To reduce customer churn in B2B sales, watch what each customer buys, not whether they have cancelled. Distributor and supplier customers rarely send a notice. They order less often, in smaller amounts, from fewer product lines. Catch that drop in your invoices and visit within two weeks.
This article gives you the churn rate and attrition rate formulas with a worked example, what a lost account costs a distributor, the four numbers to pull from your invoice export, a way to rank which customers to see first, who owns an at-risk account, and a visit plan for the conversation. It also covers the first 90 days of a new account, pricing terms that hold a customer without a discount spiral, and how to win back lapsed customers.
B2B customers leave by buying less
Software companies count churn as cancellations because a subscription ends on a date. Your customers have no subscription. A contractor who buys from you every week can move half the order to the supplier down the road, and nothing in your system changes. No ticket, no email, no goodbye.
That makes churn invisible in the tools most sales teams use. A sales report shows total revenue, and total revenue hides it: three new accounts can cover for two that are fading. You find out when the last order has been missing for four months and the customer says they 'found someone closer.'
So treat a drop in buying as the cancellation notice. It arrives weeks or months earlier than a lost account does, and it is in data you already have: the invoice export from your accounting or order system.
Measure churn and what it costs: rate, attrition and net revenue retention
Pick a period, count the customers you had at the start, and count how many of them you lost. That is the churn rate. Gainsight gives the formula as customers lost divided by customers at the start of the period, times 100, over a month or a year. Customer attrition rate is the same number under another name, so use whichever your finance team already says.
A distributor needs one extra decision, because no customer cancels. Decide when an account counts as lost. A workable rule is no invoice in 90 days for an account that used to buy every month, or 1.5 times a customer's usual gap between orders if it buys less often. Write the rule down and apply it the same way every period.
The rate counts accounts. Dollars tell you more, so measure those too. All figures below are a hypothetical distributor with 200 active accounts at the start of the year and $4,000,000 in sales to them, an average of $20,000 an account.
| Measure | Formula | Example |
|---|---|---|
| Customer churn rate | Customers lost ÷ customers at the start × 100 | 14 ÷ 200 = 7% |
| Revenue churn | Sales last year from accounts now lost ÷ starting sales × 100 | $280,000 ÷ $4,000,000 = 7% |
| Contraction | Sales lost from accounts still buying, as a share of starting sales | $240,000 ÷ $4,000,000 = 6% |
| Expansion | Extra sales from accounts that bought more | $180,000 ÷ $4,000,000 = 4.5% |
| Net revenue retention | (Starting sales − lost − contraction + expansion) ÷ starting sales × 100 | ($4,000,000 − $280,000 − $240,000 + $180,000) ÷ $4,000,000 = 91.5% |
Read the last row. A 7 percent churn rate sounds survivable. But 91.5 percent net revenue retention means this book of business shrinks by 8.5 percent, $340,000, before a single new account is won. Accounts that stayed but bought less took $240,000, nearly as much as the accounts that left. That is why this article watches buying, not only cancellations. Gainsight's page describes net revenue retention as the measure that counts both losses and expansion.
What it costs to replace the lost buying
Continue the example. To stand still, the team must find $340,000 of new sales. At the average of $20,000 an account, that is 17 new accounts, more than the 14 that left. Say each new account costs $2,400 to win (the formula is in sales KPIs for field teams). That is $40,800 before the first order. New accounts also start small. If a new account buys half the average in its first year, the 17 bring in $170,000, and the hole is still $170,000 deep.
Now add margin. At a 25 percent gross margin, the $340,000 of lost sales is $85,000 of gross profit gone. If visits to the accounts that bought less had recovered a quarter of the $240,000 contraction, $60,000 of sales and $15,000 of gross profit would have stayed. That is why reps should visit the fading accounts as often as they chase new ones. Gainsight makes the same point about keeping customers costing less than winning new ones.
