4 ms·

More than just a channel appearing to underperform, you can get misleading information without cross-platform attribution. Many common user behaviors for online

More than just a channel appearing to underperform, you can get misleading information without cross-platform attribution. Many common user behaviors for online shopping consist of browsing/searching on a mobile device, and then converting (buying) on their desktop computer when they get home. This type of behavior is very valuable to know about, but difficult to track properly and may lead you to believe you have little mobile engagement and that your web traffic is killing it with the organic conversions, when really it's totally reversed and your mobile product is the major driver for desktop conversions.

themartorana 10y ago
TV advertisers knew this - called Effective Frequency - the "number of times a person must be exposed to an advertising message before a response is made and before exposure is considered wasteful." Thing is, we try to track every damn thing, when there's still effective audio and visual advertising that, while not necessarily driving a measured online conversion, may still have you reach for a Pepsi over a Coke at a Wawa while getting gas. So where naive advertisers might make such a correlation and believe their organic traffic is spectacular, I would think season advertisers would know an ad campaign's effectiveness isn't always in an immediate and wholly measurable conversion. I could be wrong. https://en.wikipedia.org/wiki/Effective_frequency?wprov=sfsi1 https://en.wikipedia.org/wiki/Effective_frequency?wprov=sfsi...
- shostack 10y ago
  So I would humbly consider myself a seasoned advertiser. I know there's a major cross-channel and cross-device picture happening. The challenge often comes when needing to report on metrics. Unfortunately, the saying "what gets measured gets managed" cuts both ways. So I can know with all my heart that something is driving latent activity, but if I can't report on it, often times that isn't sufficient because people want to see measurable ROI when it comes to digital media. View-through and assist conversions try to help with this, but then you get into the valuation question of "what is the value of a view-through or assist conversion?" Similarly, "what is the value of a Like?" Ultimately a company needs to build a model to back things out against their targets, and so unless you have solid incremental lift testing data, or other MMM-type data, you're faced with the really hard task of trying to justify this stuff that isn't always showing measurable performance.
shostack 10y ago
Yep, this was what I was getting at. Understanding the role of the device or channel in your path to conversion is absolutely critical for ensuring that you are making intelligent decisions and not misattributing performance to another channel (Direct and Organic being common culprits). There's a surprising number of devils in the analytics details depending on a given site's setup (including things like payment processors showing up as referrals unless you exclude them) that can mess with things. Thankfully a lot of the big ones are "known unknowns" now that you can try to deal with or account for.
abraae 10y ago
Quite the problem, just how do you handle this? Are there any workable approaches that don't involve trying to get the user to login on both the mobile and then the desktop in order to correlate?
- ssharp 10y ago
  Some ad networks are already taking in user data through hashed email addresses. There are other things that can be done though. Google Analytics supports User Id's, so you can do a better job at attributing to specific users across multiple devices. GA has traditionally discouraged passing this PII into it's platform but is now supporting this limited ID information. Other analytics packages have allowed this type of identification information for a lot longer. Specific to advertising, there are also other tactics you can use to value your impressions that don't involve individually identifying people.

themartorana 10y ago

TV advertisers knew this - called Effective Frequency - the "number of times a person must be exposed to an advertising message before a response is made and before exposure is considered wasteful." Thing is, we try to track every damn thing, when there's still effective audio and visual advertising that, while not necessarily driving a measured online conversion, may still have you reach for a Pepsi over a Coke at a Wawa while getting gas. So where naive advertisers might make such a correlation and believe their organic traffic is spectacular, I would think season advertisers would know an ad campaign's effectiveness isn't always in an immediate and wholly measurable conversion. I could be wrong. https://en.wikipedia.org/wiki/Effective_frequency?wprov=sfsi1 https://en.wikipedia.org/wiki/Effective_frequency?wprov=sfsi...

shostack 10y ago
So I would humbly consider myself a seasoned advertiser. I know there's a major cross-channel and cross-device picture happening. The challenge often comes when needing to report on metrics. Unfortunately, the saying "what gets measured gets managed" cuts both ways. So I can know with all my heart that something is driving latent activity, but if I can't report on it, often times that isn't sufficient because people want to see measurable ROI when it comes to digital media. View-through and assist conversions try to help with this, but then you get into the valuation question of "what is the value of a view-through or assist conversion?" Similarly, "what is the value of a Like?" Ultimately a company needs to build a model to back things out against their targets, and so unless you have solid incremental lift testing data, or other MMM-type data, you're faced with the really hard task of trying to justify this stuff that isn't always showing measurable performance.

shostack 10y ago

Yep, this was what I was getting at. Understanding the role of the device or channel in your path to conversion is absolutely critical for ensuring that you are making intelligent decisions and not misattributing performance to another channel (Direct and Organic being common culprits). There's a surprising number of devils in the analytics details depending on a given site's setup (including things like payment processors showing up as referrals unless you exclude them) that can mess with things. Thankfully a lot of the big ones are "known unknowns" now that you can try to deal with or account for.

abraae 10y ago

Quite the problem, just how do you handle this? Are there any workable approaches that don't involve trying to get the user to login on both the mobile and then the desktop in order to correlate?

ssharp 10y ago
Some ad networks are already taking in user data through hashed email addresses. There are other things that can be done though. Google Analytics supports User Id's, so you can do a better job at attributing to specific users across multiple devices. GA has traditionally discouraged passing this PII into it's platform but is now supporting this limited ID information. Other analytics packages have allowed this type of identification information for a lot longer. Specific to advertising, there are also other tactics you can use to value your impressions that don't involve individually identifying people.