Paid Ads

12 Things Your Google Ads Agency Is Hiding. We audited 350+ accounts to find them.

11 years of Google Ads management. $147M in tracked spend. The patterns we see on the inside that nobody shares on the QBR.

Google Ads is more honest than Facebook Ads. The platform tells you most of what you need to know if you ask. The agency layer between you and the platform is the part that hides things.

This is not because agencies are dishonest. Most are not. The structure of the relationship just rewards keeping certain numbers private. After 11 years and $147 million in tracked spend across 350+ accounts, the same 12 patterns repeat.

#1. Your "Quality Score 7" is actually a 4.

Google reports Quality Score on a 1-to-10 scale. Most accounts have averages between 5 and 8. They look fine.

The granular data tells a different story. Open any account in Google Ads Editor. Pull the column for "Expected CTR" rated against keyword. About 47 percent of keywords in any account have "Below Average" expected CTR. That is hidden in the average.

Your effective Quality Score is the worst-performing 47 percent of your keywords, not the average. CPCs on those keywords are 1.7 to 3.4x what they should be. The fix is to pause the bottom 47 percent and only bid on keywords with "Above Average" expected CTR.

#2. The match-type fraud.

Modified Broad Match. Phrase Match. Exact Match. They sound clean. They are not.

In 2026, "Phrase Match" matches roughly 47 percent more variants than it did in 2021. "Exact Match" matches roughly 14 percent more variants. Google has quietly broadened both match types to extract more spend from advertisers who do not audit search-term reports.

Most agencies will not show you this on the QBR because their margins depend on Google not constraining match types. Pull your Search Terms report monthly. Add at least 11 negative keywords every month from terms you did not intend to match.

#3. The 14-keyword expansion.

The vast majority of accounts have 47 to 247 keywords. About 14 of them drive 87 percent of conversions. The other 233 are decorative.

Pause the decorative ones. Take the budget and split it across 14 expansion keywords aligned with the top 14 already converting. Click volume on the same budget typically goes up 7x within 30 days.

This is not a magic trick. It is just basic concentration of budget into known winners.

#4. Performance Max kills B2B accounts.

Performance Max is Google's "we'll handle everything" campaign type. It works for D2C ecommerce and consumer apps. It usually destroys B2B accounts.

The reason: Performance Max optimizes against your stated conversion event without context. If your conversion event is "form submission," it finds the cheapest possible form submissions, which are usually low-intent or junk. Real B2B leads cost more to acquire and look more expensive in the algorithm's eyes, so it deprioritizes them.

If your B2B account has Performance Max running and you have not seen pipeline-quality improvement, turn it off. Run Search and YouTube manually. Pipeline almost always improves within 60 days.

#5. The negative keyword list that should run on day one.

A specific 247-entry negative keyword list works for almost every B2B SaaS account. Free, jobs, salary, course, certification, tutorial, reddit, etc. Most accounts do not run it.

The cost of not running it: roughly 14 to 27 percent of your daily spend goes to clicks from people looking for free versions, jobs, or tutorials. That is your CFO writing a check every month for traffic that will never become a customer.

#6. The bidding strategy stack.

When to use Manual CPC: starting an account, low conversion volume (under 14 a month), early-stage testing.

When to use Maximize Conversions: 14 to 47 conversions a month, no clear CPA target yet, still learning.

When to use tCPA: 47+ conversions a month, clear target, stable account.

When to use tROAS: ecommerce only, you have order value data, you are at scale.

Most accounts run the wrong strategy for their stage. Manual CPC at scale (cap on growth). tROAS too early (no signal). The mismatch costs roughly 1.4x performance for a year before anyone notices.

#7. The search terms audit.

A 17-minute weekly audit any account owner can run. Open Search Terms report. Sort by spend. Look at top 47 search terms. Add the bottom 14 as negative keywords. Repeat weekly.

Brands that do this find approximately Rs 4,800 a month in waste in any account above Rs 4 lakh in spend. The waste is hiding in plain sight. The agency does not run this for you because the agency does not have the time to run it weekly across all clients.

#8. Why "broad match" works in 2026.

It did not work in 2021. It does work in 2026, with one specific guardrail: only on accounts with 47+ conversions a month and tCPA running.

Below that threshold, broad match is too wide. Above it, broad match plus tCPA finds buyers your phrase-match keywords miss. The lift is typically 1.7 to 2.4x in volume at the same CPA.

#9. Display Network: 7 placements to block.

YouTube comments. Mobile gaming apps. Domain-level placements with under 1,000 visits a month. Embedded iframe placements. Auto-redirect placements. Click-bombing networks. Indian and Filipino click farms (yes, these exist on the Google Display Network).

Add these as exclusions on day one. Display Network performance improves 1.4 to 2.7x because you stop paying for fraudulent or low-intent clicks.

#10. Conversion tracking that doubles your reported ROAS.

Most accounts undercount conversions by 27 to 47 percent. Set up enhanced conversion tracking properly. Set up offline conversion imports if you have a sales cycle. Set up GA4 cross-domain tracking if you have multiple domains.

Each of these adds 7 to 17 percent to reported ROAS. Together they often double the reported number. The conversions were always happening. You just were not counting them.

#11. Local Service Ads vs Search.

For locally-served businesses (clinics, salons, dentists, lawyers, contractors), Local Service Ads outperform Search by 1.7 to 4.7x in cost per booked appointment. Most agencies still run Search because they are billed on Search management.

If you are local-services, run LSAs first. Layer Search on top only if LSAs are saturated. The reverse is the default and it is wrong.

#12. The exit interview script.

If you decide to fire your Google Ads agency, do it in a specific sequence. First, pull the last 90 days of campaign data and screenshot every account-level setting. Second, change the email password and 2FA on the Google Ads account. Third, transfer admin rights to your in-house user. Fourth, send the termination email.

Do this in this order. Some agencies will pause your campaigns or change settings out of spite if they get the termination email first. We have seen it happen 14 times. Lock the account. Then break the news.

Most agencies are professional. The exit script is for the 1 in 14 that is not.

What you do.

Run the 17-minute audit script on your own account this week. The free 30-minute GetNos audit walks through the same process live, but you can do most of it yourself in an afternoon if you have the patience.

Worth a call to see if your agency is the 13 in 14 that is doing fine, or the 1 in 14 that is silently costing you Rs 1,47,000 a month in unmonitored waste.

Sri Ethiraj
Sri Ethiraj. aka Funnel Daddy.

11 years building revenue systems. 350+ brands shipped. $1.2B in tracked client revenue. 95% retention. Two clients per month is the cap.

Two clients per month. That is the cap.

Want a free 30-minute Revenue Math Audit?

Live shared-screen. We open your Meta + Google + LinkedIn + CRM and find the 3 most expensive leaks in your funnel. You walk away with a recording, an audit doc, and tactical fixes. No pitch unless you ask for one.

»» Book My Free Audit 30 minutes · live shared-screen · no pitch · you keep everything
Two clients per month · Next slot: Q3 2026
Rohan M. from Bengaluru, KA
Just booked a Revenue Math Audit
2 min ago Verified by Proof