Data Integrity & Accuracy

We’re Here to Help!

If you need help navigating the Hawke AI, please do not hesitate to contact the Hawke AI team for support.

Overview

With any data aggregation platform, it is important to know how, and when, that platform is accessing connected data feeds. At Hawke AI, we want to ensure that our users are aware of how data feeds are managed to ensure the highest level of accuracy and consistency when comparing the data in Hawke AI to the individual ad or analytics platforms.

Below we share how we access connected data streams and offer methods for you to validate that your Hawke AI data, and therefore insights, are accurate.

Data Freshness of Connected Platforms

GA4

• All data accessed on display of the page in Hawke AI via API.
• Google does note that there can be up-to a 48 hour delay, depending on the quantity of data in an individual Google Analytics account.

Google Search Console

• All data accessed upon display of the page in Hawke AI via API.

Google Ads

• Data within 4 days accessed via API.
• Data older than 4 days accessed via data store.

Meta Ads

• Data within 7 days accessed via API.
• Data older than 7 days accessed via data store.

Microsoft Advertising

• Data within 24 hours accessed via API.
• Data older than 24 hours accessed via data store.

LinkedIn Ads

• Data within 24 hours accessed via API.
• Data older than 24 hours accessed via data store.

TikTok Ads

• Data within 7 days accessed via API.
• Data older than 7 days accessed via data store.

Pinterest Ads

• Data within 7 days accessed via API.
• Data older than 7 days accessed via data store.

Shopify

• Data within 24 hours accessed via API.
• Data older than 24 hours accessed via data store.

Maropost

• Data within 7 days accessed via API.
• Data older than 7 days accessed via data store.

Matching Your Data

The best way to match individual platform data with that of Hawke AI is to use the individual platform reports, found under the Reporting menu. Set the same date range in Hawke AI and in the platform to compare high-level metrics.

It is best to select historical date ranges beyond 24 hours (72 hours for Google Ads matching) to ensure that the data in the platform matches what is provided via API to Hawke AI.

Mismatch follow steps: date range correct? Channel reports live, reporting summary daily, Caching etc.

Google Analytics 4:

• In Hawke AI, navigate to Reporting –> Google Analytics 4
• To validate sessions and engagement rate:
  • In Google Analytics, navigate to Reports –> Acquisition –> Traffic Acquisition
  • Ensure the same date ranges are selected
  • Compare the values for sessions and engagement rate
• To validate conversions
  • In Google Analytics 4, navigate to Reports –> Engagement –> Conversion
  • Ensure the same date ranges are selected
  • Compare the values for conversions

GA4 and thresholding

As you’re likely aware the move from Universal Analytics to Google Analytics 4 brought with it the introduction of data ‘thresholding’. The purpose of this thresholding is to power a function known as Google Signals.

What is Google Signals?

Google Signals is basically how Google has attempted to tackle how browsers/sites/users have become hostile to cookies. It’s a way to collect user data without requiring as much user consent and then allows Google to aggregate behaviour across other harder-to-track users. You can then see and do a bit more with your user demographics in GA4 – hence why it is so preoccupied with anonymizing data, which it does with thresholding. This is why when you’re looking at a view on GA4 with a single dimension, then add a second dimension, the metric totals misalign. Thresholding is being applied at each step.

Can I turn off Google Signals?

If you don’t particularly use the more detailed demographics info on GA4 – and it can be used for a lot of powerful functions, reporting, creating audience lists that you import into Google Ads etc. – then you are able to turn it off. You will then not have thresholding applied but you also will have reduced access to demographic information and potentially a less clear picture of total conversion activity.

Should I turn off Google Signals?

Every situation has its own nuance, so it is difficult to advise. Generally speaking Google Signals is probably particularly useful for clients with larger budgets and significant traffic through their site; where the specific volumes aren’t as important as the general trends. The benefit is that it allows you to then use its audience data in remarketing, reporting, and tracking across devices – which is incredibly useful. 

For smaller clients with small budgets and limited traffic, Signals could make things pretty confusing. Often with smaller clients the specific numbers are important and GA4 will spit out something completely different every time you add a new parameter and it goes ahead and applies its thresholding differently.

What’s the impact to GA4 metrics in Hawke AI?

