The Stakeholder’s Guide to Information Architecture in the Language of Business

The Hidden Costs of Poor Information Architecture

Boards and executives respond to numbers, so let me give you some. A 2022 study of 6,000 online shoppers across the United States and Europe found that 60 percent had abandoned a purchase due to a poor user experience. Within that group, 37 percent cited poor navigation or page layout as the specific reason. The same research showed that 42 percent of shoppers decide whether to stay on a Web site within ten seconds, and 20 percent decide within five.

The mobile navigation picture is sharper still. Research from the WP Dean usability program reports that mobile abandonment rises to 61 percent when hamburger menus require more than two taps to reach the target content, as shown in Figure 1. This is not a styling problem. This is an architecture problem dressed up as a styling complaint.

There are five common reasons that customers leave digital products. Three of them are direct symptoms of poor information architecture. Internally, findability problems also bleed your budget. Jakob Nielsen and Peter Morville modeled the cost of poor enterprise IA decades ago, using a simple formula:

Number of employees X average salary X average number of intranet pages visited per day X seconds of confusion per visit

Plug in real numbers from a mid-sized company and the annual cost could run into the hundreds of thousands. This cost is invisible because nobody bills for it, but it shows up in slower decisions, repeated questions, and shadow systems that staff members build to work around the official system.

These figures matter because IA work is usually presented as a soft investment with vague returns. It is not that. Every menu item, every category label, and every search results page either accelerates customers toward your business goals or slows them down. This slowdown has a price, and that price compounds across every visit.

IA Is a Strategic Issue, Not a Design Detail

Stakeholders sometimes ask whether you can delegate IA decisions to a content manager or junior designer. You cannot, because IA decisions cascade outward into territories that designers do not own.

Consider the label on a single navigation item. If your Web site calls a category Wealth Management because that is what your private banking division calls it, but customers search for investment advice, several things follow. Your search-engine optimization (SEO) underperforms because organic traffic searches for the customer’s language, not that of the business. Your marketing-copy efforts fragment because performance teams write according to what customers search for, while brand teams write according to what the executives say. Customer-service routing breaks because callers describe their needs in language the phone tree does not recognize. Sales-enablement materials inherit terms that prospects do not use, so sales representatives end up translating them in real time. This one IA decision has four downstream costs—none of which is visible in a design review.

Figure 2 shows two trees, side by side. The internal taxonomy reflects the organizational chart. The customer’s tree reflects their intent. Most Web sites publish the first and quietly punish the second.

This is why creating an IA requires stakeholder governance, not just a designer’s ownership. Someone with a view across product, marketing, sales, and support has to confirm that the language you use in a digital architecture matches the language of the people you’re trying to reach. That person is rarely the designer. It is usually a senior stakeholder who has either chosen to engage with IA work or perhaps chosen not to. The companies that win the long game choose to engage in IA.

How to Tell Good IA from Bad Without Being an Expert

You do not need a UX design certification to evaluate IA. You just need the three checks shown in Figure 3, all of which a nondesigner stakeholder can run in a single afternoon.

1. The five-second test—Show the page to a colleague who has never seen it before. Show it for just five seconds, then take it away. Ask your colleague what the page is for and what she thinks she should do next. If she cannot answer, the IA is failing this basic comprehension test that every customer applies on arrival.
2. The search-log review—Pull the last ninety days of internal site-search queries. Look at the top fifty terms. If these terms do not match the labels in your top navigation, you have a vocabulary mismatch. Customers are telling you, in their own words, what they expected to find and where they expected to find it. The gap between their words and your labels is the size of your IA problem.
3. The path analysis—In your analytics tool, look at the most common user journeys to your most valuable conversion pages. Count the clicks. Count the back-button presses. Count the dead ends. Anywhere users repeatedly moved backward or sideways, the IA is making them work for something that should be effortless.

Any senior stakeholder can lead this three-check IA evaluation, without any specialist tools or training. Together, these three checks can tell you more about an IA than any review of design comps. They also reframe the conversation. Instead of asking the designer whether the structure is good, you arrive with evidence about whether customers can use it.

Five Questions Every Stakeholder Should Ask

When a designer presents IA work, the questions you ask determine whether you receive a surface validation or genuine insights. The following five questions are enough:

1. What does our search data tell us about how customers describe what we sell? This question forces the designer to reconcile an internal taxonomy with external vocabulary. Anyone who cannot answer it has not done their foundational IA research.
2. How did we test this structure with real users and with how many? Tree testing and card sorting are inexpensive. If the answer is “We did not,” you are looking at one designer’s hypothesis that is dressed up as a recommendation. A confident IA proposal cites at least fifteen to twenty users who have interacted with the structure before the IA brought it to your desk.
3. What customer task is each top-level navigation item supposed to support? If the designer cannot match each menu item to a job the customer is trying to do, the menu reflects internal thinking. Every top-level label that exists for political reasons rather than user reasons is a tax users pay every day.
4. What happens to customers whose mental model does not match this structure? A good IA includes safety nets, including search, related links, contextual cross-links, and clear error states. If a structure assumes everyone arrives with the right vocabulary, it will fail real visitors who almost never do.
5. When will we test this again? IA is not a one-time deliverable. Customer language evolves, products change, competitors reframe categories. Conducting an IA review every twelve to eighteen months is reasonable. Every five years is negligence.

How to Invest in IA Without Overspending

The most common stakeholder mistake is treating IA as either too small to fund or too academic to deliver value. Neither is true.

Industry benchmarks suggest that you should allocate 10 to 20 percent of a digital products’ total budget to UX work, including discovery, IA, interaction design, and validation. Within that UX budget, IA-specific work—including IA research, taxonomy development, card sorts, tree tests, and labeling reviews—typically lands in the 3 to 5 percent range of the total project cost. Spending less moves the cost downstream into development rework, customer support, and lost conversions. Spending much more rarely produces additional returns unless you are operating in a complex, regulated domain such as healthcare, financial services, or enterprise software.

For ongoing operations rather than new builds, treat IA as a quarterly-review activity. Doing a two-day review that a senior IA practitioner runs every quarter typically costs less than a single targeted advertising campaign and protects the value of every campaign you run over the next three months.

When choosing whom to involve, prioritize IA practitioners who can show you a search-log analysis and a tree-test report from a previous engagement. Anyone who shows you only a sitemap is selling you the artifact, not the thinking. Creating the artifact is the easy part. The thinking is what you’re paying for.

The Quiet Competitive Advantage

Information architecture is invisible when it works well. Customers find what they need, complete what they came to do, and leave after solving their problem. But they won’t be writing you a thank-you message about your menu structure.

This invisibility is exactly why IA is a competitive advantage that few companies bother to develop. The companies that lead in their product categories are not always the ones with the best products. They are often those whose customers never get lost looking for their products. They have decided, at the stakeholder level, that the way in which they organize their offerings is as important as what their offerings contain.

Treat IA as a business decision rather than a design output. Ask the questions I’ve discussed in this article. Run the three checks. Fund the IA work properly. You’ll recover revenues that competitors are still leaving on the floor, and you’ll do this without a single new feature, ad campaign, or product launch. You’ll simply make it easier for the people who already want to buy an organization’s products to find what they came for. This is what good information architecture does. It is also what a good business does.