Some advertisers swear by them.
Others avoid them entirely.
The truth? Broad match can be incredibly powerful — if tested correctly.
Instead of blindly switching campaigns to broad match, one of the smartest ways to evaluate performance is by running a 50/50 experiment inside Google Ads. This allows you to split traffic evenly between:
- Your original keyword setup (exact/phrase)
- A version of the campaign that adds broad match keywords
In this article, we’ll walk through real results from six different accounts that ran broad match experiments — including wins, losses, and inconclusive outcomes.
You’ll see exactly when broad match worked… and when it didn’t.
How 50/50 Broad Match Experiments Work
Inside Google Ads, you can:
- Go to All Campaigns
- Click Experiments
- Create a campaign experiment
- Split traffic 50/50 between:
– Base campaign (your existing keywords)
– Trial campaign (adds broad match versions)
NOTE: Use recommendations as shown in the video above to quickly create a 50/50 broad match experiment.
This setup allows you to measure:
- Cost per conversion
- Total conversions
- Conversion value
- Return on ad spend (ROAS)
The beauty of experiments?
You’re not guessing. You’re measuring.
Important: Not Every Account Is Ready for Broad Match
Broad match tends to perform better in mature accounts that have:
- Strong conversion tracking
- Consistent volume
- Smart bidding enabled
- Clear historical data
If you try this on a brand-new account with limited data, results may be unpredictable.
Now let’s look at the real experiments.
Account #1: Photographer (Florida)
This was a relatively low-spend account.
- Total clicks across test: ~800
- 30-day experiment
- Mature campaign structure
Results:
- Cost per conversion ↓ 29%
- Conversions ↑ 42%
- Cost per lead: $18 → $13
Google labeled this “inconclusive” due to statistical thresholds. But practically speaking?
- Lower cost per lead
- Significantly higher volume
For a small local service like photography, that’s meaningful.
Decision: Applied broad match.
Even if sample size was modest, the directional trend was strong enough to justify it.
Account #2: Emergency Center (Texas)
This account had stronger volume but showed more modest results.
Results:
- Cost per conversion ↓ 3%
- Conversions ↑ 8%
Technically positive — but not dramatically.
When using broad match, you’re usually accepting:
- More reach
- Potentially broader search intent
If the lift is only single digits, you must ask:
- Is quality declining?
- Are we trading precision for noise?
In high-stakes industries like emergency medical services, quality matters more than volume.
Decision: Did not apply.
Sometimes “slightly better” isn’t good enough.
Account #3: Garbage Pickup & Dumpster Service (Texas)
This one was clear.
- ~1,400 clicks total
- Larger sample size
- Strong statistical confidence
Results:
- Cost per conversion ↓ 23%
- Conversions ↑ 30%
- Cost per lead: $32 → $25
- Conversions: 133 → 173
This is the dream outcome:
- More leads
- Lower cost
- Statistically significant
Dumpster rental is a service with wide keyword variations — broad match likely uncovered high-intent queries not originally targeted.
Decision: Applied immediately.
Account #4: Floor Tile Cleaner (Texas)
This experiment was more complex.
Results:
- Conversions ↑ 130% (33 → 76)
- Cost per conversion ↑ 21%
More than doubled conversion volume — but at a higher cost.
This introduces a strategic question:
Would you pay more per lead to get double the volume?
Sometimes the answer is yes.
Especially if:
- Lead quality is strong
- Closing rate supports the increase
- Client wants growth
But this is where qualitative analysis matters.
Broad match may bring:
- Lower intent searches
- Price shoppers
- Broader informational queries
If quality drops, volume alone isn’t enough.
Decision: Paused for deeper evaluation.
This is where communication with the client becomes critical.
Account #5: Bridal Shop (Wisconsin)
This account tracked conversion value, not just lead count.
They measured:
- Appointment clicks
- Phone calls
- Relative value assigned to each
So instead of just cost per conversion, we evaluated:
- Conversion value over cost (ROAS)
- Total conversion value
Results:
- ROAS ↓ 60%
- Total conversion value ↓ 58%
Even though Google labeled results “inconclusive,” the directional trend was clearly negative.
Broad match likely expanded into:
- Lower purchase intent
- Research-phase searches
- Non-local traffic
In retail niches like bridal, intent is everything.
Decision: Did not apply.
Broad match failed this account.
Account #6: Fertility Center (Midwest)
This was the standout winner.
This account also tracked conversion value.
Results:
- Conversion value ↑ 175%
- ROAS ↑ 178%
- Total conversions: 37 → 78
- Cost per lead: $17 → $50
Yes — cost per lead increased.
But conversion value increased dramatically more.
That’s key.
When tracking value properly, broad match may:
- Attract higher-value leads
- Capture new high-intent searches
- Expand into overlooked queries
Even though Google labeled results inconclusive statistically, practically:
- More conversions
- More value
- Stronger ROAS
Decision: Applied.
Why Broad Match Performs Differently Across Industries
The results above show something important:
Broad match is not universally good or bad.
It depends on:
1. Industry Search Behavior
Dumpster rental and fertility clinics have broader search variation.
Bridal retail may have more defined intent signals.
2. Account Maturity
Smart bidding needs data to guide broad match properly.
3. Conversion Tracking Depth
Accounts tracking conversion value tend to get smarter optimization.
4. Budget Size
Low budgets may not gather enough data for broad match to stabilize.
The Real Advantage of 50/50 Experiments
Instead of debating broad match philosophically, experiments give you:
- Objective comparison
- Equal traffic distribution
- Real performance metrics
- Controlled risk
You don’t destroy your base campaign.
You test intelligently.
This is far better than:
- Turning everything to broad match overnight
- Relying on gut instinct
- Following Google’s recommendations blindly
When Broad Match Tends to Work Best
Based on real experiments, broad match performs strongest when:
- Smart bidding is active (Max Conversions or tCPA)
- Conversion tracking is accurate
- Search volume is moderate to high
- Negative keywords are well managed
- Industry search terms are diverse
It struggles when:
- Account data is thin
- Conversion quality varies widely
- Intent is very narrow and specific
- Budget is small
The “Inconclusive” Label Explained
Google uses statistical confidence intervals to determine significance.
Often:
- Smaller accounts = “inconclusive”
- Short test windows = “inconclusive”
But advertisers must apply practical judgment.
If:
- Cost per conversion drops significantly
- Conversions increase meaningfully
You may not need 99% statistical confidence to act.
Real-world performance matters.
The Risk: Quality Over Quantity
One danger of broad match:
It can increase volume while quietly lowering quality.
Metrics may show:
- More leads
- Similar cost
But if:
- Close rates decline
- Lead intent weakens
You’re actually losing efficiency.
That’s why:
- Listen to sales teams
- Track revenue when possible
- Monitor ROAS, not just cost per lead
Broad match isn’t just about traffic — it’s about profitable traffic.
Key Takeaways From 6 Real Accounts
Here’s the pattern:
| Industry | Outcome |
| Photographer | Strong win |
| Emergency Center | Slight lift (not enough) |
| Dumpster Service | Clear win |
| Tile Cleaner | Mixed (volume ↑, cost ↑) |
| Bridal Shop | Loss |
| Fertility Center | Major win |
Broad match succeeded in 3 out of 6 clearly.
That’s not universal dominance — but it’s powerful when it works.
Should You Test Broad Match?
If your account is mature and stable?
Yes.
But do it strategically:
- Use experiments.
- Split 50/50.
- Run at least 30 days.
- Evaluate both cost and quality.
- Don’t auto-apply broad match recommendations.