NPS survey best practices treat an automated score as a trend line, not as the reason behind it. The reason takes a phone call. Call detractors, passives, a sample of promoters and silent key accounts, and never let the score decide who is asked for a review.
In This Article
NPS survey best practices keep the automated score as the trend line and add phone follow-up calls to learn the reason behind it. An automated survey shows where B2B customers sit on a 0 to 10 scale and whether that number moved. It cannot say why. In an account where several people share the buying decision, it cannot say whose view the number holds either.
A phone follow-up adds the reason. The market research outsourcing service page describes phone administration of NPS and CSAT surveys on a continuous or scheduled cadence. Detractors are flagged as the scores arrive. The sections below follow the order in which a research lead meets this decision, from the gap to the cost.
Twenty terms carry the argument, and each keeps one meaning from the first call to the final report.
- Net Promoter Score (NPS): A net promoter score NPS survey asks how likely a customer is to recommend the company, on a 0 to 10 scale. The score is the share of promoters minus the share of detractors, as Bain’s page on measuring the Net Promoter Score sets out.
- Promoter, passive, detractor: Respondents who score 9 or 10, 7 or 8, and 0 to 6 on the NPS scale.
- CSAT: A customer satisfaction rating of one interaction or one attribute, scored 1 to 5 in this post.
- Relationship survey: An NPS or CSAT survey sent on a fixed calendar that asks about the account as a whole.
- Transactional survey: A survey triggered by one event, such as a closed support ticket, a delivery or the end of onboarding.
- Study: The whole research project, from objective to report.
- Sample frame: The list a sample is drawn from: here, the automated survey’s respondents plus the key accounts that stayed silent.
- Sample: The specific contacts chosen from the sample frame to receive a follow-up call in one round.
- Follow-up call: One phone interview after the automated score, with a respondent or a silent key account, to learn the reason behind the score.
- Fielding: The period in which an automated survey is open or follow-up calls are being made.
- Wave: One round of fielding against one sample, for the automated survey or for the follow-up calls.
- Complete (n): A follow-up call finished through the last required question; n is the count of completes.
- Instrument: The written questions in their fixed order, with the probes an interviewer reads after each score.
- Disposition: The recorded outcome of one sampled case, such as no answer, refusal or complete, under AAPOR’s scheme.
- Contact rate: The share of all sampled cases in which an interviewer reached the named contact, as AAPOR’s outcome-rate formulas define it.
- Cooperation rate: The share of reached and eligible contacts who finished the call, again under AAPOR’s definitions.
- Margin of error: The range around a score that sampling error alone can explain, at a stated confidence level.
- Saturation: The point where new calls stop adding reasons.
- Verbatim: The respondent’s answer, captured word for word.
- Coding: Sorting verbatims into a fixed list of reasons, the codebook, so the reasons can be counted.
What does an automated NPS survey score tell you, and what does it leave out?
An automated NPS survey score tells you how respondents split across the bands and whether that balance moved. It leaves out the reason, the customers who stayed silent, and whose view the score holds.
That first half is worth having. A CSAT survey after every support ticket and a relationship NPS survey twice a year give a trend line at little cost. The trend is where questions start. On a small base, a move between two waves can sit entirely inside the margin of error.
The second half is the problem. The reason behind a 6 sits in an optional comment box, which gets whatever a busy buyer has time to type. That is a word, a phrase or nothing. Silent customers leave no trace in the score. Who answers by mode, and the denominator behind each mode’s rate, is laid out in the phone versus online survey comparison.
B2B adds a third gap. A consumer’s score speaks for one household, but an account’s score speaks only for whoever opened the email. The champion who chose the product, the finance lead who signs the renewal and the end users can hold three different views. The automated survey holds one.
None of this makes the score wrong. It makes the score a question rather than an answer, and the next decision is when that question earns a call. Check the margin of error before calling anyone.
When is an automated NPS survey not enough?
An automated NPS survey is not enough when the score moves and nobody can name the cause. The same holds when the largest accounts stay silent, or when one contact stands in for a buying group.
Each signal shows up in the data first. Five signals mark the point where the score needs a reason attached, and each one can be read off the dashboard:
- The score shifted between two waves, and the comments name no cause behind the shift.
- The largest accounts by revenue answered least.
- Detractors cluster in one segment or region.
- The comment text repeats a word such as support or pricing without saying what about it.
- A renewal, a price change or a product decision depends on what the score means.
None of them replaces the automated survey. Teams that automate NPS surveys for customer feedback keep the volume and the trend line, and the phone layer sits on top of both. It calls the respondents who carry the decision.
