A customer pays for a bike service on Tuesday afternoon, chats with the mechanic about the new brake pads, and walks out happy. On Friday morning an email arrives in their inbox: "How likely are you to recommend us to a friend or colleague?" It sits between a shipping notification and a newsletter, and by then the visit is a three-day-old memory. Most recipients never open it. The few who do are disproportionately the delighted and the furious, so the monthly NPS swings ten points on a handful of responses and nobody trusts the number.
The email NPS survey has a second, quieter problem: it can only reach customers whose addresses you hold. In walk-in retail, cafes, pharmacies, service desks, and repair shops, that is often a minority of the people who actually stood at your counter. The program samples your mailing list, not your customers.
There is a simpler place to ask the question: the counter itself, in the twenty seconds while the card reader thinks. A printed NPS card and a pen capture the score at the exact moment the experience ends, from customers present at the counter, including some you cannot contact by email. The historical catch, typing hundreds of cards into a spreadsheet, is what PaperSurvey.io largely removes. You print the card, collect it, scan the stack, and the promoter, passive, and detractor counts land in a live dataset with the NPS already calculable in one click or one formula.
Why the inbox dilutes your NPS
Net Promoter Score is a difference between two percentages, promoters minus detractors, which makes it unusually sensitive to who answers. The score depends on who responds, so assess the collection process as well as the arithmetic.
- Low and falling response: across a meta-analysis of 114 experimental comparisons, web surveys averaged about 12 percentage points lower response than other survey modes such as mail, phone, and in-person (Daikeler et al., 2020). That average does not predict the response rate of a retail NPS program.
- Delay erodes the signal: the score you receive on Friday measures the memory of Tuesday, filtered through everything that happened in between. NPS is supposed to reflect the experience; a lagged survey reflects the recollection.
- Selection bias at the extremes: people with strong feelings click through, the satisfied middle deletes the email. With a small, skewed sample, the promoter and detractor percentages both wobble, and the score whipsaws month to month.
- Coverage stops at your mailing list: cash customers, guest checkouts, and anyone who unticked the marketing box simply never get asked.
None of this means the NPS question is broken. It means the channel is. For the broader evidence on paper versus screen collection, see our comparison of online and paper survey response rates.
The moment of experience is at the counter
The counter has something no email campaign can buy: a customer who is physically present, seconds after the experience ended, with a natural pause in front of them. A card handed over with the receipt asks for two pen marks and a line of handwriting before they pocket their wallet.
- Broader invitation coverage: staff can offer a card to visitors without collecting an email address. Refusals, missed visitors, timing, and staff selection can still bias the responding sample.
- The answer is fresh: the customer rates the visit that is still in front of them, when you are asking specifically about that visit. Keep the reference period consistent and distinguish visit-level feedback from a question about the overall relationship.
- No device friction: no phone to unlock, no page to load, no login. A paper option can help customers who cannot or prefer not to use a web form; accessible language and print still matter.
- A visible ask carries weight: a card handed over by a person signals that the answer will be read. A no-reply email signals the opposite.
Consistency matters more than enthusiasm here. Ask every customer, or every nth customer during busy periods, rather than letting staff choose. A card offered only to smiling customers manufactures a flattering score and defeats the point of measuring at all.
What a scannable NPS card looks like
An NPS card is the easiest survey you will ever design, because the methodology fixes the core question for you. The craft is in making it scan cleanly.
- The 0 to 10 scale as eleven checkboxes: one row, clearly labeled from "not at all likely" to "extremely likely," marked with a single cross. Optical mark recognition reads the selected rating from the scan.
- One follow-up question: "What is the main reason for your score?" with a generous comment box. AI handwriting recognition transcribes cursive and print, so the reasons behind the number arrive as searchable text rather than a shoebox of cards.
- A location or till number field: a small numeric box, filled in by staff or pre-printed per site, lets you segment the score by branch later.
- A unique per-page identifier: printed automatically on every card, readable codes connect each scan to its questionnaire and help detect duplicates. Check missing or unreadable codes before treating the batch as complete.
Keep it to one side of a small card. The instrument is two questions; resist the urge to bolt on a satisfaction grid, or you trade away the thirty-second completion time that makes counter collection work.
From a drop box of cards to a calculated score
Collection stays deliberately low-tech: a stack by the register, a pen, and a drop box so nobody has to hand a critical answer back to the person they are criticizing. The automation starts when the box gets emptied.
Batch the cards weekly, or nightly in high-volume locations, and capture them however is convenient: feed them through the office scanner or copier, email the scans to a dedicated address, drop a PDF in from Dropbox, or have a manager scan the stack with a phone's document scanner. OMR reads the ratings, and a card with two scores ticked on the 0 to 10 scale is flagged for a person to resolve against the scan. Clear those, and glance through a sample of accepted cards, before the week's NPS goes to the branch managers.
From there the arithmetic is the standard one. Respondents scoring 9 or 10 are promoters, 7 or 8 are passives, 0 to 6 are detractors, and NPS equals the percentage of promoters minus the percentage of detractors. Because every scanned card is already a structured row, the chart view shows the distribution as it accumulates, and an export to Excel or Google Sheets turns the calculation into a one-line formula you can pivot by week, by till, or by branch. Watch the sample size alongside the score: there is no universal cutoff at 15 or 400 responses. Consider uncertainty, selection, and the decision you intend to make.
Running a counter NPS program in PaperSurvey
You do not need to design the card from scratch. The NPS Survey Card template below is ready to print, or open it in the editor and make it yours:
View the full NPS Survey Card template.
Swap in your logo, adjust the follow-up wording, add a store-number field, and print on plain paper from an office printer, with no proprietary forms or scanning hardware to buy. The same instrument also runs as a web survey, so delivery customers or online orders can answer the identical two questions through a link, and paper and web responses land in one combined dataset instead of two spreadsheets you reconcile by hand.
As scans arrive, responses appear in the dashboard with the 0 to 10 distribution charted, comments transcribed, and everything filterable by the location field. Exports cover CSV, Excel, SPSS, R, and Google Sheets, and webhooks or the REST API can push each new detractor straight into a ticketing or alerting workflow, so a 2 written at the counter on Saturday becomes a callback on Monday morning. For the wider point-of-service picture beyond the NPS question, see paper customer satisfaction surveys that scan to a dashboard.
The Bottom Line
A counter card can support visit-level NPS collection when that is the reference you intend to measure. A printed card at the counter asks the question while the experience is still on the counter between you, offers another way to respond, and has printing, collection, and review costs you can estimate in a pilot. Scanning removes the only real objection paper ever had: nobody types up the stack, the promoter and detractor counts assemble themselves after a quick review, and the score you report is built on hundreds of fresh answers rather than a trickle of delayed ones.
References
- Daikeler, J., Bosnjak, M., & Lozar Manfreda, K. (2020). Web Versus Other Survey Modes: An Updated and Extended Meta-Analysis. Journal of Survey Statistics and Methodology, 8(3), 513-539.
Start your free trial and put an NPS card next to your register this week.