Four signals that a customer is drifting
Compare each customer's last three months with the nine months before. Three months is long enough to smooth out one slow week and short enough to act on. If a customer buys seasonally, compare the same three months a year ago instead.
Order frequency
Count orders per month. A customer who ordered three times a month and now orders once is telling you they buy the rest elsewhere. Frequency is usually the first signal to move, because small top-up orders are the easiest to hand to a competitor.
Order size
Divide invoice dollars by the number of orders. A falling average means smaller baskets: fewer units, or the big items going elsewhere. This signal alone is the weakest, since one large job can inflate the old average. Use it with the others.
Product mix
Count the product categories on the customer's invoices. A customer who bought tile, adhesive, grout, tools and trim, and now buys only tile, has given the other four lines to someone. Dropped lines often point to the reason: a better price, a stock-out on your side, or a rep at another supplier who asked. Measuring this across all your customers is share of wallet work.
Days since last visit
This one comes from your own records, not the invoice. A customer on a four-week cycle who has not seen a rep in 90 days has been left alone long enough to be approached by everyone else. A long gap with falling orders is the pattern to treat first. Call cycle planning covers how to set the interval.
| Signal | What to pull | Flag when |
|---|---|---|
| Order frequency | Orders per month, last 3 months against the 9 before | Down by a third or more |
| Order size | Invoice dollars divided by orders | Down by a quarter or more |
| Product mix | Product categories bought | Down by a third or more, or any core line gone |
| Days since last visit | Last check-in date | More than 1.5 times the planned cycle |
Warning signs the invoice does not show
Invoices lag. A rep standing in the customer's yard sees things months earlier. Ask reps to log these in the visit note, and count the notes that carry one.
- A new buyer or owner. The person who liked you has moved on and nobody introduced the replacement.
- A competitor's product or truck on site. Note the brand and the line it replaced.
- Slower payment or more disputed invoices. Your accounts receivable team sees this first, so ask them for a weekly list of customers whose payment slipped.
- More complaints, credits or returns. Two late deliveries in a month are a warning even when orders have not moved yet.
- Quote requests that stop turning into orders. The customer is shopping you.
- A job that ended, or a shop that is quiet. Not every drop is a loss to a competitor. Ask before you assume.
Give each sign a tag in the note, so a manager can count them. A customer with one invoice signal and one visit sign is a stronger flag than one with two weak invoice signals.
A worked example with five customers
Say you run a building materials distributor with a four-week visit cycle. Five contractor customers, last three months against the nine before. All names and figures are made up to show the method.
| Customer | Orders a month | Average order | Categories | Days since visit | Signals | Monthly gap |
|---|---|---|---|---|---|---|
| Ridge Builders | 3.0 down to 1.0 | $1,850 to $1,790 | 5 to 3 | 71 | 3 of 4 | $3,760 |
| Paloma Homes | 1.0 down to 0.3 | $5,200 to $4,900 | 6 to 4 | 94 | 3 of 4 | $3,567 |
| Mesa Kitchens | 2.0 to 2.0 | $2,400 to $1,300 | 4 to 4 | 22 | 1 of 4 | $2,200 |
| Dunn Remodeling | 2.0 down to 1.0 | $1,600 to $1,500 | 4 to 4 | 30 | 1 of 4 | $1,700 |
| Cole Roofing | 4.0 to 4.0 | $900 to $950 | 3 to 3 | 15 | 0 of 4 | $0 |
The monthly gap is orders a month times average order, before minus now. Ridge Builders went from 3.0 orders at $1,850 ($5,550 a month) to 1.0 order at $1,790 ($1,790 a month). That is $3,760 a month of buying that went somewhere else.
Read the table. Ridge and Paloma trip three signals and have not been visited in two to three cycles. They are the two visits to make first. Mesa Kitchens orders as often as before but each order is 46 percent smaller: a different conversation, probably about one big product that moved. Dunn halved its orders and nothing else changed, so ask. Cole Roofing is healthy and needs nothing beyond the normal cycle.