Sizing that layer is a scoping question. The market research outsourcing page sells phone research in hour blocks sized to each project. The same page names four cost drivers: sample size, call length, respondent type and reporting requirements. The wave’s question sets each one.
When none of the five signals is present, the automated score is doing its job. A phone wave then buys unused detail.
After an NPS or CSAT survey, whom should you call?
Call detractors first, then passives, then a sample of promoters, and add the key accounts that never answered. Each group has a different job.
Table 1 sets out the plan by group.
Group | Why call | What the call is for | Order |
|---|---|---|---|
Detractors (0 to 6) | The score carries risk and the reason is missing | The event or pattern behind the score, and whether it is still open | First, within days of the score |
Passives (7 or 8) | The group closest to moving either way | What would move the score up, and what a competitor offers | Second |
Promoters (9 or 10), sampled | What earns the 9 or 10 is the thing to protect | The concrete reason, in the customer’s wordss) | A sample, called to saturation |
Key accounts with no response | Silence from a large account is a finding of its own | Whether the survey reached the right person, and that person’s view | Alongside detractors |
Other contacts in a scored account | One respondent does not speak for a buying group | The view of the finance, user or executive contact | When the account’s score drives a decision |
Two rules keep the plan honest. At the start of the call, the respondent hears who the research is for and whether answers are attributed to them. Names stay with the research team. A detractor’s name reaches the account team only with that detractor’s spoken agreement on the call. Both rules track the Insights Association Code of Standards, which bars acting on your respondents’ answers without their consent.
The second rule separates research from rescue. A retention conversation is the subject of the post on renewal and re-engagement calls, and it runs on its own list afterward.
Sizing follows the wave’s question. To explain a drop among detractors in one segment, every reachable detractor in that segment gets a call. Promoter calls continue to saturation. Neither group is sized to a margin of error, because a follow-up wave explains a score rather than estimating one.
AAPOR’s Standard Definitions set how every sampled case’s final disposition is recorded, from no answer to refusal to complete. Contact and cooperation rates come from those dispositions. Put plainly: count everyone you tried to reach.
For your follow-up wave, a reason count means nothing without the dispositions behind it for that group. An illustrative wave shows the failure. It reports 40 detractor completes and never states that 200 detractors were sampled for the wave. Its reasons describe the reachable fifth.
Three safeguards follow from the standard. Record a disposition for every sampled case, from a wrong number through to a complete. Report contact and cooperation rates per group. Then compare reached and unreached detractors on what the automated survey already holds: score, segment and revenue.
The order matters less than the coverage. A wave that calls only detractors explains the bottom of the distribution, and the next wave has no baseline for what promoters were protecting.
Element | NPS survey email example (as typed) | Sample NPS survey question (as spoken) |
|---|---|---|
Stem | How likely are you to recommend [Company] to a friend or colleague? | On a scale from 0 to 10, where 0 means not at all likely and 10 means extremely likely, how likely are you to recommend [Company] to a colleague? |
Scale | Eleven buttons, both ends labelled on screen | Both anchors spoken once; repeated if the respondent asks |
Follow-up | Optional text box: What is the main reason for your score? | Required probe, read after the number is recorded: You gave a [n]. What is the main reason for that number? Then one neutral follow-up: Anything else? |
Position | First item on the form; nothing precedes it | After the screener and before the attribute ratings, so no rating primes it |
Three details separate the NPS survey examples in Table 2, and each one does work on the phone. To a colleague replaces the friend-or-colleague object. A B2B respondent recommends to peers, and the shorter object is easier to hold in memory while the scale is still being read. The probe repeats the number back. That tells the respondent the rating was recorded, and it asks for the reason behind that number rather than a general opinion. And the probe is required. That is the point of asking by phone at all: the score sorts the respondent, and the reason tells you what to do. Sample NPS survey questions that work in a form, such as what could [Company] do better, produce a sentence on the phone. The spoken probe produces a reason.
A planned post on writing survey scripts covers the wording around the items: the introduction, the transitions, the respondent who answers out of order.
Tell us your research scope
A follow-up wave starts from a decision. Bring the segments, the accounts and the deadline, and the scope comes back as a call plan and an hour block.
What is the best time to send an NPS survey, and how often?
The best time to send an NPS survey depends on which of the two kinds it is. A transactional survey goes out right after its event, and a relationship survey runs on a fixed calendar clear of renewals.
Transactional timing is the easier half. A CSAT rating about a support ticket measures that ticket only while the customer remembers it. It goes out the day the ticket closes. The follow-up call for a low rating belongs within days, while both sides can still retrieve the details.
