In mid-July 2026, a surgical specialty clinic in the greater Tokyo area experienced a sudden surge in invalid clicks from specific IP addresses.
Its invalid click rate jumped from 7.5% to 43.5%, driving up its cost per acquisition (CPA) while new patient numbers declined. Manually excluding IP addresses was no longer enough to keep up. After implementing Spider AF, the clinic reduced its CPA by 59%, from ¥17,000 to ¥7,000, and increased the number of new patients visiting the clinic by 60%, from 46 to 74.
We spoke with the advertising agency specialist who manages the clinic’s campaigns about what happened and how the team addressed it.
Results at a Glance
- 59% lower CPA: ¥17,000 → ¥7,000
- 60% more new patients: 46 → 74
- 91 IP addresses automatically blocked as of August 23: More than five times the 17 addresses excluded manually
- Virtually eliminated manual access log analysis and IP exclusions
-
The Challenge: An Invalid Click Rate of 43.5% and a Manual Process That Couldn’t Keep Up
Thanks for joining us. Before implementing Spider AF, how familiar were you with ad fraud and the tools available to address it?
My understanding was fairly basic: these tools automatically identify invalid clicks and prevent them from being charged.
What specific challenges were you facing?
Starting in mid-July 2026, we saw a sudden spike in invalid clicks from particular network connections.
Invalid clicks rose from around 60 in June, representing an invalid click rate of 7.5%, to roughly 800 in July, or 43.5%. On some days, the invalid click rate exceeded 70%.
There were other clear anomalies, too. Click-through rates reached 40–50% in certain geographic areas. This led to a sharp increase in CPA—to approximately ¥17,000 in July—and a decline in new patients.
An invalid click rate of 43.5% is substantial. What did you do after spotting those anomalies?
We retrieved the server access logs, extracted and compiled the IP addresses, and manually added IP exclusions at the account level in Google Ads.
But the source kept switching IP addresses every few days. We couldn’t keep up by pulling logs and adding exclusions manually, and we were reaching the limits of what that process could handle.
Why Spider AF: No More Manual IP Exclusions
With manual measures no longer keeping up, what first caught your attention about Spider AF?
My immediate reaction was that we wouldn’t have to keep excluding IP addresses by hand.
Spider AF connects to Google Ads through its API and automatically adds detected sources of fraudulent traffic to the exclusion list. That addressed the exact problem we were facing: we couldn’t manually keep pace with someone who kept changing IP addresses and starting again.
It also looks at behavioral patterns, rather than relying solely on IP addresses. And because we could deploy the tag through Google Tag Manager, there was very little need to modify the website. That made implementation feel straightforward.
Automation sounds like a strong fit for the problem. What else mattered as you evaluated the tool?
We had three priorities, in this order:
- Detection accuracy: The ability to identify sources that keep switching IP addresses, using signals beyond the IP address itself, such as behavioral patterns.
- Workload: Automating the process from detection through exclusion using the Google Ads API. We’re a small team managing multiple clients, so continuously analyzing logs and adding exclusions manually simply wasn’t sustainable.
- Platform coverage: We run campaigns on both Google Ads and Yahoo! Ads, so we needed support for both.
What ultimately led you to move forward?
After the spike in mid-July, we tried managing the situation through access log analysis and manual IP exclusions. But each time the source changed IP addresses, we fell several days to a week behind. It became clear that this wasn’t solving the underlying problem.
We connected Spider AF to Google Ads through the API and installed the tracking tag in July. In August, we connected Yahoo! Ads and enabled audience exclusions.
The two deciding factors were the ability to automate detection and exclusion, and the ability to manage protection across both Google and Yahoo! in one place.
Knowing that it was a Japan-based solution with Japanese-language support also gave us confidence.
What the Analysis Revealed: More Sources Were Slipping Past Google’s Detection Than Expected
It sounds like the limits of manual intervention were a major factor. Once you started measuring traffic with Spider AF, did anything surprise you?
Before implementation, we knew that Google Ads was classifying more than 700 clicks per month as invalid, with an invalid click rate of 43.5%. What we couldn’t determine was how much additional traffic was going undetected by Google and still being charged.
Because Google was already flagging nearly half of the clicks as invalid, we assumed the amount slipping through would be relatively small.
But by August 23, Spider AF had automatically blocked 91 IP addresses—far more than the 17 we had been able to identify and exclude manually. We realized that more sources of fraudulent traffic were slipping past Google’s detection than we had expected.
