Beck Glass
SEM relaunch · CRO audit · 4-tier web ops
Five-campaign SEM rebuild, three-page CRO audit, and a new squeeze page. 27 conversions inside Week 2 of relaunch.
Objective: relaunch demand capture across Beck Glass and Beck Auto Glass with a measurable SEM motion, a diagnosed website, and a managed operating cadence, consolidating campaign ops, CRO, and reporting into a single accountable layer.
The engine as sources, processors, sinks.
Each node is something this engagement actually did. Signal in on the left (sources feeding processors, processors producing sinks), outcome on the right.
What we walked into.
Beck Glass operates auto and residential glass-services properties across Vancouver and the GVRD, including ICBC-billed windshield work, a category with sharp local intent and clear urgency signals. Demand capture and web maintenance were running as separate efforts: no shared rhythm, paid spend not converting (paid social ~96% of paid sessions, zero recorded contacts in 30 days), the homepage absorbing ~85% of pageviews but converting at 0.13%, and broken /services/... URLs leaking targeted geo traffic.
The brief was to consolidate SEM, CRO, and web ops into one operating model (diagnose, rebuild, run) rather than a string of one-off project engagements.
The shipped work.
SEM relaunch across Beck Glass and Beck Auto Glass. Five-campaign architecture (Generic Repair, Brand, Competitor Conquest, Retargeting, and Local / Google Maps) weighted ~60% to core repair services, with growth bets on ADAS Calibration and defensive plays on brand and competitor terms. Manage spend up to $20K/mo with weekly optimisation, conversion-led bid strategies, and tight negative-keyword hygiene.
Three-page CRO audit benchmarked against Unbounce and the Ascend2 SMB Landing Page Optimisation Report (264 SMB marketers, 2026). Home page scored 42/100, services page 38/100, and we shipped a new windshield-replacement squeeze page at 76/100: benefit-first headline, dual hero CTA, 5.0★ Google rating above the fold, sticky mobile CTA bar, ICBC differentiator, and a no-nav layout built for paid traffic.
Four-tier website support model (Bronze → Platinum, $7,500 → $22,500/mo) packaging campaign ops, CRO iteration, and runbook-driven support into one recurring cadence.
Call attribution wired end-to-end. Google Ads → HubSpot via RingCentral so phone-call conversions flow into the CRM and campaigns optimise against real sales outcomes, not form fills. HubSpot and GA4 channel labels reconciled so reporting tells one story. Weekly update cadence covers best performers, optimisations implemented, and what to watch next.
What it produced.
Week 2 of SEM relaunch: 27 conversions across the account as campaigns exit the learning phase, primarily direct calls to the business. Generic||Repair leading: 173 clicks, 11 conversions. PMAX||Local trailing second: 93 clicks, 6 conversions.
Channel diagnostic settled by data. Paid search converting at 1.45% session-to-contact in HubSpot vs paid social at 0% recorded contacts from 4,652 sessions, making the case to reallocate spend from paid social into search and the new squeeze page.
New squeeze page scoring 76/100 on CRO criteria (Above the Fold 85, Message Match 82, CTA & Form 78), built to receive paid traffic on a no-nav, dual-CTA layout with multi-layer trust signals (ICBC + 47yrs + OEM + 5.0★).
Operating cadence live across both properties: weekly campaign tuning, fortnightly check-ins, monthly performance review, quarterly retro. One accountable layer, three reporting rhythms.
The honest read.
Diagnose, then rebuild. Standing up new SEM into a website converting at 0.13% on the homepage would have wasted spend. The CRO audit and squeeze-page rebuild ran in parallel with the campaign relaunch, so paid traffic landed on a page built for it, not a homepage built for everyone.
Paid search vs paid social, settled by data. Paid social was 96% of paid sessions and 0% of recorded contacts; paid search was 4% of sessions and 100% of paid contacts. The mix correction is now an operating decision, not a debate.
Call attribution as table stakes. A glass-services business converts on phone calls. Wiring RingCentral → HubSpot meant the win column finally matched the actual sales motion, and campaigns optimise against real outcomes, not form fills.
Four tiers gave structure to both sides. The Bronze → Platinum model gave the engagement clear scope boundaries from day one: no ambiguity about what's included, what triggers an upgrade, and what sits outside.
Weekly cadence kept the engine honest. A short weekly update (best performers, optimisations implemented, what to watch next) meant the loop between SEM signal and web ops never drifted.
The tools wired into this engine.
Vendor list reads flat. The configuration is what makes it operate.
Want one of these for your engine?
Each of these started as a single audit or a scoped setup sprint, not a blank retainer. Book a call and we will map the smallest useful step.