Android Webview & Safari (in-app) without any referrer information in Google Analytics

[SarahLK]

I've got a client that gets a considerable amount of traffic that:

1. Enters on an interior page (that we are running properly tagged ads for on various platforms)
2. Bounces
3. Has a browser of either Android Webview or Safari (in-app) which I believe indicates the website is being viewed within a mobile app
4. Has no referrer data: (direct) / (none)

We've tested several scenarios (Facebook app, Gmail app, etc.) and ruled them out. Anyone know what this might be?

Thanks in advance!!

[JonoAlderson]

Oh, excellent work!

Good to know for the future!

[SarahLK]

UPDATE: We were able to get tracking code on the Pandora ads and determined that it was the source of nearly all of those visits. So, it WAS Pandora all along.

[JonoAlderson]

Ah, that's interesting!

We don't get Pandora over here in the UK so it's been a while since I touched it. Does it still require/run in flash? That might go some way to explaining weird referrer data...

[SarahLK]

Hey! I haven't resolved the issue yet, but I'm working with the client to add some tracking to all of their Pandora ads. I'm thinking that might be the culprit. Turns out the are running several ad types so there are some that may not be tagged after all.

The implementation isn't the issue, because the only time we see this 11 issue is from this specific traffic source. Everything else behaves normally and typically aligns with metrics we get from other sources.

I will definitely keep you posted about updates. (And will probably throw a party when I finally get it resolved!)

[JonoAlderson]

Hey Sarah,

Just checking back to see if you're still struggling with this?

[JonoAlderson]

Hah!

Ah, now, that's particularly interesting. Got a link to the site handy? Might be worth exploring the implementation.

[SarahLK]

The more I learn about this, the less I know.

It appears that each individual session is actually being counted as 11 sessions. Unless, of course, 11 people from the same small town visited the website on the same day, viewed the same pages and entered the exact same typo into the coupon code box.

Everything my mystery traffic does appears to happen in multiples of 11. I'd love to think that it's just the Internet reminding me that Spinal Tap is awesome, but I suspect there may be more to it.

[JonoAlderson]

Yikes, definitely sounds like there's some exploring to be done!

I'm going to see if I can rope in some smart people to help, and get back to you.

[SarahLK]

Thank you so much for your response!

Yep. I've looked at it by screen resolution, device, location, ISP, demographics...just about everything. Every indication is that this is NOT bot traffic. The distribution percentages of the various attributes are all fairly similar to the "regular" traffic. The only exception is that my phantom traffic appears to be a little older. The traffic patterns also follow the regular traffic, and those patterns are usually influenced by offline advertising efforts.

This feels like some kind of rogue mobile ad campaign, but we have ruled out all mobile campaigns that we are knowingly running (Facebook, Instagram, Google Display on Pandora, etc.). We are also seeing bounce rates significantly higher than the rest of our mobile campaign traffic which would theoretically rule out the possibility that our tracking code is getting stripped out somehow on some of our campaigns. We have ruled out both promotional and transactional emails.

This is easily the most frustrating analytics mystery I've ever encountered!

[JonoAlderson]

Hey there,

My gut feeling is that this sounds like bot traffic.

Many generic systems represent themselves as safari-compatible agents, given that this is one of the main mobile web platforms.

I'd be interested in understanding how this traffic looks when you segment it by other attributes; specifically, screen resolution, colour depth and a few other of the 'system' variables in GA tend to give some good clues as to whether this is human or bot traffic. Have you explored these variables, and seen unusual patterns (such as everybody using the same browser, or the same screen resolution)?

Even if there's variation in these areas, it's worth looking at the patterns over time - there are a lot of systems which, for example, monitor site performance, and as part of this legitimately emulate a variety of browsers... However, they tend to do so consistently, e.g,. at the same time of day.

I'd definitely do some digging to see if you can see patterns which suggest that this isn't real traffic, and then, either way, work out what the next steps are.

I presume that you've enabled the setting in your GA profile to exclude common bots/agents?