---
title: Paper Cards or Screens? How to Choose the Right Medium for Card Sorting
lang: en
source: https://mindsprt.dev/en/knowledge/research-brief-paper-vs-digital-card-sorting/
---

# Paper Cards or Screens? How to Choose the Right Medium for Card Sorting

*Mai Strategy Lab · 8 min read · 2026-08-02*

> Whether structuring an information architecture, catalog navigation, or website menus, card sorting is almost inevitable. But should you hand participants a deck of physical paper cards, or send them an online tool? This article breaks down the cognitive conditions, data quality risks, and ideal scenarios for each medium so you can make the right choice at kickoff, before finding out your data is unusable

**Quick answer:** Whether structuring an information architecture, catalog navigation, or website menus, card sorting is almost inevitable

## Overview

A client wants to redesign a 120-page product catalog and revamp their website navigation at the same time. You propose running a round of card sorting first, writing item names onto cards and letting internal sales reps and distributors group them. The goal is to see how far their mental models sit from the company's internal part-number logic.

A problem pops up immediately: participants are scattered across three cities. Some are willing to visit your office, while others can only join online. Do you force everyone around a single table to keep the medium consistent? Or do you mix things up with half on paper cards and half on digital tools?

This is not an administrative dilemma, but a methodological one. Whether the medium influences sorting results determines if you can pool the two datasets together for analysis. Here are four practical questions that unpack the decision criteria.

## Do Paper Cards and Computer Tools Really Produce Different Results?

Differences exist, but not enough to render either approach invalid, according to comparative research. A study systematically comparing computerized and paper-based card sorting setups addressed this exact question, focusing on whether categorization results from both mediums can be considered equivalent [1].

In practice, the difference lies less in accuracy and more in how the sorting process unfolds:

・With a hundred paper cards spread across a table, participants spot unassigned cards in their peripheral vision and freely break apart existing piles at any time.

・Digital tools are constrained by scrolling, pagination, and drag-and-drop viewports, so participants tend to file cards they see first and slot later cards into existing piles.

In concrete terms, paper sessions often spark unexpected third categories, whereas digital sessions tend to consolidate into a few broad groups. Paper works better when exploring unknown structures, while digital is cost-effective and sufficient for validating existing setups.

## When Should I Use Paper Cards, and When Should I Go Digital?

The decision hinges on research intent, not budget or participant locations. Split the scenarios into two distinct buckets:

When to choose paper cards

・Open sorts where you are still unsure how many categories should exist.

・Think-aloud protocols where you want to observe hesitation and second-guessing, not just the final sorting outcome.

・Card items require tactile cues, such as paper weight, texture, or post-press finishing samples. Here, cards are not mere information carriers but the actual physical products, and going digital erases test variables.

・Small sample sizes (10 to 15 participants) where your goal is qualitative interpretation rather than statistical clustering.

When to choose digital tools

・Closed sorts where categories are predefined and you need to verify item placements.

・Distributed participants and sample sizes scaled past 30 people for statistical analysis.

・Step-by-step logs are needed, like how many times each card moves and how long participants pause, which paper sessions cannot capture without video recording.

For that catalog redesign project, a sensible approach is to run paper sessions with 8 to 10 internal sales reps and veteran distributors to establish a candidate architecture, then run a closed digital sort with 40 or more external customers to see if the structure holds up against real users. Two phases, two mediums, two goals, rather than jumbling the two datasets together.

## What Goes Wrong When You Mix Both Mediums?

The biggest risk is not about validity, but mistaking differences in medium for differences among participants.

If the paper group creates seven categories while the digital group produces only four, you cannot tell whether the two groups have different mental models or if the two mediums simply impose different interaction costs. In mixed data collection, the medium turns into an uncontrolled variable entangled with the exact behavior you want to observe.

A pragmatic approach operates on three levels:

1. Avoid mixing whenever possible. Keep the medium consistent within a single round of research.

2. If mixing is unavoidable, record the medium as a dedicated field. Stratify your analysis first to confirm that structures match before merging datasets.

3. Disclose the breakdown of mediums in your final report. This satisfies basic reproducibility requirements and sets clear boundaries for anyone referencing your findings.

It is worth clearing up a common terminological mix-up here: "sorting data / selecting cases" in data analysis software refers to sorting and filtering observations within existing datasets [2][3], which belongs to post-collection analysis. Meanwhile, card sorting in design research is a data gathering method aimed at generating categorical structures [1]. Both use the word sorting, but one happens before data generation and the other happens after. Mixing these keywords during cross-disciplinary literature searches will pull in lots of irrelevant results.

## As Digital Tools Improve, Do Paper Cards Still Have Irreplaceable Value?

Yes, primarily under one specific condition: when the items being sorted possess physical materiality.

When sorting paper swatches, post-press finishing samples (foil stamping, embossing, spot UV), or business cards of varying weights, participants base their groupings on tactile feel, thickness, reflection, and sound. These attributes cannot be digitized on screen, not because rendering is poor, but because digital mediums simply cannot transmit them. When investigating how customers mentally organize paper stocks, paper cards are not just one option, they are the only option.