Relationship timing takes more judgment. When to send an NPS survey in B2B comes down to two tests, and both concern the calendar. First, the account has used the product long enough to judge it, which rules out the weeks of onboarding. Second, nothing nearby will set the score. A survey that lands in the week of a renewal negotiation, an invoice dispute or a major release measures that event instead.
How often to send an NPS survey is an account question before it is a contact question. A B2B account gets the relationship survey, a CSAT survey after every ticket, onboarding check-ins and now a follow-up call. They land on the same few people. Count every touch against one account limit.
CSAT survey frequency follows the same rule. A rating after every ticket is defensible for the ticket, but a contact who opens four tickets in a month gets one survey. Ongoing cadence is a program decision. The customer insight surveys service page lays out monthly, quarterly and event-triggered schedules, built to keep the same customers from being over-surveyed.
Program cadence across a whole customer base gets its own treatment in the post on how often to call customers.
Dialing rules apply before the first question. Under the TCPA’s autodialer provision, an autodialed or prerecorded call to a mobile number on your list needs prior express consent. A live, hand-dialed call needs none. From November 25, 2026, the FCC’s revocation order lets you name one disclosed opt-out method for autodialed calls. Honor each revocation within ten business days. A follow-up carrying an offer counts as a sales call, and the state mini-TCPA reference guide lists the rules your wave then meets.
Which NPS survey best practices apply to the follow-up call itself?
NPS survey best practices for the follow-up call serve one purpose: getting the reason behind the score in the customer’s own words. Five practices protect that purpose.
- Name the caller and the sponsor first, and explain that the call is about the score the customer gave. Naming the score explains why the phone rang.
- Ask for the reason before anything else, then probe to one example: the ticket, the delivery, the meeting. A reason without an example cannot be coded.
- Ask whether anyone else in the account would score differently, and who that person is. That question maps the buying group.
- Capture the answer as a verbatim and code it afterward, so the interviewer listens instead of sorting. Recording follows state law. The guide to call recording laws by state sets out which states require every party’s consent before your call is recorded.
- Close without an offer. Nothing is sold, and nothing is promised beyond passing on what the customer asked to have passed on.
Item wording and probe design are covered in the guide to designing a B2B survey for phone administration, with a full instrument. The call returns two datasets. One is the rating; the other is the reason, with the example that makes it codable.
How do follow-up calls become NPS survey analysis?
NPS survey analysis begins when every verbatim is coded against one fixed codebook, so reasons can be counted by group. The counts then sit beside the score.
The codebook comes first. A short list of reasons, drafted from the first batch of calls and then frozen, lets two waves be compared. A codebook that changes weekly compares nothing. A second coder then recodes one verbatim in ten without seeing the first codes, and the disagreements show which codes are too loose.
The report carries three layers. The score and its movement sit on top, with the count of each reason by respondent group underneath. Below that sit three verbatims per top reason. Quoted as said, they let a reader check that each code means what it claims to mean. Count reasons per respondent, not per mention. That way, one talkative customer cannot outweigh three brief customers in the final reason counts.
Keep phone ratings apart from automated ones. In a 2026 randomized trial in the Journal of Medical Internet Research, 3,200 Korean hospital inpatients were assigned to telephone or mobile web. Telephone totals averaged 84.9; mobile web, 81.5. Hospital staff identified eligible adults after discharge, with no screener, for a 21-item survey of unreported length: a B2C setting. The authors’ untested explanation is social desirability. For your follow-up wave, a rating asked on the call goes in its own column and never into the automated trend line.
One reading rule carries the analysis. A reason named by detractors and promoters alike does not drive the score, and a reason named only by detractors does.
Do CSAT survey best practices allow asking happy customers for a review?
CSAT survey best practices keep review requests out of the follow-up call that is built on the score. Asking only satisfied respondents runs against FTC guidance, and a sentiment-tied incentive breaks the rule.
The rule is the FTC’s final Rule on the Use of Consumer Reviews and Testimonials, 16 CFR Part 465. It took effect on October 21, 2024. Section 465.4 bars compensation or other incentives conditioned on a review expressing a particular sentiment, positive or negative.
The FTC staff questions and answers on the rule cover the two practices a phone follow-up invites. The first concerns who is asked. Asking only customers a business thinks are happy has no specific ban in the rule. Staff say it can break the FTC Act. The second is paying a customer to remove a truthful negative review, which sits in the same place. Incentives for reviews are allowed under Section 465.4 only when nothing, stated or implied, requires a particular sentiment. The 2022 guidance goes further. The FTC’s guide for marketers on soliciting reviews tells businesses not to ask only the customers expected to leave positive reviews. Put plainly: the score cannot pick the reviewers.
