Pre-call account review
A customer's account status on one page before a sales call: purchases, orders, complaints and overdue balance
For companies that sell to other businesses and only notice that a major customer is leaving when year‑on‑year sales fall, months after orders begin to drop.
- Data sources
- ERP (the company's management system) · CRM (the customer management system)
- ERP (the company's management system)
- CRM (the customer management system)
- Approval
- No approval: the salesperson decides whether to use the review, makes the call and records what was agreed on the customer record
- First step
- The order frequency rule is run on the last two years of orders and the result is shown before any contract is signed
Pre-call account review
In preparationReady to view- Order volume down on the previous twelve months
- One open delivery issue
- No overdue balance
- One complaint pending resolution
- Recommended action: resolve the complaint before making new offers
The pre‑call account review sets out on one page the status of a customer who is ordering less often: the items the customer buys and has stopped buying, orders pending delivery, delivery performance on orders already delivered, open complaints and the overdue balance.
The review is triggered by a drop in a customer's order frequency, recalculated every night from a synchronised copy of the data in the ERP (the company's management system). The salesperson makes the call and records what was agreed in the customer record in the CRM (the customer management system).
On this page
The manual process
In business‑to‑business sales, a major customer is lost gradually: purchases shrink, part of the range moves to another supplier, and the fall only appears when sales are compared year on year. The warning signs appear months earlier: longer gaps between orders, items that stop appearing and complaints resolved late.
Preparing for a call by hand means looking at five screens across two systems: sales by period, orders pending delivery, open complaints, credit and the latest orders. The usual result is a call made without the full picture: the salesperson does not know about the overdue balance, the late delivery of the latest order or the open complaint.
Data sources and the company map
The company map is the written description of where each piece of data is held, what each source contains and how each business situation is analysed. The agent is the software that performs each task; the agent checks and calculates only what the map defines.
Show the figure
Sources, calculations and actions defined in the map
Show details
For this use case, the map defines two sources: the ERP, which holds orders, invoices, payments and credit used; and the CRM, which holds the customer record with the credit limit, contacts, visits and complaints, and where the salesperson records what was agreed on the call. From these two sources, the map defines four calculations and two actions reserved for the salesperson: the call and the record of what was agreed.
The agent reads both sources from a synchronised copy and writes nothing to the ERP.
Detecting the change in order frequency
Each customer's order frequency is the usual gap between their orders, derived from the months in the last year in which the customer placed an order. The agent recalculates it nightly over a fixed period, and the signal is raised when the time since the last order exceeds that usual gap by a defined proportion.
Show the figure
The usual gap and the time since the last order
Show details
The reference is each customer's own pattern, not the portfolio's. A customer who orders every month and has gone a quarter without ordering has changed frequency; a customer who orders every quarter and has gone the same quarter without ordering has not.
The signal appears before the sales comparison shows any fall, because cumulative sales take months to reflect missing orders. The rule requires a minimum number of months with an order before a customer has a reference frequency; occasional customers raise no signal, and any fall in their purchases shows only in the sales comparison.
Checking sales, orders, complaints and collections
Once the signal is raised, the agent checks all four areas of the map together and takes the data needed for the call from each one. Every figure in the review carries its source: the screen it comes from.
Show the figure
| Area | Contribution to the review | Calculation |
|---|---|---|
| Sales | The items the customer buys and the items they have stopped buying, with the change over the period | Sales per item in the period and in the same period of the previous year |
| Orders | Orders pending delivery and the delivery performance of those delivered | Promised date against actual dispatch date, order by order |
| Complaints | Open complaints and those closed recently, with their resolution time | Time between opening and closing, complaint by complaint |
| Collections | Credit granted and credit used, overdue invoices and recent payments | No calculation: the limit is read from the CRM and the rest from the ERP |
Composing the review
The account review has a word limit and three parts: the customer's status, questions for the call and proposals supported by the customer's history.
The review adds no figures of its own: the language model (the artificial‑intelligence program that reads and writes text) writes the three parts using only the figures already shown on the screen, taken from the same calculation.
