Same budget, 2.2× the leads — rebuilding a multi-location Google Ads account from the tracking up.

The starting point: a dashboard nobody could trust

The first problem wasn’t in the campaigns. It was in the conversion setup. The account was partly counting page views as conversions, while some real enquiries weren’t being tracked at all. So the reported numbers were inflated and incomplete at the same time, and none of them carried a value. Smart Bidding was being trained on a signal that mixed people who had glanced at a page with people who wanted to book a course.

The structure had the same problem. Campaigns had piled up over the years, brand and non-brand traffic ran through the same campaigns, and many ads in the older campaigns had been disapproved. Two campaigns stood out once we took them apart:

• A national, location-agnostic campaign spent roughly a quarter of the year’s budget at about €570 per conversion.
• A single city campaign spent another quarter at about €300 per conversion.

Together they took about half the budget and delivered about 14% of the conversions. That’s using the old, flattering definition of a conversion. Against real enquiries they did even worse.

Fix the measurement before touching the bids

Optimising an account against a broken signal only gets you better at the wrong thing. So the first job was to rebuild tracking around the actions that actually mean business:

• Server-side tagging, so conversions survive ad blockers, browser restrictions and consent gaps far better than a browser-only setup.
• Real conversions only. Form submissions and other genuine enquiry actions count. Page views don’t.
• Conversion values on every lead, so the bidding algorithm can tell a valuable enquiry from a cheap one, not just count them.

For the first time, the account knew what a lead was worth. Between March and July it recorded around €13,000 in conversion value. The year before, the account had tracked practically none.

Rebuild the structure around locations and intent

In February we rebuilt the account around how people actually search for a local course provider: by location, and by whether they already know the brand.

• One brand and one generic campaign per location, so budget, bids and ad copy could follow local demand instead of being averaged across three cities.
• Retired the dead weight. The national catch-all campaign, dynamic search ads, voucher campaigns and a location outside the core catchment were switched off.
• No gap in lead flow. The old campaigns kept running alongside on Maximise Clicks until April, while the new structure built up enough conversion history to bid properly.
• Protected peak season. In June and July we added Target Impression Share campaigns to hold visibility on the searches that mattered most when demand was highest.

The results: March to July 2026 vs. the same months in 2025

We compare March to July because that’s the period the new structure was fully live, and it covers the same season in both years. Spend was almost identical: €10,272 against €10,552.

• 250 leads, up from 115 — 2.2× as many at the same budget.
• Cost per lead down 55%, from €92 to €41.
• Conversion rate up from 1.1% to 2.9%.
• Non-brand leads up 114%, at around €53 each instead of €118. The growth didn’t come from people who were already searching for the brand.
• Best month: May 2026, at €31 per lead and a 3.2% conversion rate, against €67 a year earlier.