
How Bot Traffic Wastes Ad Budget: Detecting and Filtering Invalid Traffic in Google Ads
Bot traffic wastes ad budget by generating invalid clicks and fake sessions that Google Ads partly filters but never fully removes. Google credits back only the invalid clicks it detects, so ecommerce advertisers running Performance Max and Shopping campaigns must audit placements, analyze traffic quality, and judge results by profit-based measures like MER rather than platform ROAS alone.
How Invalid Traffic Reaches Shopping and Performance Max
Google Ads defines invalid traffic (IVT) as clicks or impressions not generated by genuine user interest. This includes general invalid traffic (known crawlers, data-center traffic, and obvious automation) and sophisticated invalid traffic (hijacked devices, spoofed apps, and human-like bots). Google's systems filter general IVT in real time and credit advertisers for it, usually as "invalid clicks" in the campaign columns and as adjustments on the invoice.
The gap is in sophisticated IVT and low-quality-but-technically-valid traffic. Performance Max amplifies this exposure because it serves across Search, Shopping, Display, YouTube, Discover, Gmail, and Maps from a single budget. You cannot opt out of individual networks, and channel-level reporting is limited. A Performance Max campaign can spend meaningful budget on Display and app inventory where click fraud and accidental clicks cluster, while the final-URL conversions from branded search and remarketing make the blended ROAS look healthy.
Standard Shopping carries a different risk. Competitor click abuse and scraper bots hit high-visibility product listings, especially for high-CPC categories such as electronics, supplements, mattresses, and jewelry. Low click-through-to-session ratios are the first symptom.
Detecting Bot Traffic: Signals to Audit Weekly
Invalid traffic rarely shows up as one obvious metric. Look for patterns across Google Ads, GA4, and your server logs.
- Click-to-session gap: Compare Google Ads clicks with GA4 sessions by campaign. A discrepancy above 20–25% sustained over two weeks points to bots, bounced redirects, or tracking failures.
- Engagement rate collapse: Sessions with 0–1 seconds of engagement, no scroll, and a single pageview, concentrated in specific cities, ISPs, or device models.
- Geographic anomalies: Spend in regions you do not ship to or never targeted by intent. Check the Locations report for "Location of interest" versus "Physical location" mismatches.
- Time-of-day spikes: Click bursts at 2–5 a.m. local time with zero add-to-carts.
- Conversion quality drift: Rising lead or checkout-start counts alongside flat revenue, or orders with matching IP ranges, disposable emails, or failed payments (card testing).
- Single-product concentration: One SKU absorbing disproportionate clicks with no conversion, common in Shopping feeds targeted by scrapers and competitors.
In GA4, build an exploration segmented by source/medium, hostname, and engaged sessions. Add network type through Google Ads linked dimensions. In Google Ads, review the Invalid clicks and Invalid click rate columns (add them via Modify columns), then compare against your account's baseline. Most healthy accounts sit in the low single digits; a rate persistently above 10% on a campaign warrants investigation.
Filtering Invalid Traffic Inside Google Ads
You have fewer levers in Performance Max than in Search, but several still work.
- Account-level exclusions: Apply placement exclusions at the account level. Google supports account-level placement exclusions that cover Performance Max, which lets you block low-quality websites and apps in one list. Pull the Placement report from Insights and exclude domains with high spend, near-zero engagement, and no conversions.
- Brand exclusions and negative keywords: Performance Max supports account-level negative keyword lists and brand exclusions. Add irrelevant queries (such as "free," "jobs," "manual pdf," and competitor-adjacent terms) to prevent junk query traffic.
- Location settings: Choose "Presence" rather than "Presence or interest" so you only reach people physically in your ship-to regions.
- Customer Match and first-party signals: Feed audience signals from purchasers and high-LTV customers so Google's model optimizes toward real buyers rather than cheap clicks.
- Conversion hygiene: Count only purchases as primary conversions. Demote add-to-cart and page-view events to secondary. Bots trigger soft events constantly, and the bidding model learns from whatever you mark primary.
- Feed segmentation: Separate top-spend, low-conversion SKUs into their own asset group or listing group so you can see and cap them.
For IP-level blocking, Search and Display campaigns support IP exclusions (up to 500 per campaign). Performance Max does not, so use them on standard Shopping-adjacent Search campaigns and rely on exclusion lists elsewhere. Third-party click fraud tools such as ClickCease, Lunio, and TrafficGuard fill part of that gap, but they rely on tagging and cannot control Performance Max delivery directly. Evaluate them by the refund or exclusion volume they produce against their monthly fee.