We had excluded just 17 IP addresses manually. Spider AF automatically blocked 91—more than five times as many. There were more sources going undetected by Google than we had anticipated.
The Results: CPA Fell 59%, from ¥17,000 to ¥7,000, While New Patient Visits Increased 60%
So there was more traffic getting past Google’s detection than you expected. Once Spider AF was running, which changes stood out most?
Four things really stood out:
- We could connect to both Google Ads and Yahoo! Ads through their APIs and manage protection in one place.
- The dashboard showed detected IP addresses and repeat click counts. That gave us concrete results to share with the client, including exactly how many IP addresses had been automatically blocked.
- We virtually eliminated the work of retrieving and analyzing access logs and manually adding exclusions.
- Initially, it took two to three days from detection for blocking to take effect. That time shortened as we continued using the tool, and by late August, we could stop traffic from a new source within a day of its first appearance.
Alongside those improvements in workload, reporting, and response time, what changed in the performance metrics?
CPA fell from ¥17,000 to ¥7,000—a 59% reduction.
The number of new patients visiting the clinic, which is our key performance indicator, increased from 46 to 74—a 60% increase.
How did the clinic respond to those results?
The clinic director had personally spotted the anomalies in the dashboard’s geographic data and was understandably concerned. Being able to provide regular updates with concrete figures—such as how many IP addresses the tool had automatically blocked—helped reassure them.
By late August, we were able to report that the invalid click activity had subsided. Since September, we haven’t received any further inquiries about invalid clicks. Our meetings have returned to routine campaign improvements, such as reviewing geographic targeting and increasing the budget.
Along with the recovery in new patient numbers, the clinic’s decision to increase its advertising investment was the clearest sign of its confidence in the results.
With concrete results to report and discussions returning to campaign improvement, did the way you interpret data and make decisions change as well?
Significantly. During the spike in invalid clicks, the data feeding automated bidding was being contaminated, and our geographic and search term data no longer reflected genuine demand. After implementation removed that noise, we were able to make decisions such as:
- Resuming campaigns in areas we had temporarily excluded and redesigning our overall geographic targeting, because regional click-through rates and conversion counts were reliable again.
- Reviewing negative keywords and keyword match types, because we could analyze search term reports based on genuine demand.
We could finally see the data that mattered. Previously, we spent a lot of time trying to determine whether a number reflected invalid click activity or real demand. Now, we can use the data directly to make campaign decisions.
Who Would Benefit from Spider AF?
Being able to separate out the impact of invalid clicks and use your data with confidence sounds like a major change. Based on your experience, which businesses would benefit most from Spider AF?
First, businesses in industries with high conversion values or high costs per click, such as healthcare. Even a small number of invalid clicks can significantly affect the data used by automated bidding, so the value of putting protection in place can be especially clear.
Second, local businesses that advertise within a defined geographic area. When campaigns target a limited area, they can be particularly vulnerable to concentrated activity from competitors or other malicious sources.
I also think protection is important for advertisers using automated bidding.
And finally, agencies and consulting firms like ours, where a small team manages multiple accounts. Even when you don’t have the capacity to analyze logs and manage exclusions manually, automation allows you to maintain protection consistently.
It’s clear that your team has moved from constantly responding to invalid clicks to focusing on campaign management and improvement.
Thank you for sharing your experience!
From Responding to Invalid Clicks to Investing in Growth
This surgical specialty clinic’s experience highlights three key points:
- The challenge: The invalid click rate surged from 7.5% to 43.5%, and CPA rose to approximately ¥17,000. Manual IP exclusions couldn’t keep pace with sources that kept changing network connections.
- The solution: Spider AF connected to both Google Ads and Yahoo! Ads through their APIs, automating detection and exclusion. It automatically blocked 91 IP addresses—more than five times the 17 excluded manually.
- The results: CPA fell 59% to ¥7,000, while new patient visits increased 60%, from 46 to 74. Reducing the noise from invalid clicks enabled the team to make campaign improvements based on reliable data.
For healthcare providers with high costs per click or high conversion values, even a small number of invalid clicks can have a significant impact on campaign performance.
If you’ve noticed a sudden increase in CPA or unusually high click-through rates in certain locations, start with a free assessment to understand what’s happening with your ad traffic.