A second scenario involves decision environments rooted in physical space. Exhibition traffic flows, retail shelf layouts, and catalog spread-flipping sequences share spatial isomorphic traits with cards spread across a table, giving paper cards higher simulation fidelity.

Conversely, if you are sorting pure information items like page titles, feature names, or article tags, physical materiality plays no role. Digital tools hold the upper hand in cost, scale, and data completeness. Clinging to paper cards in that situation mistakes ritual for methodology.

## Next Steps for Implementation

Write three elements into your research plan at project kickoff instead of waiting until the week before fieldwork:

・Whether the goal is exploration or validation determines an open or closed sort.

・Whether sorted items carry physical properties determines if paper cards are irreplaceable.

・Whether sample size and analytical approach favor qualitative interpretation or statistical clustering determines the medium and session count.

Once these three conditions align, the choice of medium usually boils down to a single sensible answer without further debate.

Scope and limitations: These criteria assume participants have no major operational hurdles with either medium. If digital proficiency varies widely among your target group, such as veteran production technicians mixed with young marketing specialists, the cognitive burden of digital tools will contaminate your findings. In such cases, prioritize paper cards or on-site assistance even when your goal is validation, and document this adjustment in your report. Likewise, in cross-linguistic or cross-cultural studies, non-medium variables may outweigh the medium itself, meaning the criteria in this article must be recalibrated.

## Key Takeaways

Differences between paper and digital card sorting center on the process and field of view rather than categorical accuracy. Existing research treats both as comparable methodologies [1].

Prioritize paper cards when exploring unknown structures, while digital tools are sufficient and cost less for validating established structures.

When sorted items involve tactile qualities, paper weight, or post-press finishes, paper cards cannot be replaced by digital mediums.

Keep the medium consistent within the same research round. If mixing is unavoidable, record the medium as an analysis field and examine layers separately.

Card sorting in design research and sorting/selecting cases in statistical software belong to different project phases. Avoid confusing keywords during literature searches [1][2][3].

## Further Considerations

For print manufacturers, choosing card sorting mediums highlights an overlooked reality: paper and finishing samples are research tools, not just price-quote attachments. Designing sample books in detachable, stackable formats turns customer categorization logic into visible data, creating direct value for product planning and catalog architecture. For designers, a two-phase workflow (paper exploration → digital validation) offers a cost-controlled standard process that avoids forcing qualitative methods into statistical expectations. The opportunity for AI integration sits in process data from digital sessions: drag trails, hesitation duration, and revision frequency. These signals are mostly discarded today, yet structured logging turns them into rare high-quality inputs for clustering models. On the SaaS side, the gap is obvious: current card sorting tools almost exclusively record final results, failing to capture the full process or provide stratified analysis for mixed-medium data. Whoever builds "medium as a controlled variable" into their product will fill the biggest missing piece in this methodology.

## References

[1] Goetz (2114). Different or alike? Comparing computer-based and paper-based card sorting. International Journal of Strategic Innovative Marketing. DOI: 10.15556/ijsim.01.01.003

[2] [Sorting Data and Selecting Cases](https://doi.org/10.1017/cbo9780511804786.006). Data Analysis Using SAS Enterprise Guide. DOI: 10.1017/cbo9780511804786.006

[3] [Selecting, Sorting and Weighting Cases](https://doi.org/10.4135/9781446249390.n10). Applied Statistics with SPSS. DOI: 10.4135/9781446249390.n10

## FAQ

### Do card sorting results differ when using paper cards versus online tools?

Differences exist, but primarily in the process rather than the final conclusions. Paper cards allow participants to view all items at a glance, making it easier to branch into unexpected categories. Digital tools, limited by scrolling and screen boundaries, tend to converge into a few broad groups. Existing comparative studies systematically tested the equivalence of both mediums [1].

### When is it necessary to use physical paper cards for card sorting?

When the sorted objects carry physical materiality. Factors such as paper weight, texture, or finishing samples like foil stamping and embossing cannot be communicated through a screen. Going digital in these cases erases core test variables.

### Can data from paper sessions and online sessions be combined for analysis?

Direct merging is not recommended. Differences in medium become entangled with participant differences, making it impossible to tell whether variations stem from mental models or interaction costs. If combining them is necessary, record the medium as an analytical field, check whether the two group structures align through stratified analysis before merging, and disclose the medium distribution in your report.

### Which mediums should be paired with open and closed card sorting?

Open sorts explore unknown structures and work best with paper cards, allowing participants to assemble and dismantle piles freely. Closed sorts validate established categories, where digital tools offer clear advantages in scale, cost, and process logging.

### Is card sorting in design research the same as sorting data in statistical software?

No. Card sorting in design research is a data gathering method used to discover organizational structures [1]. In statistical software, sorting data and selecting cases refer to operations performed on existing datasets during the analysis phase [2][3]. They occur at completely different stages of the research workflow.


---

> HTML version: https://mindsprt.dev/en/knowledge/research-brief-paper-vs-digital-card-sorting/
> MINDS — 麥思印刷整合有限公司 · https://mindsprt.dev