Why ROAS Hides Bot Waste and MER Exposes It
Platform ROAS counts attributed conversion value divided by ad spend inside Google's attribution window. It is easily inflated: Performance Max claims view-through and branded-search conversions that would have happened anyway, and bot waste on other inventory is subsidized by those conversions. MER (marketing efficiency ratio) is total revenue divided by total marketing spend across all channels. It does not depend on attribution, so waste appears as a decline in blended efficiency.
| Measure | Formula | Where it misleads | Best use |
|---|---|---|---|
| Platform ROAS | Google-attributed revenue / Google spend | Over-credits brand and retargeting; hides invalid spend inside blended campaigns | Day-to-day bid and creative decisions |
| MER | Total revenue / total marketing spend | Ignores channel mix and margin differences | Business-level efficiency and budget scaling |
| Contribution-margin ROAS | (Revenue minus COGS, shipping, fees) / ad spend | Requires accurate cost data | Setting true break-even and target ROAS |
| New-customer ROAS | First-order revenue / spend on new customers | Excludes repeat value | Judging Performance Max prospecting |
A practical example: a store spends $50,000 a month on Google and $30,000 on Meta, with $320,000 in total revenue. MER is 4.0. Google reports a 6.0 ROAS on $300,000 of attributed value, which would imply Google alone drove nearly all revenue. The gap shows over-attribution. If you cut $8,000 of identified junk placements and MER holds at 4.0 or improves, that spend was waste. If MER drops, those placements were assisting sales.
Tuning Bidding Around Clean Data
Smart Bidding learns from conversion data, so bot-driven noise corrupts target ROAS and maximize-conversion-value strategies. Follow these rules:
- Set tROAS from margin, not from hope. Break-even ROAS equals 1 divided by contribution margin. At a 35% margin, break-even is 2.86. Target above it to cover overhead and generate profit.
- Do not change targets more than 15–20% at once. Larger swings reset learning and can mask whether bot filtering helped.
- Wait 2–3 conversion cycles after exclusions. Evaluate on 14–28 days of data for most stores.
- Use new-customer acquisition goals with a value bonus so the algorithm stops overpaying for returning buyers while prospecting traffic is audited.
- Exclude fraudulent orders from conversions. Import offline conversion adjustments (retractions or restatements) for cancelled, fraudulent, or refunded orders so the model stops rewarding them.
Google Shopping Optimization to Reduce Exposure
Feed quality is both a performance lever and a fraud-control lever. Tighten product titles so they match purchase-intent queries, which reduces irrelevant impressions. Use custom labels for margin tier, best sellers, and price band, then split bids or asset groups accordingly. Remove out-of-stock products promptly, because clicks on unavailable items waste budget regardless of source. Review the Search terms and Listing groups reports monthly, and pause any product with spend above 2–3 times its target CPA and no sales.
A Monthly Bot Traffic Audit Routine
- Export the Google Ads Invalid clicks column by campaign and compare with the prior month.
- Reconcile ad clicks to GA4 sessions and engaged sessions.
- Review placements and apply new account-level exclusions.
- Check orders for fraud markers: repeated IPs, mismatched billing and shipping regions, and card-testing patterns.
- Compare MER and contribution-margin ROAS before and after changes.
- Document each change with a date so you can attribute results.
Bot Traffic and Google Ads FAQ
Does Google refund money lost to invalid clicks?
Google automatically credits invalid clicks it detects and filters, shown as adjustments on your invoice. You can also file a request for undetected invalid activity through Google's invalid activity contact form, supplying logs and dates. Credits apply only to clicks Google confirms as invalid.
Can I block IP addresses in Performance Max?
No. IP exclusions are available for Search and Display campaigns, up to 500 addresses per campaign, but not for Performance Max. Use account-level placement exclusions, negative keywords, and location settings instead.
Is MER better than ROAS for ecommerce?
MER is better for business-level decisions because it uses total revenue and total marketing spend, independent of attribution. ROAS remains useful for in-platform optimization. Use both: ROAS to steer campaigns, MER to confirm that spend produces profit.
What invalid click rate is considered high?
Most healthy accounts record low single-digit invalid click rates. A sustained rate above 10% on a campaign, combined with a click-to-session gap above 20%, signals a problem worth auditing.
How long until filtering improves Performance Max results?
Expect 2–4 weeks. Smart Bidding needs time to relearn from cleaner conversion data, so avoid additional major changes during that window and evaluate MER over at least 28 days.


