HTML5 IMA SDK - DFP event pixels dispatched on incorrect VPAID events

[wnev...@innovid.com]

Hi IMA team,

Innovid has been encountering cases of discrepancies with publishers utilizing the native IMA SDK or another player wrapping IMA SDK. In particular, we are running into discrepancies on the 'impression' event. Please see the below inline image (attached as well):

To explain, we are seeing each VPAID Event in the 'Event' column correspond with the following DFP pixel triggered by IMA. The video player tested on was Springboardvideo, wrapping HTML5 IMA SDK, linked here: <http://goo.gl/fk81YO>.

The above results come from testing different JS VPAID ad units with the VPAID events suppressed from being sent to the player. For example, we no longer saw the 'vast_creativeview' DFP pixel fire when the ad unit suppressed sending the 'AdStart' VPAID event to the player.

Would you be able to look into and/or confirm the above? Additionally, could you please confirm which pixel is in charge of which metric (e.g. which label listed above corresponds with the 'start' metric in DFP reporting)? We are encountering discrepancies on 'impression' metrics, as the 'vast_creativeview' pixel which corresponds with the impression is not being fired alongside the 'AdImpression' VPAID event.

Please let us know any other information you need from us, or if you have any questions. We're happy to jump on a call if necessary or if you feel it would be beneficial.

Cheers,

Will Neville

Innovid Tier II Engineer

[Tyler Sidell (IMA SDK Team)]

Hi Will,

Thank you for sending over those details. I can confirm that some of the IMA events do not necessarily correlate in a one-to-one relationship with the DFP label/metric. You have it correct that Q1, Q2, Q3, and Q4 respond to videoplaytime25, videoplaytime50, videoplaytime75, and videoplaytime100 respectively. 

We can talk from the IMA-side and only confirm if the impression pings are occurring.  We've checked your page and have noticed that the pings are in fact occurring. You can confirm this by opening the Developer Tools on the browser and checking the Network tab. After the SDK pings these impressions, it is up to the server to interpret them.  Please note that the impression ping is not part of "conversion" and is in fact part of pcs/view instead.  For more information about viewability metrics please follow the DFP documentation.

Thanks,

Tyler Sidell

IMA SDK Team

[wnev...@innovid.com]

Hi Tyler,

Thank you for your response. The above table was generated using the Chrome 'Network' tab under Developer Tools in conjunction with Fiddler.

The issue from our point of view is not whether or not the impression pings are occurring, but rather on what event they are occurring. We set up custom VPAID creatives to respond to the the test page linked; each of these VPAID creatives suppressed notifying the player of VPAID events (AdStart, AdImpression and AdVideoStart).

We noticed that the impression ping still occurs even when the 'AdImpression' VPAID event is not fired from the ad unit. Additionally, we noticed that the impression ping does not occur if the 'AdStart' VPAID event is suppressed. Per the IAB VPAID spec, the impression ping should occur only in response to the AdImpression VPAID event.

Would you be able to confirm at what point/alongside which event the impression ping is fired by the SDK?

Let us know if you need any more information, thanks!

Cheers,

Will

[SLee]

My two cents: AdVideoStart and Ad Impression are pretty much the same. As one who'd developed VPAID creatives I've sometimes wondered 'when in the workflow should I dispatch AdImpression ?' It seemed redundant with  AdVideoStart. I would usually end up dispatching it after the video was buffered and one frame of the video had been painted to the screen. Some would say it should be after one or two seconds of the ad video have rolled.

Anyway it seems the SDK has made this question moot because looking at your chart, it is reporting Impression to the DFP server on either AdStart or AdLoaded .

[Tyler Sidell (IMA SDK Team)]

@Slee, thank you chiming in with that information. 

So I understand that you are concerned as to why there is no impression event.  The issue seems to lie in the wrapping of the SDK by Springboardvideo player so I would suggest checking in with them as well.  I've came to this conclusion because if you use your tag in our VSI you will see the AdImpression Event does get fired and pings the impression. When you place your tag into the VSI make sure that you are changing HTTP to HTTPS.

Example:

<VASTAdTagURI>
<![CDATA[
https://rtr.innovid.com/r1.57a372270d2d97.77583285;cb=444849452
]]>
</VASTAdTagURI>

Thanks,

Tyler Sidell
IMA SDK Team

[wnev...@innovid.com]

Hi Tyler,

Just to confirm that we're on the same page, the concern on our end is that the IMA SDK's impression ping is occurring on an event other than 'AdImpression'.

Would you be able to provide an example of the impression ping that the SDK sends out as viewable in the Chrome 'Network' section under Developer Tools?

