18% of your August sessions were bots (and it’s tanking your eCommerce conversion rate)

In August, across the 10 stores we track closely, 18% of sessions were bots. In July, it was 6%. So the share of fake traffic tripled in a single month, and nobody touched a thing on the site. And then, surprise, your eCommerce conversion rate drops off a cliff. Makes sense.

I see this kind of panic all the time. A client calls me in full meltdown mode: "my conversion rate collapsed in August, we need to rebuild the product pages, something's broken." And in 7 cases out of 10, the site is perfectly fine. It's just the denominator that ballooned.

Why a bot kills your conversion rate without breaking anything

You calculate your conversion rate the dumb-simple way: orders divided by sessions. If your session count inflates with traffic that will never buy, the ratio mechanically collapses. Nothing to do with your UX, your pricing, or your homepage photo.

And bots don't buy. In our August data, bots convert at 0.19%. Humans, at 1.86%. So the moment you double or triple the share of bots in your total sessions, your blended conversion rate drops on its own, even though your real human traffic converts exactly the same as before.

Take 100 sessions, 2 orders, you're at 2% conversion. The next month, same 2 human orders, but 50 bot sessions on top. Now you're at 150 sessions, 2 orders, 1.33%. You "lost" a third of your conversion rate. And yet you have the exact same number of real customers. It's that stupid.

That's why the average eCommerce conversion rate you read about everywhere means nothing if you don't know how it was calculated. A site at 1.5% with clean traffic can convert better than a site showing 2% stuffed with bots. Until you're comparing the same thing, the benchmark doesn't help you.

Where this fake traffic comes from

Bots have always existed. There are the good bots: Google crawling your content to rank your pages, Instagram and Facebook pulling previews when someone shares a link. And there's everything else: all the various scrapers that just vacuum up your site with zero intent to buy. That's the usual background noise.

What changed in August is the volume. Across our panel, direct traffic picked up roughly a million sessions between June and August. And honestly, I don't believe a million new people suddenly showed up in the middle of summer. When you dig in, a big chunk of that direct "growth" is bot traffic that platforms dump into direct by default because they don't know where it came from.

The trap is that growing direct traffic feels reassuring in the moment. You tell yourself awareness is climbing, people are typing your brand name. Except direct that explodes with no awareness campaign behind it is rarely a good sign. Look at your direct traffic conversion rate in detail. If it collapses while the volume climbs, you have your answer.

Shopify's bot filter before you rebuild anything

The reflex when conversion rate drops is to want to tear everything down on the site. You rebuild the product page, change the CTA, test three homepage versions. And meanwhile, nobody checked whether the traffic was even real.

It used to be that Shopify didn't distinguish between human and bot traffic in its analytics. Everything was lumped into the same bucket. Now you can filter. In Shopify Analytics, you apply the "Human or bot session" filter and exclude the bots. You recalculate your conversion rate on human traffic only, and then you know whether your site actually has a problem or not.

Personally, to avoid bias, I always cross-check multiple sources. I look at the Shopify view, the Google Ads view, the Meta view directly on each platform, and I compare over identical windows. I don't use third-party tools like Google Analytics to judge conversion, except to get a rough order of magnitude on sessions. By the way, your Google Analytics conversion rate and your Shopify conversion rate will almost never show the same number, because each platform counts sessions its own way. If you mix everything without paying attention, you draw the wrong conclusions.

Another thing that helps: look at sessions by country. If you see a country that makes zero sense for your business, like Singapore jumping ahead of France or Germany when you sell nothing there, that's a signal. Often it's bot traffic, or at least traffic that weighs nothing in orders and just dilutes your ratio.

To sort out your real problems cleanly, start with your audit on your P&L, not your campaigns. You start from real money, not vanity metrics.

And on the ad budget side, it matters too

Something people often forget. If your campaigns bring in bot traffic, you're paying for clicks that will never convert. Google or Meta count the volume, your ROAS looks decent on click count, but behind it there's no purchase. When you clean up your traffic on the analytics side, look in parallel at the conversion rate by acquisition source. A campaign that drives a lot of sessions and converts at basically zero is often where your budget vanishes into thin air.

Watch out for the September 21 trap

Heads up, this one can wreck your comparisons. On September 21, Shopify changed its default setting. Now, the sessions shown correspond to human visitors directly, without you needing to apply the filter.

In practice, that means your sessions and your conversion rate are going to look different from that date on. Mechanically, fewer sessions shown (since the bots drop off), so a conversion rate that goes back up.

The danger is right there. If you compare September to August without accounting for this change, you'll conclude "great, the conversion rate recovered, our actions paid off." When really, you just changed the way the denominator is counted. It's just the counting method that changed, nothing else.

The simple rule. When you compare two periods straddling September 21, drop an annotation on your dashboard, and recalculate both periods on the same basis (humans only). Otherwise you're talking nonsense in the meeting and making decisions on an illusion.

If your conversion rate had "dropped" in August, here's what I'd do

Before panicking, you verify that your drop is real. You start from clean data, not a gut feeling.

First reflex, I open Shopify Analytics, Human or bot filter, and I recalculate conversion on humans only. Then I look at the evolution of direct traffic in volume and conversion rate, because if the volume climbs while conversion collapses, it's bots. I also check sessions by country, an off-market country climbing the rankings is suspicious.

• I cross-check Shopify, Google Ads and Meta over identical windows, without mixing counting methods.
• And I annotate September 21 on every comparison straddling that date.

If after all that your human conversion rate really did drop, then yes, now you can start looking at the site, the merch, the campaigns. And at that point, the conversion and Shopify performance work is exactly what we do day in, day out.

But most of the time, your site is fine. It's just that you counted robots as customers.