For your follow-up program, three operating facts follow. A phone interviewer hears the score first, which makes a gated review request simple to run and hard to catch in a script review. A detractor’s credit must carry no review condition. Whether a business buyer’s rating on a B2B review site counts as a consumer review under the rule is a question for counsel. Staff place gating and removal under the Act. Since the Supreme Court’s Loper Bright decision in 2024, courts settle a statute’s meaning themselves and treat agency views as guidance only. Your program cannot lean on staff answers alone.
A 2022 case shows the pattern. In January 2022 the FTC alleged that Fashion Nova misrepresented its product reviews, posting four- and five-star reviews while holding back lower-rated ones. The conduct ran from 2015 to 2019. It was the FTC’s first case over concealed reviews; the company paid $4.2 million in a settlement and disputed the allegations.
The final rule’s preamble records more than 4,500 merchants that automatically published only 4- or 5-star reviews. Enforcement under the rule has begun. On December 22, 2025, FTC staff sent warning letters under the rule to 10 unnamed companies. The published template warning letter cites civil penalties of up to $53,088 per violation of the rule. The letter is not a formal determination.
Three safeguards follow from the rule. Keep review requests out of the follow-up call, and send one invitation to every customer through the client’s own channel. Route each detractor’s remedy through the account team. Log it, and attach no review condition to it in the script, the email or the account manager’s follow-up. Counsel reads the closing script and any incentive language before the first wave goes out.
This describes the rule and staff guidance. It is not legal advice, and a program’s own facts decide how the rule applies to it.
When is a phone NPS or CSAT survey worth the extra cost?
A phone NPS or CSAT survey is worth the extra cost when a decision rides on what the score means. The automated comments must also leave it unexplained.
The four cost drivers apply to a follow-up wave exactly as they do to any other phone study. A follow-up wave keeps sample and length small. It calls a targeted group with a short instrument instead of re-surveying every customer by phone.
Who calls matters as much as how many. A customer explaining a 5 to the account manager who sold the contract has a renewal price riding on the answer. A neutral interviewer has nothing to trade. Wording that gets candid answers is the subject of the post on writing survey scripts that get honest answers.
Ask any provider for five things:
- Contact and cooperation rates for each respondent group, calculated under AAPOR Standard Definitions rather than a house formula.
- The frozen codebook, plus the share of verbatims that a second coder changed on review.
- Three recorded calls, chosen by the buyer.
- The median call length and hours per complete.
- The script’s opening and closing lines, read aloud.
A provider that cannot produce all five is selling a score, not a study of what drives it.
A quarterly wave is a program. At that point the timing and frequency rules above become its operating rules, and the cadence question moves to the program.
Frequently asked questions
Is an automated NPS survey enough, or do we need phone follow-up calls to find out why B2B customers gave the score they did?
An automated NPS survey covers the trend. Phone follow-up calls earn their cost when the score moves without a named cause, or when large accounts stay silent. A renewal riding on the reason counts too. A follow-up wave aimed at detractors, passives, a promoter sample and silent key accounts adds the reason without replacing the automated survey.
What is the best time to send an NPS survey, and how often can we survey the same B2B account before response quality drops?
A transactional survey goes out right after its event, such as a closed support ticket. A relationship survey runs on a calendar. It goes out once the account can judge the product, and away from renewals, invoice disputes and major releases. Frequency is set per account. Every NPS survey, CSAT survey and follow-up call to that account counts toward one shared limit.
After an NPS or CSAT survey, should we call the detractors, the passives, the promoters, or the customers who never answered?
A follow-up plan calls all four groups, each one for a different reason and with a different job. Detractors come first, for the event. Passives come next, for what would move them, and a sample of promoters shows what to protect. Silent key accounts are a finding. They are called alongside the detractors to learn whether the survey reached the right person at all.
When is a phone NPS or CSAT survey worth the extra cost over an email or in-app survey?
A phone NPS or CSAT survey is worth the extra cost when a decision depends on the reason behind the score. The automated comments must fail to give it. Cost scales with sample size, call length, respondent type and reporting, so a follow-up wave stays affordable by calling a targeted group. Its instrument stays short.
Can an NPS or CSAT follow-up call ask happy customers to leave a review?
Asking only satisfied respondents for a review runs against FTC staff guidance on the review rule. Staff say it can break the FTC Act. An incentive tied to a positive review breaks 16 CFR 465.4, whether the condition is stated or implied. Keep review requests out of follow-up calls. Send one invitation to every customer instead, through the company’s own email or account channel.
Discuss your research project
Some scores explain themselves. A short scoping call sorts out which of yours need the phone, and how many calls that takes.
