local mode vs. javascript: huge difference

[faklork]

Hi!

When I use local mode for invocation, I get much higher imps and ctrs. However, there is a warning that this wold not comply with IAB rules. Can somebody please explain?

 

Cheers,

Alex

[andrewatfornax]

Hi @faklork,

With most invocation methods, when Revive Adserver delivers a banner, there's a logging beacon that is delivered immediately after the banner, so that an impression is only recorded once the banner has actually been displayed. However, with local mode invocation, Revive Adserver simply records the impression at the point that the tag is called to display the banner.

That's why you get higher impressions counts with local mode invocation - it's impossible for the request for a banner to come into Revive Adserver, but for the user to then "navigate away" from the page before the impression is recorded. 

This (i.e. how local mode tags record impressions) is not how IAB guidelines state impressions should be recorded.

See also: https://documentation.revive-adserver.com/display/DOCS/Requests+Higher+than+Impressions

(Although I note that we don't mention local mode tags there - something that needs updating; unless, of course, I am totally wrong about all of this!)

Please see thread below - I was wrong about this.

[faklork]

Thanks for your time to answer this!

 

10 hours ago, andrewatfornax said:

See also: https://documentation.revive-adserver.com/display/DOCS/Requests+Higher+than+Impressions

(Although I note that we don't mention local mode tags there - something that needs updating; unless, of course, I am totally wrong about all of this!)

Yes. I knew that page but since it did not mention local mode ...

I get it now, but isn't this somewhat counterbalanced by higher CTR? I guess after all most  banners ARE served because the CTR definitely IS higher. In your scenario the CTR would be less, or not? No Banner, no click ...

I think that it could be possible that local mode serves as a sort of "workaround" for Ad Blockers, since there is no JS to block, and if you rename your directories there wouldn't be much to detect. I am wrong?

 

Cheers,

Alex

Edited April 11, 2017 by faklork

[andrewatfornax]

Hi @faklork,

No banner, no click, certainly - but also no impression! So, the sum change in that case is zero, so I don't see how that would affect the CTR, either up, or down. The CTR is only calculated off the impressions that actually happen.

How are you comparing the CTR being higher with local mode tags vs. something else? That is:

• What other tag(s) are you comparing it against? and
• Are you doing in-place A/B testing of identical zones on identical pages, and comparing the CTRs?

[faklork]

On 11.4.2017 at 2:21 PM, andrewatfornax said:

Hi @faklork,

No banner, no click, certainly - but also no impression! So, the sum change in that case is zero, so I don't see how that would affect the CTR, either up, or down. The CTR is only calculated off the impressions that actually happen.

How are you comparing the CTR being higher with local mode tags vs. something else? That is:

• What other tag(s) are you comparing it against? and
• Are you doing in-place A/B testing of identical zones on identical pages, and comparing the CTRs?

"How are you comparing the CTR being higher with local mode tags vs. something else? "

Local mode shows more impressions AND more clicks. JS shows fewer impressions AND *a lot* fewer clicks. So CTR is higher in local mode.

"Are you doing in-place A/B testing of identical zones on identical pages, and comparing the CTRs?"

Only one Zone in use, invocation the same on all pages. Comparing imps/clicks/ctrs per banner from days/weeks before switching local mode to JS mode.

I just looked at one of my banners. The day I switched from local mode to JS, daily average impressions fell to about 1/3, and clicks went down to almost ZERO. I compared that with other campaigns, banners, it is roughly the same.

While I understand that impressions are lower with JS invoking (as you explained above), I don't get why the clicks are also down? In local mode I have so much more clicks. Clicks should be the same, or not? click is click, after all?

Puzzled,

Alex

[andrewatfornax]

HI @faklork,

Yes, I'm puzzled too - that doesn't make any sense at face value.

So, let's work through it as methodically as we can.

1. If you test the JS tags on your site yourself, can you confirm that across browsers that you see visiting your site, that clicks do actually work, and are actually recorded? You may want to set up a separate zone on a hidden test page on your site to confirm this, and be able to keep your test traffic separate from anything that visitors to your site are generating.
2. Assuming that everything is okay, could you split your delivery at random, 50/50? That is, in your live site, set up duplicates of your zone(s), with the same banners linked, and 50% of the time, randomly deliver one of the zones - one always delivered using the localmode tag, and one always delivered using the JS tag. Then the traffic will be split 50/50 across the two zones, and you can compare fairly the number of impressions in each (how well do they compare in terms of total # of impressions delivered?) and also the CTR between the two. The delivery in each will be happening at the same time of day, in the same place, with the same content, so it would be a completely fair comparison.

[andrewatfornax]

... and following up, I was wrong above. Local mode uses a logging beacon, just like JS does.

The reason local mode tags are not okay as far as the IAB is concerned is because the call to display the ad is not initiated by the client (web browser), but rather by the server (web server).

 

... All of this means that there's no reason why local mode and JS should be vastly different in terms of impression count and CTR, so something is going wrong. We just need to track down what, and why.

[Josh]

This is due to bots. Bots most often won't execute js tags but will follow <img> (logging beacon) and <a> (click tracker). You can confirm this by looking at your web server logs.