Lastly, please see the attached screenshot. Here, I am using Fiddler to auto-respond to the tag running with a custom creative that does not send the 'AdImpression' or 'AdVideoStart' events to the player. As a result, these events do not show up the event log beneath the player - this log appears to be just displaying what events the ad unit is sending to the player, and not what the player is doing in response to those events. I bring this up because under these same conditions (i.e. using the same creative with the VPAID 'AdImpression' and 'AdVideoStart' events prevented from being sent to the player) we still see the DFP impression pixel, leading us to believing that the IMA SDK impression ping is being sent on something other than 'AdImpression'.

Please let me know of the above regarding viewing the SDK's impression ping in Chrome, or if you have any questions on the above.

Thanks so much!

Cheers,

Will Neville

[wnev...@innovid.com]

Quick followup - forgot to attach the image. Apologies!

[Tyler Sidell (IMA SDK Team)]

Hi Will,

You could use the "VPAID testers ad" from our VPAID sample page to test the events and see what gets pinged.  I've set up a simple working example in our VSI, that you can use to test the events firing and I'd suggest to open up the Network tab to see what gets returned.  I tested this out and it looks like the impression url that is found in the tester XML is pinged on "Ad Started".  I'll speak to the rest of the team about the functionality and discuss if this should be fired on "Ad Impression".

In the meantime, you can also check out the event firings of your specific ad tag in this other VSI example that I have set up for you.  In this example, you should see in the log events that the Ad Event: Impression is indeed getting fired.

Thanks,

Tyler Sidell

IMA SDK Team

[wnev...@innovid.com]

Hi Tyler,

Great, thank you for your message. Indeed, the impression url being pinged on 'AdStarted' rather than 'AdImpression' is what we believe to be the cause of the discrepancies on our end.

Would you be able to provide an ETA regarding this issue? We have several clients affected by this issue that are waiting to launch HTML5 JS VPAID campaigns.

Thanks!

Cheers,

Will Neville

Innovid

[Tyler Sidell (IMA SDK Team)]

Hi Will,

Glad to hear that we got to the bottom and figured out the reasoning to your issue.  I cannot provide ETAs but I will bring this information back to the rest of the team and will update this thread if I have further information.

Thanks,

Tyler Sidell

IMA SDK Team

[wnev...@innovid.com]

+Steve from WebMD,  Devin from Intellisphere, Tory from US Magazine

Hi Tyler,

Hope you've been well!  Wanted to check in to see if you have been able to regroup with your team regarding this issue? There is a sense of urgency as certifications for HTML5 JS VPAID are being stalled due to the discrepancies caused by the IMA impression ping being fired alongside AdStart rather than AdImpression.

Let us know if you have any questions or if there are any issues.

Cheers,

Will

Innovid Tier II Engineer

[Tyler Sidell (IMA SDK Team)]

Hi Will,

We understand your concerns and are working to get this up to date according to the specs.  We have no available timeline right now but I have let the team know and will update once the fix is live.

Thanks,
Tyler Sidell

IMA SDK Team 

[Artan Nikaj]

Hi,

Can you please provide an ETA by when this will be fixed? 

This issue is preventing our certification with Innovid / VPAID which is the only thing hurdle left for us to go HTML5 first instead of flash. 

Thanks,

Art

[Tyler Sidell (IMA SDK Team)]

Hi Artan,

I've confirmed with the rest of the team that the fix has gone live.  Testing on our VSI, you can see that the AdImpression event gets fired correctly.

Thanks,

Tyler Sidell

IMA SDK Team

[Gil Atlan]

Hi all,

I'm not seeing the expected result:

While testing with VSI tester, I select the AdStarted button, and fire this event by clicking on the 'trigger event' button, I see that the player fires the following pixels:

http://www.example.com/impression

http://www.example.com/creativeView

These 2 pixels should only fire when the VPAIDEvent.AdImpression is dispatched by the VPAID ad.

Additionally, if I start fresh and choose AdImpression + trigger the event, none of the VAST pixels are fired.

Expected result: 

When the VPAID ad dispatches the VPAIDEvent.AdImpression, the player should fire its impression pixel.

-Gil

Innovid Tier II Engineer

[Tyler Sidell (IMA SDK Team)]

Hi Gil,

We've tested this and it was working earlier.  I will bring this back to the team and keep you posted with updates.

Thanks,
Tyler Sidell

IMA SDK Team

[Artan Nikaj]

Hi,

Can you please provide an update on this issue?

Need to update stakeholders.

