It usually happens when a campaign hits a plateau. The metrics are stable, the leads are consistent, but the growth has stalled. In the world of Google Search Ads, this is the moment the algorithm starts whispering its favourite suggestion: "Switch to Broad Match to reach more people."
For most specialist service providers, this suggestion is the digital equivalent of opening Pandora’s box.
I recently sat down with a sleep consultant based in the UK who was facing this exact crossroads. Her campaign was a model of precision. We were using Exact Match keywords exclusively, targeting parents specifically looking for professional intervention. The results were excellent, but we were "Limited by Budget" and, more importantly, limited by the sheer volume of people typing in those exact, high-intent phrases.
We had the capacity to scale, but I was hesitant. In the delicate niche of child sleep consulting, the line between a parent ready to hire a professional and one looking for a free blog post on "how to get a baby to sleep" is razor-thin.
The Fear of the Informational Floodgates
My caution stemmed from years of watching Broad Match go rogue in the service sector. Broad Match is designed for reach; it looks at the intent behind a search rather than just the words. While this sounds sophisticated, it often lacks the nuance required for high-ticket services.
The risk was clear: if we switched to Broad Match, Google might decide that someone searching for "free sleep training tips" or "why is my baby crying at night" is "close enough" to someone searching for a "sleep consultant UK".
In the former scenario, you pay for a click that has zero chance of converting into a client. You become a public library, paying Google for the privilege of educating people who have no intention of opening their wallets.
The Scientific Solution: The Google Ads Experiment
Instead of making a "blind" switch based on a gut feeling, we opted for surgical precision. We ran a controlled experiment.
In Google Ads, an experiment allows you to split your traffic. We took the existing, proven campaign (the Control arm) and ran it alongside a trial version where the top-performing keywords were switched to Broad Match (the Treatment arm).
This is essentially an A/B test for your budget. For 50% of the searches, the old rules applied. For the other 50%, the algorithm was given more freedom.
Understanding Statistical Significance
When you run these tests, you cannot just look at the numbers after three days and make a call. You are looking for statistical significance.
Statistical significance is the mathematical way of proving that the results aren't just a fluke or a lucky Tuesday. It tells us that if we ran this test a hundred times, we would get the same result. Google’s system monitors the data until it reaches a confidence level—usually 95%—that the Treatment arm is genuinely performing differently than the Control.
Without this confidence, applying a change is just gambling with your marketing budget.
The Verdict: When the Machine Wins
After weeks of monitoring, the data for our sleep consultant delivered a surprising verdict. The Treatment arm, which utilised Broad Match, outperformed the rigid Control arm in nearly every meaningful category.
The results were clear:
- Conversions: Up by 5.1%.
- Cost per Conversion (CPA): Down by 4.6%.
- Clicks: Increased by a staggering 86.3%.
While the total cost of the campaign rose slightly by 0.2%, the sheer efficiency gained in the Treatment arm was undeniable. The experiment proved that in this specific instance, Google’s AI was able to find additional high-intent users that our rigid Exact Match keywords were missing.
We weren't just getting more traffic; we were getting more of the right traffic at a lower price point. Because we had a solid foundation of negative keywords already in place, the "informational" searches I feared were largely filtered out.
The Takeaway for Service Providers
Does this mean Broad Match is the new gold standard? Not necessarily.
We only earned the right to use Broad Match because we started with the discipline of Exact Match. We had months of conversion data that taught the algorithm exactly what a "lead" looked like. If we had started with Broad Match on day one, we likely would have burned the budget on definitions and DIY seekers. I have seen lots of occasions where it exactly did that.
Scaling isn't about flipping a switch; it's about running the experiment, respecting the statistics, and only handing over the keys to the AI when it has proven it can drive as well as you do.
Book a consulting session with me on Fiverr or send me a message on LinkedIn if you want to test your hypothesis!