Bad CX practices rob your growth potential by misleading managers.
GIGO is an acronym for “garbage-in, garbage-out”: in any system, the quality of output is determined by the quality of the input.1
AI relies on data quality: you assume AI outputs are truth, but if your data is junk, then AI recommendations are likewise junk.
Voice of the Customer (VoC) is — or should be — a primary input to customer experience practices: journey maps, human-centered design, messaging, engagement incentives, digitalization, sales conversion, onboarding, service, support, loyalty incentives, CRM, etc.
VoC should also be a primary input across your enterprise to decisions, development, reviews, and rewards of all types of products, policies, processes, plans, people, partners, and performance.
Why? Customers pay your salaries, budgets, and profit-sharing. Investors reinforce this only to the degree that customers’ spending increases.
So, is GIGO rampant in VoC? Yes, let’s explore this.
This is the second of a 7-part series:
Part 1) Insults: repel customers: the opposite of your goals.
1) Requesting a Score: disrespects customers’ truth.
2) Automating Interactions: makes CX hard in unique situations.
3) Upselling After CX Fail: is self-serving, not good CX.
Part 2) GIGO VoC: misleads managers.
4) Bias: shows your self-centicity.
5) Inconsistent Interpretations: lead to shaming customers.
6) Population Misrepresentation: makes CX data irrelevant.
4) Bias
The purpose of investing in customer analysis is to proactively manage truth better than your competitors.
You don’t know what customers think — and why they behave as they do — unless you study it.
- By observation (ethnography), focus groups, interviews with interactive exercises, surveys, and digital body language (e.g. recording calls for sentiment analysis, website visit sessions, apps usage, product usage, etc.).
- And by analyzing findings for actionable guidance to managers of all types in your company.
Bias is anything that might mask truth.
Biased invites to participants:
— “We trust this was an exciting session, and invite you to rate it.”
— “You’ll soon receive a survey, and we aim to excel in our service to you.”
— “If you cannot give a 9 or 10, please tell me now, so I can help you.”
— “Your top score means my kids will not miss out on Christmas presents this year.”
— “A low rating might cause me to lose my job.”
— “Get a gift card (or enter our raffle/drawing) by scoring us 5 stars.”
More examples of bias:2
— Hand-picking favorable participants or data sets.
— Collecting data only after positive situations.
— Averages in analysis rather than percentages.
— Initial groupings by preconceived categories rather than letting the data tell the story.
(e.g. Promoters/Detractors, high/low-volume, product line, geography, etc. should be secondary groupings after searching for new patterns in the data as primary grouping)
“I’ve seen brands mandate NPS from the C-suite down and do whatever it takes to get the number to be what they want it to be.” — Wanda Mills3
“I’ve seen it happen where a 7 or 8, due to local interpretation, have been marked as promoter scores, in order to create “realistic scores”. Now that’s a great way to cover up the issues at hand and block further improvement.” — Waios Estafthiou3
Really?!?
What’s so desperate to invest in collecting data, only to make the data untrue?
Super waste of everyone’s time and efforts, along with precious budget.
Keys to Banning Bias:
1) Remove surveys from compensation and performance evaluations.
Instead, use surveys to inform you what to manage proactively.
Track “What are you doing about it?” for performance and compensation.
2) Disallow “ads” in any statement about study participation.
Cease phrasing in the Bias list above.
The only “ad” allowed is:
“Based on past feedback, we improved X by Y%”.
Say: “We want to learn from your perspective.”
Say: “We’re curious to see your viewpoint.”
Say: “Let us know what’s important to you.”
Say: “Help us help you.”
Nothing more.
3) Instill curiosity among managers.
How can we manage smarter than competitors?
How can we stay ahead of disruptive competitors?
(e.g. Uber displacing taxis)
How can we free-up budget tied-up in responding to what’s bothering customers?
4) Learn data science for accurate analyses.
Data integrity.
Multivariate analysis.
Forrester reported in August 2025:
“Only half [of CX teams] feel confident in their ability to analyze what they collect.”4
“Only half of teams can successfully link CX metrics to business outcomes.”4
“Less than a third can set realistic targets.”4
“Even fewer report the ability to find signals in data.”4
Data quality essentials shown by Tech Target
To drive higher growth in 2026, stop GIGO:
- By stopping bias!
- By pre-testing consistent interpretations.
5) Inconsistent Interpretations
Your scales and phrases may mean different things to different participants’ personalities, cultures, use cases, and roles.
One-size-fits-all data collection is misguiding you.
Examples:
“Cultural context shapes the way people respond to metrics like NPS. In Brazil, there’s often an unspoken understanding that giving a low score could have real consequences for the person or team involved — sometimes even risking someone’s job. As a result, Brazilians may avoid giving ‘harsh’ feedback, even when they see room for improvement, which can inflate scores. Meanwhile, in Japan, there’s a cultural value placed on humility and continuous improvement. Japanese respondents may give a ‘7’ — not because they’re dissatisfied, but because they genuinely see it as encouragement for even better service. For them, ‘perfection’ is an ongoing pursuit, not a destination.” — Fabio Albiero de Faria3
“There will always be variation, even within a single country. I live in Brazil — it’s like many countries in one, with vastly different cultures. Expectations may shift, as well as what drives someone to give a certain rating. Not to mention differences among people of different ages, tastes, and life experiences.” — Gustavo Nava Stechinski3
Inconsistent interpretation of scales and phrasing is not just geographic — especially in B2B.
