Description: This article answers commonly asked questions when using InnovidXP.
What does InnovidXP need to attribute response to adverts?
Attributing response to adverts refers to the process of determining which marketing or advertising efforts are responsible for specific user responses or actions, such as clicks, conversions, or purchases. It involves identifying and measuring the effectiveness of various advertising channels, campaigns, and touchpoints.
For this attribution to happen, InnovidXP needs the following data:
- Response data: from your website, phone, app, or SMS
- Spot data: from your spot logs or through our spot detection services
- Impression data: via a compressed CSV file sent to an S3 location.
Please contact Innovid if you need help with this at email@example.com.
What is the difference between a click-through and a response?
A click-through is a simplistic, last-touch method of determining success that is typically used in Search Engine Optimization (SEO).
A response is a sophisticated modeled approach that is used to determine the efficacy of a TV campaign. It includes several signals and frequencies that influence the decision-making process and, therefore, the perceived performance of a marketing campaign.
- Responses are typically calculated within a 7-day attribution window, i.e., a visitor can click on a website multiple times in that period, which is measured as one response from that visitor.
- Click-throughs are counted individually, i.e., each time a visitor clicks on a website, it is included in the measurement.
What is a placement group?
This is the grouping used to place ads, it might be the target audience (e.g. Males 18-30) or by a particular channel grouping (e.g. sports cluster), or timing (e.g. July portion), etc.
What do the spot configuration settings mean?
When uploading spots, the Configuration page provides several options, depending on what your file contains and your history of uploading files:
- Apply Settings allows you to apply a previous configuration setting to this file. Just select the file with the settings you want to copy, and they will be applied to the rest of the configuration settings.
- Worksheets will be applicable if the uploader detects that you have more than one sheet in your file with data on it. You will be asked to select the sheet with spot data on it that you want to upload to your InnovidXP portal. Please note that data on other worksheets will not be uploaded.
- Audience can be uploaded in 000s or as a nominal value (1:1). Your choice of upload will affect how the numbers are displayed in the portal.
- Timezone is essential to anyone broadcasting local spots or in a country that spans several time zones. Please select the one appropriate to your file. Local spot times will be converted to UTC and will be used to assist attribution for local spots if you select ‘Regional/Market Area level spots are provided in their local timezone’. If you do not have regional or Market Area level spots, select ‘All spots are provided in the ... timezone’.
What is a BARB clock number?
The Broadcasters Audience Research Board (BARB) uses clock numbers to identify TV ads. A clock number is a unique ID assigned to an ad and is used from when an ad is created to when the ad is aired and then reported. TV broadcasters use it to identify the right ad when they count the number of times it is aired.
We have taken the following example from a Sound Lounge blog about clock numbers:
A clock number comprises 13 letters and numbers in a combination that identifies the agency, the advertiser, the ad version number, and the length, for example, AMVBZBB003030.
The first three letters identify the ad agency. The next four letters are the advertiser. The next three numbers identify which version of the ad this is, and the final three numbers are the length of the ad.
Looking at this example, which is not a real Clock Number, the agency is Abbot Meade Vickers, and the advertiser is BT Broadband. (This advertiser code is made up by the agency, so the brand is often obscure.) This is the third version, and it is a 30-second ad.
What do I do when I receive a tag drop alert?
The web and action tags fire on the page(s) they are placed while obeying the required opt-in/opt-out mechanisms. Any comparison of web volumes and Google Analytics is on a like-for-like basis as the tag is intended to interact in a consistent way with the consent criteria.
Web tags should be placed on the client's website by the web manager. We recommend installing tags two weeks before your campaign goes live to allow time for data QA.
Why did a tag drop occur?
If you have recently performed an update on your website, this may have caused the tags to be dropped.
What should I do?
If you have received an email about a potential tag or response drop, you should reinstall your tags. If you are unsure how to do this, access our article on Placing Tags with Google Tag Manager.
Where do I find my tags?
If you receive an email from Innovid about a tag or response drop, use the tags provided in the email (sent either via a link, a PDF, or within the email itself) to reinstall.
You can also find your tags in your platform within the Data menu, under Responses > TVS Tracker Instructions.
How do I know the tags have been successfully installed?
To check you have successfully reinstalled your tags, you can:
- Load your website using Google Chrome.
- Right-click and select Inspect to display the Inspection window.
- Select the Network tab, reload your site and check for TV2track.js.
You can also download the Chrome extension (excluding US clients) from the Chrome web store and autorun a tag check from the extension icon's dropdown. This check may take a while, so feel free to let it run in the background. Check the final report for tagging alerts, visit tags, and action/KPI tags.
How can I avoid this happening again?
The tag will stay in place until actively removed by your web development team or tag manager administrator. Using comments/labels within the code or tag manager can give your colleagues information on what it is for and why it should be kept in future site optimizations.
Why is my TV influence zero?
It is uncommon for the TV Influence metric to display as zero, but if it does happen, there are several things you can check.
