Candela sells aesthetic laser and energy devices to dermatologists, plastic surgeons and med spas, where one lead can become a five-figure device sale. We rebuilt their North American search account. Two weeks later, the same spend was buying more than twice the leads.
The account overview for the two weeks before the rebuild went live, against the first two weeks after.
Read-only access, no commitment. You keep the findings either way.
The account was returning about 34 form-fill leads a week, at $876 each. Before touching a single bid, we went through every keyword, every negative and every automated change made since January. Four problems stood out.
Auto-applied recommendations had added 126 broad match keywords that spent about $9,450 for three leads, stripped protective negatives and removed a converting brand keyword.
Bare device names like Vbeam, PicoWay and Nordlys sat in the brand campaign on an uncapped impression share strategy. They had spent $47,000 since January, at up to $7,411 per lead, and landed on a generic page.
Shared negative lists were blocking the client's own product names. Nearly 700 Candela-branded negatives sat in a single campaign, and one campaign's best historical keyword, 86 leads at $163 each, could not serve.
Key product campaigns ran Maximize Conversions with no target, so the algorithm chased volume at any price. The hair removal line was running at $2,609 per lead.
Smart Bidding is only as good as the signal it learns from, so measurement came first. Then we stopped the account changing itself, took out the spend with no return, and rebuilt the structure so every product searcher met the right ad and the right page.
Moved lead tracking onto a server-side setup and switched on enhanced conversions for leads, so every form fill reaches Google Ads with first-party data attached and survives browser and cookie restrictions. We also built the bridge to Candela's CRM for offline conversions, so bidding can learn which leads go on to become device sales, not just which ones fill in a form. Every new campaign launched with Candela's own UTM convention, so each lead lands in their CRM tagged with the campaign and keyword that produced it.
Turned off auto-applied bid and keyword recommendations and restored the negatives they had stripped, so the account only changes when we change it.
Paused more than 130 keywords with no return across a full year of data, and brought back historical winners as phrase match after checking each one against its landing page.
Moved product shoppers out of the brand campaign into three dedicated campaigns pointing at device-specific pages, with fresh keywords, cross-negatives and a two-angle ad test: clinical authority against the practice owner's fear of losing patients.
Removed self-blocking negatives across Glacē, Matrix and the brand campaign, and routed high-converting branded terms like "candela medical" and "candela laser hair removal" to the campaigns built for them.
Buyers searching "Glacē vs HydraFacial" were seeing ads that never mentioned HydraFacial. We named the comparison in the copy and moved all Glacē traffic to the landing page converting at $554 per lead instead of $982.
Added target CPAs to the uncapped campaigns and raised budget only where efficiency was already proven.
None of this was a one-off clean-up. In the first three weeks, the account change history logged:
Measured against the 13-week average before the rebuild, not only the two weeks before it, the first two weeks after still delivered nearly twice the leads at 45% lower cost.
Clicks cost the same. Nearly twice as many of them turned into leads.
Average cost per click barely moved, $8.80 before and $8.86 after. The gain came from sending the budget to searches and pages that convert. The brand campaign spent 22% less per week and delivered 49% more leads once product shoppers had their own campaigns, and every product line got cheaper:
| Campaign | Leads / week before | Leads / week after | Cost per lead before | Cost per lead after |
|---|---|---|---|---|
| Brand | 16.4 | 24.5 | $877 | $459 |
| Glacē | 4.6 | 11.7 | $884 | $382 |
| Gentle (hair removal) | 2.4 | 7.0 | $2,682 | $785 |
| Better Than Diode Laser | 2.0 | 3.5 | $524 | $347 |
| Matrix (RF microneedling) | 1.7 | 2.3 | $1,114 | $241 |
| Aesthetic portfolio | 6.6 | 6.9 | $247 | $187 |
| Vbeam, PicoWay, Nordlys (new) | 0 | 9.1 | n/a | $773 |
| Account | 33.8 | 65.0 | $876 | $481 |
The three new device campaigns added nine leads a week that the old structure was not capturing at a sensible cost.
Read access and forty-five minutes. You keep the findings either way.
Book a free audit