— End-users may differ from purchasers.
— English-as-2nd-language may differ.
— Frequent users vs. infrequent users.
— Different generations.
— Differing digital expertise.
— etc.
To fix this, it’s popular to educate participants.
Example:
“Colors, icons, and instructions can cause customers to feel ashamed or guilty or pressured to say something that does not truly respect their thoughts. Then your data is inaccurate. It’s a waste of everyone’s time and efforts.”5
The last thing you want to do in customer experience is make customers feel bad!
Anytime you’re using scales or words that aren’t natural for participants, you’re making it a self-centric ordeal instead of making it a positive customer-centric experience in itself.
Stop being heavy-handed: your data is biased.
Keys to Consistent Interpretations:
1) Pre-test scales and phrases.
Make a word strip for each study phrase.
Meet 1-on-1 different types of customers.
Say: “Arrange these any way you like.”
Be ‘a fly on the wall’ — don’t bias them.
Ask: “Tell me what it means?”
Take lots of notes! This is gold.
Change your scales and phrases to match what makes the most sense to them.
Who cares about industry norms or your trend data? Throw it out. It was GIGO.
2) Use question branching or separate questionnaires.
Use screening questions to identify differences.
Send participants to the right questions for them.
3) Use more almost-free data and less survey data.
90% of human thought is non-verbal.
So, why are we over-relying on verbal studies?
You have data overload in what they already told you.
You recorded every word and move (apps, web pages, etc.).
Data-mine this extensively.
Use surveys only to learn what you couldn’t otherwise.
To drive higher growth in 2026, stop GIGO:
- By stopping bias!
- By pre-testing consistent interpretations.
- By sampling for population representation.
6) Population Misrepresentation
Every customer is valued, yet:
- Value differs widely by customer.
- Viewpoints differ by customer type.
Your Marketing and Sales approaches respect this, but popular VoC practices do not.
Unless you get data from 100% of your customers, your report is a sample of the population.
What happens when you strive for response from 100% of customers?
— Happiest and angriest customers participate.
— Many types of customers are under-represented.
— It’s a waste of everyone’s time and money: GIGO.
Keys to Representative Participants:
1) Segment them the way Marketing does.
Which customers are spending more?
Focus on response rate for them.
You’ll be relevant to growth!
Findings will matter more.
Actions will grow ROI more.
2) Use a sample size table.
It’s in your stats textbook inside cover.
Use the 90% column (+/- 10%).
3) Use stratified random sampling.
Manage invitations separately by customer type.
Randomly invite participants.
Get response rates by sample size table.
4) Allow anyone to give feedback, any way, any time.
Welcome every customer to give you inputs.
Allow photos, sketches, video, audio, comments, ratings.
Data-mine what you get.
Look for patterns.
Stream findings to relevant parties in your firm.
8 Top Sins
In summary, stop these bad CX practices in 2026 for Voice of Customer, Voice of Employee, Voice of Partner:
1) Inviting everyone to your survey.
2) Asking how you are doing.
3) Using your phrases.
4) Color-coding scales.
5) Giving everyone the same questions.
6) Expressing your high standards.
7) Asking only for a recommend rating and comment.
8) Segmenting by demographics.
“These practices cause garbage-in, garbage-out in non-customer-centric products, processes, policies, and mindsets.”6
“Bad data collection practices cause missed opportunities, unnecessary costs, re-work, repetition, redundancies, scrap, friction, self-centered actions, lower growth, obsolescence.”6
It leads to more bad CX costs and less good CX growth:
“Businesses around the world risk $3.8 trillion in sales due to bad customer experiences — a figure $119 billion higher than last year.”7
“More than half (53%) of consumers say they will cut spending after a bad customer experience, and admit that one in 10 (12%) of their brand interactions don’t live up to expectations.”7
GIGO risks tremendous bottom-line and top-line growth.
Let’s stop GIGO CX practices in 2026 and beyond!
1What is garbage in, garbage out (GIGO)?, TechTarget, June 14, 2023.
2That NPS You’re Chasing? It’s Probably a Lie, by Wai Au, LinkedIn, 2025.
3Hot Take: NPS is Broken by Design, by Wai Au, LinkedIn, 2025.
4Six Gaps Hold Feedback Management And CX Measurement Programs Back, by Colleen Fazio and Maxie Schmidt, Forrester, August 8, 2025.
58 Voice of Customer Keys to CX ROI, by Lynn Hunsaker, ClearAction.
6Garbage-In, Garbage-Out (GIGO) Stunts Growth, by Lynn Hunsaker, LinkedIn, January 27, 2025.
7Bad Customer Experiences Put Nearly $4 Trillion at Risk in Global Sales, Qualtrics, November 19, 2024.
This is the second of a 7-part series: 26 Bad CX Practices to Stop in 2026.
- Part 1: Insults
- Part 2: GIGO Metrics
Coming Soon
- Part 3) Benchmarking
- Part 4) Prioritization
- Part 5) Leading Indicators
- Part 6) Goal Silos
- Part 7) Metric Silos
Image licensed to ClearAction Continuum by Shutterstock.