Thanks,

Art

[Tyler Sidell (IMA SDK Team)]

Hi Artan,

The issue is being worked on but we do not have an ETA as of right now.  We will communicate any further updates further here.

Thanks,

Tyler Sidell

IMA SDK Team

[Abhijit Pande]

Hi IMA Team,

We are having almost 90% discrepancy between Google DFP and LKQD tags.  We provided the following tags to the DFP and they are recording more impression than us.  So I'm not sure if this issue is related to the earlier discussion where DFP impression ping was occurring on some other events.  Could that be reason for such high discrepancy?

I think all events are firing correctly as far as I can see though not sure why DFP is counting more impression than LKQD.  

You can find the attached VAST/VPAID tag and test page URL where this tag is up.  On this test page ideally we should see LKQD default ad is loading which means everything is working correctly from LKQD and something is not working on DFP.  Would be great if you could take a look on this.

Let me know if you need anything.

Best

Abhijit

[Tyler Sidell (IMA SDK Team)]

Hi Abhijit,

The tracking of impressions by DFP and LKQD are out of the realm of our expertise.  For accurate information on these metrics, I would suggest reaching out to DFP and LKQD. 

I did however notice that the lkqd tag that you provided does not have any impression node which could lead to the discrepancy.

As for the original issue with impressions firing on AdStarted vs AdImpression, our team is currently investigating and assessing the impact.  I will reach out once I have additional information.

Thanks,

Tyler Sidell

IMA SDK Team

[Abhijit Pande]

Hi Tyler,

Thanks for your reply.  I just gave the sample tag and what DFP guys have uploaded is with macros with &pageurl=%%REFERRER_URL_ESC_ESC%%&rnd=%%CACHEBUSTER%% it should pick up.  So my point is could this be reason for the discrepancy as impression firing event adStarted Vs AdImpression has issue?

Thanks for taking a pain to look at it.  Post this I will reach out to DFP team certainly.

Best

Abhijit

[Tyler Sidell (IMA SDK Team)]

Hi Abhijit,

The macros that are added by DFP are independent of the AdStarted vs AdImpression issue that we are investigating.  I will reach out once we have further information on that issue.

Thanks,

Tyler Sidell
IMA SDK Team

[Abhijit Pande]

Sure thanks

[wnev...@innovid.com]

Hi Tyler,

Hope you're doing well! Wanted to reach out to see if you had an updated ETA?

Additionally, to give some visibility into how this issue affects live reporting, please see the attached DFP report. This report is for an HTML5 JS VPAID tag that was run on Desktop where the HTML IMA3 SDK is utilized. Please note the difference between the 'Impression' and 'Start' metrics; per the IAB spec, a video impression should be counted when the video begins playing, and as a result these numbers should be very close if not equal.

Let me know, thanks!

Cheers,

Will Neville

Innovid Tier 2 Engineer

[Tyler Sidell (IMA SDK Team)]

Hi Will,

While we understand the urgency of your request, we do not have a definitive ETA at the moment.  Although we need to be compliant with the spec, this affects impressions across multiple VPAID providers so our team is currently calculating the impact.  Rest assured this issue is being addressed and being worked on but we don't have a solid update at this time to provide.  I will pass along the DFP report that you attached to the rest of the team.

Thanks,

Tyler Sidell

IMA SDK Team

[Fred Nostro]

+ Andrew Contreras from Rocketfuel

[Gil Atlan]

IMA team,

Do you have any updates for this issue?

The IMA player is currently causing discrepancies with VPAID 2 vendors as it does not fire its impression pixel according to video/VPAID specs.

If you ask for severity, please have a look at the (partial) list of publishers who we CC'ed to this thread.

Innovid is already certified with hundreds of publishers/SSP/DSP and whenever IMA/DFP is involved, we see these discrepancies, showstoppers to any campaigns.

-Gil

[Chris Feldman (IMA SDK Team)]

Hi Gil,

I don't have an update yet. This is something that our team is still working on. I will be sure to let you know as soon as there are any changes.

Regards,

Chris Feldman

IMA SDK Team

[Deepika Uragayala (IMA SDK Team)]

Hello Gil,

Just a heads-up, the fix is live. Let us know if you still see the issue.

Regards,

Deepika Uragayala

IMA SDK Team

On Thursday, February 16, 2017 at 5:07:26 AM UTC-5, Gil Atlan wrote:

[Gil Atlan]

I can confirm I'm seeing the expected result while viewing it on http://ryanthompson591.github.io/vpaidExamples/

We would monitor our next impression tests with the publishers and let you know if we see anything funky.

Thank you for fixing this issue.

-Gil