Show the figure
The review, in its three parts
Pre-call account review
Only the figures shown on screenCustomer status
- Order frequency
- Change in sales
- Orders pending delivery
- Overdue balance
Questions for the call
- Items no longer ordered by the customer
- Complaint resolved late
- Order delivered late
Proposals
- A substitute item, a delivery date or a visit, drawn from the customer's history
Reading the review, calling and recording what was agreed
The review reaches the salesperson as a “Pre‑call account review” alert with four actions: view the review, add a comment, reject it and, after the call, record what was agreed. Recording what was agreed is the only one of the four actions that changes any data; the other three change nothing. The comment adds what the salesperson knows and the system does not. The detailed view links every figure to its source screen.
Show the figure
The alert to the salesperson and the record of the call
Alert to the salesperson
Pre-call account review
- Status, questions and proposals
- Every figure linked to its source screen
- Customer record in the CRM: the note of what was agreed
- Operations log: the review's calculations and the note
Show details
The salesperson makes the call; the agent neither calls nor writes to the customer. The review requires no approval: the salesperson decides whether to use it. Management can schedule the review for the whole sales team and see each day which reviews are ready, which are partial and which could not be prepared; scheduling writes nothing to the customer record.
After the call, the “Record what was agreed” action saves the note in the customer record in the CRM and in the operations log; the agent writes nothing else. The next review of the same customer includes that note.
Design criteria
The order frequency rule is the same for the whole portfolio and can be checked step by step. If each salesperson used a different definition, reviews could not be compared and discussion would centre on the definition rather than the customer.
The review has a word limit so the salesperson can read the whole review before the call.
Company decisions
Three decisions specific to each company determine which customers are flagged and what the review contains.
Portfolio scope
A list kept by the company excludes customers in special channels, such as direct sales or export, from the salespeople's portfolios. Without that list, a salesperson receives alerts about customers who are not theirs.
Frequency signal thresholds
The rule has two settings, the same for the whole portfolio: how far the time since the last order must exceed the usual gap to raise the signal, and the minimum number of months with an order. Before changing them, the company can see which customers the new values would add and which they would drop; the company decides whether to make the change, and the next nightly recalculation applies the new values to every customer.
Language model and hosting region
During rollout, the company chooses the language model that writes the review, its provider and the region where it runs. The model receives only the data the screen already shows, and the company signs the contract with the model provider.
Rollout
The rollout begins by running the order frequency rule on the company's orders from the last two years, and the result is shown before any contract is signed: the customers the rule would flag today and the customers already lost that the rule would have flagged in advance. Hellomatik builds the map from the data the company supplies; the company neither accesses the technical console nor writes the map. The use case has no fixed timetable.
- 01
Company map
Hellomatik defines the use case's sources: orders, invoices, payments and credit in the ERP, and customers, contacts, visits and complaints in the CRM. The connection to the ERP is read‑only.
- 02
Testing the rule on past orders
Hellomatik calculates order frequency on the last two years of orders and compares the result with customers already lost: how many the rule would have flagged, how far in advance and how many customers the rule cannot assess.
- 03
Test reviews for known customers
Hellomatik prepares reviews for customers the salespeople already know, and the salespeople check each review for what is missing, unnecessary or inaccurate. Corrections are made in the map, not in the review text.
- 04
Portfolios and alerts
The company assigns each salesperson's portfolio, chooses who receives each alert and defines what is recorded in the CRM after the call. Once portfolios and alerts are assigned, the agent monitors order frequency and the salesperson makes the call.
Limits of the use case
The use case does not calculate the probability of losing a customer or how many customers the company retains. The signal comes from a written, verifiable rule, not from a program that learns from the data, and no company is running this use case yet.
The use case raises no alert for occasional customers. A customer who orders in only a few months of the year has no reference frequency, so no signal is raised; any fall in that customer's purchases shows only in the sales comparison.
The use case does not script the conversation with the customer. The review suggests questions and proposals drawn from the customer's history; the conversation and any commitments made during it are the salesperson's responsibility.
Book a meeting with a Hellomatik engineer to review what each salesperson knows today before calling a customer.
Or request a meeting here