Notice what a revenue report would have shown. Cole's steady $3,800 a month sits beside the losses and softens them. The gap column is the number a manager should read, not total sales.
Rank at risk customers by money, then by signals
You will find more drifting customers than you can visit this week. Rank them in this order and stop when the week is full:
- Monthly gap. Sort by dollars of buying lost, as in the table. A $3,700 gap outranks a customer who dropped 60 percent from a small base.
- Number of signals. Among similar gaps, the customer tripping three signals has a fuller problem than one tripping one.
- Days since last visit. Break ties toward the customer nobody has seen, because you know least about them.
- Account value over a year. If two customers tie, see the one you would hate to lose.
Give the list to the rep who owns the customer, with the numbers beside each name. A rep who hears 'Ridge is down' goes in with a guess. A rep who hears 'Ridge went from three orders a month to one, and stopped buying grout and trim' goes in with a question.
Who owns an at-risk account: a proactive outreach playbook
A flag nobody owns is a report. Name an owner for every flagged account on the day it is flagged, and give each owner a deadline. The rep who visits the account owns the conversation. The manager owns the weekly list and asks about every name on it. Service, credit and the warehouse own the fixes the customer asks for, and each fix has a name and a date.
| Trigger | Owner | Action | Within |
|---|---|---|---|
| Two signals flagged on a regular account | Territory rep | Visit with the numbers in hand, ask what changed | 14 days |
| Any signal on a key account | Rep, manager joins | Call the buyer, then visit; manager asks about it in the weekly review | 3 days to call, 7 to visit |
| No order for 1.5 times the usual gap | Territory rep | Phone the buyer and ask if anything is missing | 2 days |
| Complaint, credit or late delivery logged | Rep, then service or warehouse | Call the customer back with the fix and the date | Same day |
| Fix promised at a visit | Whoever promised it | Do it, then tell the customer it is done | The date promised |
| Account lost (90 days, no invoice) | Manager | Record the reason, move to the win-back list | 1 week |
Track three things about the playbook itself: the share of flagged accounts visited inside the deadline, the share of promised fixes done by their date, and the share of flagged accounts whose buying recovered within two months. The weekly sales report template has a place for the flagged list.
A win-back visit plan for a drifting customer
Most reps handle a fading account the same way: stop by, ask how things are going, and leave with 'all good.' That fails because the customer has no reason to tell you the truth. Use a plan instead.
- Before the visit, write down the three facts. Which orders dropped, which lines went missing, when you last visited. Bring them on the phone.
- Visit within 14 days of the flag. Each week of delay is another week of buying habits setting somewhere else.
- Open with the data, kindly. 'I noticed your tile and adhesive orders slowed in the spring. Did something change on our side?' Say it as a question about you, not an accusation.
- Ask who else they buy from, and why. Price, delivery time, a missed delivery, a new buyer, a closed job. The answer decides the fix.
- Fix one thing in the room. Offer to check a price, expedite a delivery, or call the warehouse. A promise to look into it later is where most win-backs die.
- Meet the people who decide. If the buyer has changed or the owner's son now orders, meet them. Many drifts start with a new contact nobody introduced.
- Write the notes at check-out. Reason given, who you met, what you promised, and the date you will confirm it.
- Revisit in two weeks. Check the next invoices. If orders are not rising, change the offer or the person, not the speech.
Ask why at the visit and close the loop
Surveys reach the wrong people in a distributor. A buyer who is shopping around does not answer an email. The best moment to ask is the visit, in person, with the facts on the table. Keep the question set short enough to remember, and write the answer down at check-out.