Areas to check:
- Channel mapping
- Channel/regional coverage
- Campaign size
- The first campaign being run using an impression dataset
- TV strategy of brand awareness
- High traffic websites
- Live event streaming
1. Channel mapping
It is important to ensure that all channels have been mapped correctly during platform onboarding or this will impact all results, including TV Influence.
Guidance: review your channel mapping and check it is correct.
2. Channel/regional coverage
Although market adoption for connected TVs and set-top boxes continues to grow, the number of providers involved means linear impression datasets are fragmented. Any single data source tends to have limited coverage, and accessing multiple data sources can be prohibitively expensive. There are no third-party linear impression data providers that offer full coverage across all linear channels and spots.
In cases where an advertiser is advertising heavily on channels with low/no coverage, our control groups may be impacted as households in the group may actually be seeing ads on channels we don’t cover. A similar situation may arise if the third-party linear impression dataset has poor/patchy regional coverage. Again, this may lead to the control group becoming contaminated with households that have actually been exposed to the ads.
In both these instances, the control visit rate would be inflated, which would lead to the TV Influence number being lower, possibly to the point where it is zero or negative (a negative value is always ‘rounded up’ to zero in the InnovidXP platform Campaign view).
Guidance: Innovid will be able to advise on whether there is appropriate coverage for your campaign to be measured using impression-based measurement.
3. Campaign size
If an advertiser’s TV campaign is relatively small scale, it may not be possible for us to detect any TV Influence on the brand’s website.
Guidance: Innovid will be able to advise on whether your campaign size and levels of web traffic are viable for impression-based measurement.
4. The first campaign being run using an impression dataset
If an advertiser is new to InnovidXP, or starting to use a new impression data source, then we must establish a control group to use for campaign measurement.
To do this we need to build up at least 30 days of data history for the advertiser and their viewership to establish a clean control group. Until we have this, the control group visit rate may be inaccurate, impacting the TV Influence metric.
Guidance: Review the data once the clean control group is established after 30 days. At this point, the TV Influence metric will be a more accurate reflection of TV lift.
5. TV strategy of brand awareness
Brands largely using TV to build brand awareness may not have an immediate call to action that drives consumers directly to their website/app. Given we measure TV Influence (web/app activity) over a relatively short period of time (i.e. a monthly campaign), this may result in TV Influence being calculated at zero.
Guidance: Innovid can advise on whether your campaign is appropriate for impression-based measurement.
6. High web traffic sites
Brands with naturally high website traffic may also have issues seeing a TV Influence above zero. For example, large D2C or content streaming brands will have many visits to their website from returning users. Repeat visitors will increase the noise and control group visit rate and reduce our ability to detect a TV-driven uplift.
Guidance: This can be mitigated by tracking specific actions or visits to specific pages on the website rather than looking at all visits to the website.
7. Live event streaming
The above issue can be exacerbated if the brand live streams events on its website, for example, the Super Bowl, as this will drive a natural increase in traffic to the site on that day, increasing the control group visit rate. If an ad for the brand is scheduled to air during that live-streamed event, then any potential uplift from those ads may be masked by the increased control group visit rate.
Guidance: Again, this can be mitigated by tracking specific actions or visits to specific pages on the website rather than looking at all visits to the website.
What are the differences in OTT/CTV metrics between InnovidXP and Extreme Reach?
Clients may see differences between InnovidXP and Extreme Reach metrics, and these are a result of the following:
1. Impression Discrepancies:
- Creative tagging - ensure that all the creatives are tagged with the correct InnovidXP pixel.
- Ad servers - have different methods for counting impressions and filtering valid impressions. InnovidXP counts an impression regardless of whether it was filtered.
- Ad blockers - ad blocking software can prevent an ad from being delivered. However, when the impression pixel has fired, InnovidXP counts the impressions.
- Network connection and server reliability - an ad server may fail briefly or encounter an issue that prevents it from logging an impression.
- InnovidXP can only capture video impressions - no display impressions will be reflected in the platform.
- For all OTT/CTV Partners, InnovidXP cannot attribute Youtube inventory, some set-top box/VOD inventory (e.g. Charter), or track Facebook videos.
- InnovidXP does not handle IPv6 impressions directly, but we are able to collect IPv6 impressions via our IPv4 collectors.
2. Reach/Frequency Discrepancies:
- Frequency metrics for an ad server like Extreme Reach (ER) come from placing a cookie or when an ad is served in an environment where cookies aren't a possibility, an alternative methodology.
- ER calculates campaign reach based on the number of unique cookies seen during the course of the flight.
- Frequency is defined as the average number of times any unique cookie was exposed to a video ad(s) in the campaign. In situations where cookie-based user tracking is impossible due to browsers, devices, or environments that lack cookie support, ER generates a semi-unique user identifier based on observed attributes of the HTTP/S request.
- ER cannot tag a unique device if it's ADH (Google), or a privacy situation (GDPR or CCPA impressions), or in cases like a Safari browser that automatically clears cookies, it could occasionally count as a new unique.
- InnovidXP's unique groups all household devices under one unique, whereas ER's unique counts individual devices under the same HH as separate unique.
- ER For CTV is a semi-unique user identifier based on observed attributes of the HTTP/S request and user agent of the device, instead of the former cookie method.