In Q1 2024, during a routine review of our search analytics, I found something that still bothers me. Two search queries on our official site that had absolutely nothing to do with our core products, but were pulling in steady traffic from real users.
I'm a quality and brand compliance manager for a telecom equipment manufacturer. My review process covers product specs, marketing collateral, packaging copy, help center content, and brand-voice guidelines. That's roughly 200+ items a year, and since 2021 I've rejected around 15% of first submissions—usually for inaccurate specs or inconsistent messaging. It's not glamorous work, but it's the layer that keeps small errors from reaching customers.
That Tuesday's review was supposed to be uneventful. We'd just wrapped a launch cycle for the ZTE Blade 10 phone and a new 5G CPE router. I pulled the search query data for our official site to see what users were actually looking for—half-expecting the usual mix of phone spec questions and router setup queries. Then two entries stopped me cold:
- "blood pressure monitor symbols"
- "what is on my wifi"
Neither query has any obvious connection to our products. We don't make medical devices. We make phones, routers, and network infrastructure. My first thought: bot traffic. Some analytics glitch. Maybe a scraper. But when I filtered for actual user sessions with click-through and dwell time, they didn't disappear. They showed up repeatedly, across multiple weeks. Real people, real queries, real frustration. That felt like a red flag worth chasing.
The Investigation
I went through session recordings first. Took about two hours to watch 40 or so sessions. Users started on our router setup pages, bounced to help articles, then wound up on our brand homepage—clicking back and forth, clearly trying to find something specific. Some entered through search engines, landed on our official site, and stayed for minutes without finding the answer.
The "what is on my wifi" question was the easier puzzle to solve. We have thousands of people using our 5G CPE routers as the primary WiFi gateway in their homes. The router management interface shows connected devices, but the labeling is vague. A generic icon with a MAC address and partial device name. If you're not technically inclined, that interface tells you almost nothing about what's actually on your network. So people go to Google, search for the answer, and end up back on our site—but on a page that doesn't help them.
The blood pressure monitor piece took longer to untangle. Turns out, some users have WiFi-connected health monitors, and those devices connect to our routers. When they open the router management interface, they see a list of connected devices identified by generic icons. The same little circle represented a phone, a laptop, and a medical device. Users searching for "blood pressure monitor symbols" were trying to work out which icon on their screen matched the device in their bedroom. That's a level of interface confusion we'd never accounted for.
Bottom line: our product worked perfectly. Our communication didn't.
The Gap I'd Missed
Here's the part that still gets me. Four years of quality audits, and my scope had never included the device management interface or the help center articles supporting it. I'd review the marketing copy for every phone spec comparison, every Blade 10 spec sheet, every router product page. But the interface users actually stare at? It never made it onto my checklist.
This was a textbook case of the "quality means polished marketing materials" thinking. That perspective comes from an era when the website was basically a brochure and the product lived in the physical box. That's changed, completely. For connected devices, the interface is the product. A user who can't identify their own devices on your router interface doesn't care that your marketing copy scored 95% in the brand consistency audit. They care that they can't find their blood pressure monitor.
I pulled our help desk tickets from the previous quarter. Roughly 34% of device-identification calls could have been avoided with clearer in-app labels and a better help article. The tickets themselves were vague—stuff like "which icon is my phone?" or "can't tell what's on my network." Each call cost us an average of $8 to handle. We were spending somewhere between $40,000 and $55,000 annually on support conversations that a $2,400 fix could largely eliminate. That's a painful leak when you finally spot it.
The Fix
We didn't rewrite the whole router interface. That would've been overkill and way too expensive for a partial problem. Instead, we made three targeted changes:
- Added distinct icons for known device categories—phones, laptops, printers, medical devices, smart home appliances
- Wrote a one-page help article titled "Understanding the Device List on Your ZTE Router," linked directly from the interface
- Added a "device type not listed?" button that opened a 30-second manual override where users could label devices themselves
Total cost: roughly $2,400 in design time plus 12 hours of my own review time. A no-brainer when you're looking at $50k+ in avoidable support costs. I'm glad we caught it, but three quarters of users asking the same question before we noticed is—honestly—too long.
Two months after launch, device-identification tickets dropped by 41% compared to the previous quarter. I also asked 10 users to navigate the new interface without instructions; 9 correctly identified their devices in under 10 seconds. That's not a statistically rigorous test, but it's enough to confirm we were heading in the right direction.
What I Actually Learned
Quality isn't a finish line. It's a moving target based on where your users actually spend time. I was genuinely proud of our phone product pages—accurate, consistent, aligned with brand standards. But some of our users weren't looking at those pages. They were staring at a router interface, trying to decode an icon, and eventually giving up and calling support.
The second lesson: search logs tell you more than surveys do. No one requested "better device labeling" in a customer satisfaction form. No one said "please add a blood pressure monitor icon" in a feedback survey. But their queries told me exactly where the comprehension gap was. I just had to stop filtering for terms I expected to see and start paying attention to terms that made no sense.
And the most frustrating part of all this? The "blood pressure monitor symbols" query had been appearing in our logs for at least three quarters before I noticed it. Three quarters of people asking the same question, getting a half-answer from a help article that only addressed phones and laptops.
Where This Approach Falls Short
I'm not going to pretend this works for every company. If your search volume is low—say, under a few thousand queries a month—the data won't be reliable enough to spot patterns. You'll be chasing noise and wasting time on phantom problems.
And if you're in a B2B context where customers deal directly with vendor support instead of researching their questions online, your search logs won't show the same signal. The method relies on having enough self-service users who turn to search before they call support.
If you're a startup or a niche supplier, your mileage will vary. Start with the help desk tickets first—those are more actionable at lower volume. Use search logs to confirm a pattern, not to discover one from scratch.
But if you've ever wondered where the gap between "technically correct" and "actually usable" lives, search logs are a solid place to start. Just be ready to find things you weren't expecting—and check your device interface, not just your marketing pages.