The Forgotten Estimate Pipeline cleans the denominator so the trend is honest, and Quote Aging and Follow-up Prioritizer turns what's left into this week's call list.

The reason that sequence works is that every prompt states its scope and its thresholds in the output. When the segment migration report says a Whale is a customer above a stated revenue line, and the margin profile uses the same customer set, the two reports can be read together without reconciling them first. That's a small discipline, and it's most of what separates a set of analyses from a set of numbers.

Wrapping Up

The thread through these thirteen is that a total is a claim, and a claim needs a mechanism. Net new customers has a flow behind it. Pipeline value has an age distribution behind it. Promotion revenue has a baseline behind it. When the mechanism is one prompt away, the headline stops being the end of the conversation and becomes the start of it.

All thirteen are in the paid tier of the Sonar AI Prompt Library, under Revenue and Customer Analytics.