GIVT vs SIVT: Why Your Ad Reports and Your Analytics Disagree
Two kinds of fake, two kinds of cost
The industry splits invalid traffic into two buckets, and the split is where the money is. General invalid traffic, GIVT, is the traffic everyone agrees is fake: search crawlers, known monitoring bots, datacenter scrapers with no browser. Your analytics already filters most of it, and your ad platform credits it back without an argument. It is waste, but it is recoverable waste.
Sophisticated invalid traffic, SIVT, is the rest. These are bots built to look human: residential IPs, real browser fingerprints, mouse movement, session behavior that passes a casual check. They click ads, fire conversion events, and fill forms. Lunio's 2026 Global IVT Report measured an 8.51 percent invalid traffic rate across paid channels, and 8.2 percent on Meta. On $10,000 a month, that is about $820 paying for nobody, and the platform will not credit all of it back, because some of it looks exactly like a customer.
Why the two dashboards never match
Your ad platform counts clicks. Your analytics counts sessions. A bot that clicks an ad and never loads the page shows up in Ads Manager and never appears in Shopify. A bot that loads the page and bounces in under a second shows up in both, but with different labels. A bot that fires a fake Purchase event shows up as revenue in one system and as nothing in the other.
That is why merchants see 600 link clicks in Ads Manager, 200 sessions in Shopify, and zero sales, and conclude the ads are broken. The ads may be fine. The gap between clicks and sessions is the SIVT. Segmenting traffic by behavior closes that gap: compare landing page views to link clicks, and add-to-carts to sessions, and the bots fall out of the numbers on their own.
The expensive half is the data, not the clicks
The click fraud costs money. The data poisoning costs more. Meta decides who sees your ads by studying who converted. When bots fire ViewContent, AddToCart, and Purchase events, the algorithm treats them as buyers and finds more traffic that behaves like them. Lunio calls this a negative feedback loop: the more invalid events you send, the more invalid traffic you get.
Break the loop at the event, not at the click. A server-side conversion event that only fires after a real payment clears is worth more than ten browser events. A Purchase event gated on a verified order number cannot be faked by a bot that never paid. And a weekly check of landing page views against link clicks catches the drift before a month of spend trains the algorithm on ghosts. Behavioral signals beat CAPTCHAs here, because SIVT bots solve CAPTCHAs and pass fingerprint checks. What they cannot fake is a purchase.
Frequently asked questions
What is the difference between GIVT and SIVT?
GIVT, general invalid traffic, is known crawlers and datacenter bots that tools already filter and platforms credit back. SIVT, sophisticated invalid traffic, is bots that mimic human behavior with residential IPs and real fingerprints, and it is much harder to detect and recover.
Why do my ad clicks not match my sessions?
Ad platforms count clicks, analytics counts sessions. A bot that clicks and never loads the page appears in Ads Manager but not in Shopify. Compare landing page views to link clicks to see how much of the gap is bots.
What is the average invalid traffic rate on paid ads?
Lunio's 2026 Global IVT Report measured 8.51 percent invalid traffic across paid channels, and 8.2 percent on Meta. On $10,000 a month of spend, that is roughly $820 paying for nobody.
How does invalid traffic poison ad targeting?
Ad platforms learn from conversion events. When bots fire fake ViewContent, AddToCart, and Purchase events, the algorithm finds more traffic that behaves like bots. Gate Purchase events on verified orders to break the loop.