1. I noticed [line] orders slowed since [month]. What changed? 2. Where are you buying that now, and what made it easier? 3. Has anything gone wrong with us in the last few months: a late delivery, a wrong price, a stock-out? 4. Who else should I be talking to about your orders? 5. If I fixed one thing this month, what would bring the orders back? Write down: reason in their words ____________ Reason tag: price / delivery / stock / service / contact left / job ended / competitor / other What I promised: ____________ Date: ____________
Closing the loop is the part most teams skip. Within a week of the visit, tell the customer what you did about what they said, even if the answer is that you could not change the price. Then count the reason tags every month. If 11 of 40 lost accounts name late deliveries, that is a warehouse problem, and no amount of visiting will fix it. Send the tally to whoever owns that problem.
Net Promoter Score and satisfaction surveys can add a signal once a year, but they come after the fact and from the customers who answer. Treat them as a supplement to the visit question, not a replacement. Customer visit report template has fields for the reason and the promise.
Win back lapsed customers who have stopped ordering
Some customers went from drifting to silent before anyone looked. Pull every customer with no invoice in the last 90 days who bought regularly before. Treat them as a separate list from the drifting ones, because the visit is different.
- Find out whether they stopped buying or stopped needing. A contractor between jobs and a contractor on a new supplier look the same on paper.
- Lead with what changed on your side. New stock, a new delivery day, a new rep, a price list. Reactivation works when you give a reason to try again.
- Ask for one small order. A single reorder of the line they used most. A trial order costs them little and starts the pattern again.
- Send a different rep if the relationship soured. A fresh face can reset it. History stays with the customer in your records, so the new rep walks in knowing what happened.
Be honest about what to expect. Some of these customers are gone for good: the business closed, or the owner has a long relationship elsewhere. Visit each once with a plan, record the reason, and move the account off the weekly list. Reasons collected across fifty lost customers will tell you more than any single visit.
Which lapsed customers to chase first
| How long lapsed | Chance to restart | What to do |
|---|---|---|
| 90 to 180 days | Best. The habit is not yet set elsewhere | Visit within two weeks, bring a reason to try again |
| 6 to 12 months | Lower. They have a new supplier routine | One visit with a specific offer, such as a line they used to buy |
| More than a year | Low. Treat as a new prospect | A call or a visit when the rep is nearby, and a note if the buyer has changed |
Inside each row, rank by what the account bought in its best year, then by whether the reason it left was something you have since fixed. The chances in the table are judgment, not measured rates, so check them against your own win-backs after a quarter.
A 60-day win-back sequence
- Day 1: read the history. Last order, lines bought, notes, the reason the account went quiet if anyone asked, and who the contact was.
- Week 1: call the buyer. Short and specific: 'We have changed the delivery day for your area. I would like to stop by.' Do not open with a sales pitch.
- Week 2: visit. Use the five questions above. Listen for what changed on their side.
- Week 3: send what you promised. A price check, a sample, a written delivery schedule.
- Week 4: ask for one trial order. The line they bought most.
- Week 6: visit again or phone. Check whether the order arrived and was right.
- Day 60: decide. If they ordered twice, move them back to a regular cycle. If not, record the reason and stop. A lapsed account gets three real attempts, and then it is a prospect.
A discount should be the last thing in the sequence, and only for a customer whose past volume justifies it. If price is the reason they gave, put a number on it first: what the price gap costs them a month against what the account is worth to you.
The first 90 days of a new account
Prevention starts when the account opens. The habits a new customer forms in its first quarter, including where it orders, who it calls and how often it sees your rep, are the habits you will be fighting to keep. Plan the first 90 days before the first order ships.
| When | What the rep does | Who to meet |
|---|---|---|
| Before the first order | Set up credit terms, the price list, the delivery address and how the customer will order | Buyer, and whoever pays the invoices |
| First order | Check the order is right and arrives on time. Visit within a week of delivery | The person who receives the goods |
| Day 30 | Ask for the second order. Fix anything that went wrong. Show one line they do not buy yet | Buyer, and the person who uses the product |
| Day 60 | Compare the lines bought with the lines their trade usually buys. Offer a plan for the gaps | The owner or approver |
| Day 90 | Set the account's tier and visit cycle. Write the contact map and the first account note | Everyone on the map who matters |
The second order is the number to watch. A customer who has ordered twice has formed a pattern. A customer who ordered once has done a trial. Count how many of your new accounts place a second order within 45 days, and visit the ones that have not. Account plan template can hold the 90-day plan for the larger ones.
