Dots scattered across the screen. No clear path. No obvious next step.
This is 30 days of real user taps on the Bianco's homepage. Now imagine it's 6pm, the kids are hungry, you've decided on pizza. You land here. Every second of confusion is a lost order.
Strong product. Loyal customers. A website in the way.
Bianco's is an independent, woman-owned thin-crust pizzeria in Walpole, MA. They run their own delivery drivers — a deliberate choice to stay off Uber Eats and DoorDash and protect their staff. The product is excellent. The local reputation is strong.
The problem wasn't awareness. Most orders still came in by phone, creating operational load and removing visibility into customer behaviour. The ask: find out what's blocking online ordering, reduce the friction, shift the balance.
75% mobile. 65% arriving via Google search — the referrer data from Hotjar's traffic channel report. Search traffic indicates active intent: someone looking for a place to order, not browsing passively. The site needs to confirm the decision and get out of the way.
The ordering button ranked third on its own homepage.
One month of behavioural data, August 2025, 100% session capture via Hotjar and Clarity. The top-line numbers looked fine. The click data told a different story.
The homepage had two buttons side by side: OUR MENU and ORDER NOW. OUR MENU got nearly twice the taps — 19% of all site-wide clicks versus 10% for ORDER NOW. Then there was the mobile hamburger: 27% of all taps. Open it, and the first option is the food menu. A two-tap route to the same dead end.
The hero image filled 100% of the mobile viewport with no visible scroll affordance. Users tapping the logo or background weren't lost — they were trying to activate the page, expecting a carousel or more content to appear. A full-bleed hero that hides everything below it is a design failure, not a user failure.
Clarity session recordings confirmed the pattern. Users tapped OUR MENU, landed on a wall of text with no photos and no way to order, scrolled, got stuck, and navigated back. The phone number was static — users tapping it expecting a dialer got nothing. Contact info sat below a fold most sessions never reached.
Two additional friction points compounded this. A chat-style bubble in the corner — styled to look like a messaging widget — offered “Call” and “Order” on tap. It received 28 interactions, almost all from confused users. A cookie consent banner appeared on every desktop session. Both have since been removed.
Three participants. One thing none of them were asked about.
Unmoderated study via Hotjar. Two scenarios: browse naturally, then attempt a specific order. No prompts on food photography. No leading questions.
All three participants mentioned missing food photography without being prompted. Three strangers, one session each, same observation. When that happens it stops being an assumption. It's a finding.
Shosh also found a bug on Speedline — the third-party ordering platform — where switching from delivery to pickup erased all entered address data. Outside direct control, but documented: friction at the handoff costs orders the website already earned.
Same homepage. Same first tap. Very different outcomes.
The legacy site presented a fork users didn't know was a fork. Reaching for the food menu was the natural first move — see what's available before ordering. That path was a dead end. The ordering path existed but ranked third. Mobile perspective, where 75% of sessions happened.
Users weren't avoiding ordering. They wanted to see the food first — a natural step before committing. The site treated that instinct as a detour. The fix wasn't removing the food menu path. It was making both paths lead to the same place: Speedline, which functions as both a browsable menu and an ordering system simultaneously.
Every decision traceable. Nothing cosmetic.
Same Tilda CMS, no platform rebuild. Each change is grounded in a specific finding.
| What changed | Why |
|---|---|
| Two equal CTAs → single “Start Order” | OUR MENU got twice the taps of ORDER NOW. Eliminating the split removes the dead-end path. “Start Order” over “Order Now” — implies browsing is welcome, not that you must commit immediately. |
| Phone number → click-to-call | Clarity recordings confirmed users tapping static phone text expecting a native dialer. Nothing happened. |
| Anonymous pizza grid → named, clickable specials | All three study participants independently cited missing food photography. Named items give first-time visitors a visual anchor before clicking through. |
| Business info → above the fold | Location, phone, and delivery area were hidden below a full-viewport hero with no scroll affordance. Moved above it. |
| CTA repeated mid-page | Scroll depth data showed 50%+ of sessions reaching mid-page. The CTA needs to be there to meet them. |
| Cookie banner removed | No legal requirement for a US-only local audience. Removed friction and cleaned up data capture. |
| Chat-style bubble removed | 28 taps, almost all from confused users. Engagement driven by wrong affordance is not positive engagement. |
The signals that matter. And the one still missing.
January 6 to March 4, 2026. Hotjar on free tier — 35% sampling. Session counts are estimates; ratios and behavioural patterns are valid. Clarity running at 100% capture throughout.
| Metric | Aug 2025 | Jan–Mar 2026 | |
|---|---|---|---|
| Returning sessions | 13.1% | 29.6% | ↑ Doubled |
| Session duration | 5:45 | 6:03 | ↑ Positive |
| Homepage scroll depth | 49.8% | 53.4% | ↑ More users reaching mid-page CTA |
| Rage clicks | 2 | 0 | ↑ Friction resolved |
| Bounce rate | 27.3% | 41.7% | Measurement artifact — see below |
On the legacy site, tapping OUR MENU created a second pageview — analytics recorded that as engagement, artificially lowering the bounce rate. After the redesign, a user who lands, taps Start Order, and goes straight to Speedline is a single-page visit. It reads as a bounce even when the intended action was completed. Clarity recordings validate this: sessions showing users reaching Speedline are false negatives in the bounce count, not failures.
The “Menu” navigation item is still the #1 clicked element post-redesign at 622 sessions — and still routes to the static text page. Redirecting it to Speedline is the next action. Until it ships, a significant share of engaged sessions is hitting a dead end.
The structural constraint throughout: Hotjar and Clarity stop recording the moment a user leaves for Speedline. Whether a visit became an order is invisible. Every metric above is a proxy. A UTM passthrough inquiry to Speedline is pending — if supported, it would allow website-originated sessions to be identified within Speedline's reporting.
Friction removed. Foundation to build on.
Immediate: redirect the food menu to Speedline, track Start Order as a conversion event, establish a 30-day clean data window. Then: CTA language rotation, a user test of the current experience, and a formal site map.
Longer term, there's a brand story that hasn't been told yet. Long-tenured kitchen staff. Their own delivery drivers. A deliberate choice not to join Uber Eats — made to protect livelihoods, not just margins. A content strategy rooted in that story, paired with SEO work and a marketing calendar, is the difference between a fixed website and a growing business.
The reasoning is the work.
The most important calls in this project weren't design decisions. They were interpretation calls — what the scattered tap data meant, why the bounce rate increase was a good sign, why the food menu needed to be understood as user intent before it could be treated as a design problem.
Working within a free-tier toolset, a third-party platform that couldn't be modified, and incomplete conversion data: that's what this engagement actually looked like. Navigating that clearly is the skill.