Call volume tripled in 45 days.
Call volume grew 3× while maintaining cost per call within the $6–$12 range. The account transformation was complete within 45 days of rebuild. Agent utilisation went from 60% to near-capacity. The negative keyword framework eliminated 40% of previous spend that had been flowing to non-calling queries — that budget was reallocated to the highest-intent booking terms and drove the volume increase.
Optimised for the wrong event
A call-center booking model run like an e-commerce site — tracking clicks with no link to call volume.
Rebuilt around one event
Every campaign, bid and ad rebuilt to generate booking-intent phone calls, with hard cost-per-call thresholds.
3× calls at $6–$12
Call volume tripled within cost, agent utilisation near capacity, 40% of wasted spend reclaimed.
01 The problem
A USA-based OTA operates a flight booking business built entirely on call center conversions. Unlike typical travel platforms that close transactions online, this business's revenue model depends on qualified phone calls reaching agents who complete bookings manually. The objective was never website visits or form fills — it was phone calls, at controlled cost, from people with genuine intent to book. Previous campaigns had been built and optimised as if the business were a standard e-commerce travel site — tracking clicks and generic conversion events with no correlation to actual call volume or booking rate. Budget was flowing to queries with no phone intent, agents were under-utilised, and the cost per qualifying call was far above what the business model could sustain.
One objective: the phone.
An account optimised for the wrong conversion will never perform — the architecture had to change first.
Rebuilt around calls
Every campaign, ad group, bid strategy and ad rebuilt to generate booking-intent phone calls — removing everything optimised for any other outcome.
High-intent Search
Large-scale Search targeting booking queries where the user's next action is a call to complete the booking.
Hard cost-per-call floors
Call-specific conversion tracking with hard cost thresholds, keeping every call inside the $6–$12 band.
Negative-keyword framework
Continuously expanded negatives eliminated informational and research-phase queries — 40% of old spend reclaimed.
From clicks to calls.
03 The results
Call volume grew 3× while maintaining cost per call within the $6–$12 range. The account transformation was complete within 45 days of rebuild. Agent utilisation went from 60% to near-capacity. The negative keyword framework eliminated 40% of previous spend that had been flowing to non-calling queries — that budget was reallocated to the highest-intent booking terms and drove the volume increase.
Forty-five days to triple.
Re-architecture → Search → call floors → reclaim.
Re-architecture
Rebuilt the entire account around a single conversion event: booking-intent phone calls.
Search live
Launched large-scale Search on high-intent booking queries built to drive calls.
Call-cost floors
Implemented call-specific tracking and hard cost-per-call thresholds.
Negative framework
Built and expanded negatives, reclaiming 40% of spend from non-calling queries.
3× volume
Calls tripled within the $6–$12 band; agents moved from 60% to near-capacity.
What the OTA kept.
Documented, portable, and owned by the client not locked inside our account.
- ✓Calls-only account structureEvery campaign built for the phone.
- ✓Call conversion trackingCall-specific measurement and cost floors.
- ✓High-intent keyword setBooking queries that drive calls.
- ✓Negative-keyword frameworkThe list that removed non-calling spend.
- ✓Cost-per-call thresholdsThe rules that hold $6–$12.
- ✓Agent-capacity modelHow volume was matched to staffing.
04 Why it worked
Most campaign rebuilds improve efficiency or volume — rarely both simultaneously. The insight here was structural: an account optimised for the wrong conversion type will never perform, no matter how well the ads are written or bids are set. The entire architecture had to change before any metric improvement was possible. Once rebuilt around calls, every optimisation lever worked in the right direction.
Everyone before us optimised our account like a website. The moment it was rebuilt around calls, every lever finally pulled in the right direction.