Choose customers who fit, so fewer leave
Some churn is a selling mistake. A one-off project buyer, a price-only shopper, a customer outside your delivery zone or one too small to justify a visit will leave, however good the follow-up. Learn who they are from your own history.
- List the customers you lost in the last two years, and the ones you kept longest.
- Compare them on the things you know: trade, size of the first order, lines bought, distance from the warehouse, how they came to you.
- Write down the three traits the long-term customers share and the three the short-term ones share.
- Score new prospects against them before a rep spends a visit. The score is a guide, not a gate.
- Give the best-fit prospects to reps with room in their territory. A territory map shows where those customers cluster and who covers them.
A fit check also tells you when to stop. Do not chase a customer who will never order enough to cover the cost of the visits. Windshield time explains how drive time changes that math for distant accounts.
Pricing, terms and rebates that keep an account without a discount spiral
A customer who asks for a lower price is rarely asking only about price. Ask first, as the visit plan says. When price is the issue, give something that you can take back and that is tied to buying more. A cut with nothing in return teaches the customer to ask again.
| Lever | What the customer gets | What you get | Watch for |
|---|---|---|---|
| Volume tier with a year-end rebate | A percentage back if purchases pass a set amount | More volume, and a reason to keep buying until December | A tier set below what they already buy gives away margin |
| Price held for a period | No price change for 6 or 12 months | Fewer price negotiations | Costs can move. Write in a review date |
| Standing order or delivery schedule | Product arrives without the buyer having to remember | A steady order pattern and easier warehouse planning | Review the quantities every quarter |
| Bundle of lines | A better price when they buy a set together | More lines per account, which is harder to replace | Do not discount the line they would buy anyway |
| Payment terms | More days to pay | A reason to stay with you | Check credit first, and keep the same terms for similar customers |
Here is the difference in numbers, hypothetical. A customer buys $60,000 a year at a 25 percent gross margin, which is $15,000 of gross profit. If you cut the price 5 percent to keep them, you give up $3,000 with no extra sales. If you instead offer a 2 percent rebate on the whole year if they reach $72,000, and they do, the rebate costs $1,440. The extra $12,000 of sales brings $3,000 of gross profit, so you are $1,560 ahead and the customer has bought more lines from you. If they miss the tier, you pay nothing.
Mistakes that lose accounts faster
- Watching revenue instead of customers. Total sales hide losses behind gains. Look at each customer against its own history.
- Reviewing quarterly. A customer flagged in March and visited in June has already moved its buying. Run the check every week or two.
- Sending the discount first. Cutting the price before you know the reason teaches the customer to threaten you. Ask first.
- Visiting only the loud accounts. The customers who call you with problems are not the ones leaving. The quiet ones are.
- Leaving the data with the manager. If the rep never sees the flag, nobody visits. Put it on the rep's route.
- Treating every customer the same. A $400 a month account and a $5,000 a month account should not get the same visit cadence. See key account management for how to tier them.
Make the check a weekly habit
Prevention is a routine, not a project. On Monday, export last week's invoices and update the comparison. Flag customers who trip two or more signals. Add the ones worth a visit to a rep's route for the week, ahead of routine stops. Review what the visits found on Friday.
Sales Perch lets managers plan a rep's day or a route that repeats every week, every other week, every fourth week or every month, so the regular cycle stays on the calendar while the flagged accounts get added to it. Sales route planner shows how a rep's week fits together. For your biggest customers, add a written plan: account plan template has one you can copy.
Start this week with one list. Pull the last twelve months of invoices, find the ten customers with the biggest monthly gap, and send a rep to each within 14 days with the numbers in hand.