The upshot of thresholding is that at times Hawke AI and GA4 may misalign on some metrics. To try to normalize results, Hawke AI will pull the daily data from GA4 for each day in your selected date range and then add it together to create a total. In GA4, selecting a date range will result in an applied threshold for the totals of that range. You can test it out by comparing the totals of a 7 day view and the sum of each individual day within that view – there’s a good chance the numbers will be slightly different. For more information on how Hawke AI handles thresholding, please reach out to us: support@hawke.ai

Google Search Console:

• In Hawke AI, navigate to Reporting –> Google Search Console
• In Google Search Console, navigate to Performance –> Search Results
• Ensure the same date ranges are selected
• Compare all four of the default metrics (note that Google rounds their display of these metrics)

Google Ads:

• In Hawke AI, navigate to Reporting –> Google Ads
• In Google Ads, navigate to All Campaigns –> Campaigns and ensure that there are no active filters
• Ensure the same date ranges are selected (best to test with dates older than 3 days)
• Compare the metrics from the bottom of the Google Ads page (see “Total: Account” row) to the values shown at the top of the Hawke AI page
• Potential issues:
  • Conversion-based metrics (Conversions, Conversion Rate, CPA) do not match – this is likely conversion lag due to the conversion window set in Google Ads (conversions may continue to increase in Google Ads, while Hawke AI does not monitor for these changes after 72 hours)
  • Search Impression Share values may differ between Google Ads and Hawke AI as a result of the estimated and aggregation for this metric. See FAQ for more details.
  • Hawke AI contains a subset of your Google Ads campaigns due to selective importing of campaigns during Hawke AI setup

Meta Ads:

• In Hawke AI, navigate to Reporting –> Meta Ads
• In Meta Ads, navigate to Campaigns and ensure that there are no active filters
• Ensure the same date ranges are selected
• Compare the metrics from the bottom of the Meta Ads table (see “Results from # campaigns” row) to the values shown at the top of the Hawke AI page
• When you pull in metrics from Meta Ads, please check to see whether the metric is web vs. on-meta vs. a total and ensure you pick the one you need.
• Potential issues:
  • Conversion-based metrics (Conversions, Conversion Rate, CPA) do not match – this is likely conversion lag due to the conversion window set in Meta Ads (conversions may continue to increase in Meta Ads, while Hawke AI does not monitor for these changes after 24 hours)
  • Hawke AI contains a subset of your Meta Ads campaigns due to selective importing of campaigns during Hawke AI setup

Attribution in Meta

Earlier in 2021, Apple released an update to iOS with additional privacy features which would allow users to prevent apps from certain types of tracking. In response, Meta made changes to how data was collected and reported on within their platform and to API partners, such as Hawke AI. Note: this generally impacts conversion-based numbers only (such as conversions, purchases, and add to carts), and should not impact spend, impressions, or click-based metrics.

The primary issue stems from blending data points across multiple campaigns (and/or ad sets), all of which can have different attribution models and conversion types.

To illustrate this, see the screenshot below, which notes multiple attribution settings across a set of campaigns, which then prompts a note as to why Meta (via Ads Manager) cannot sum certain metrics:

Extending upon this, when hovering over the superscript within the Results and Cost per Result data points, there is a note that the metric may not be including data from iOS 14.5 (or higher) users:

Within Meta Ads Manager, there are a lot of unknowns for as there is no delineation of those with tracking permitted and those without. Additionally, the Meta Ads API (used by platforms such as Hawke AI) does not provide clear ways to match API-provided data to that shown within Meta Ads Manager – the good news is that Meta, via the API, provides a boolean flag to API users with a new “unified” attribution setting, which does include opted-out iOS users.

Managing Unified Attribution in Hawke AI

In Hawke AI, we have made a few changes to how users can toggle the way these impacted Meta Ads metrics behave – sometimes unlocking data that is not shown in Meta Ads Manager. While the defaults often cover the majority of instances, you may need to try a few different settings to find what makes sense (and matches) for you.

You can select the type of attribution you wish to use in the Client Account Settings → Conversion Settings page, under Meta Ads Attribution. Once selected, you can then select the views and clicks attribution window relevant to your Meta Ads setup.

Use Unified Attribution Setting. The “unified attribution” setting will help users access data that includes iOS users. This number is higher than “account attribution”.

Use Account Attribution Setting. The “account attribution” setting uses what is currently set within your Meta Ads account, but due to mismatches, this number is lower than when “unified attribution” is used.

Please note, that due to attribution mismatches across campaigns, it may take some trial-and-error to dial in exactly what works for matching your Hawke AI data to what you use in Meta Ads Managers and/or other reporting platforms.

LinkedIn Ads:

• In Hawke AI, navigate to Reporting –> LinkedIn Ads
• In the LinkedIn Ads Campaign Manager, navigate to Campaigns and ensure that there are no active filters
• Ensure the same date ranges are selected
• Compare the metrics from the top of the LinkedIn Ads table (see “# campaigns” row) to the values shown at the top of the Hawke AI page
• Potential issues:
  • Conversion-based metrics (Conversions, Conversion Rate, CPA) do not match – this is likely conversion lag due to the conversion window set in LinkedIn Ads (conversions may continue to increase in LinkedIn Ads, while Hawke AI does not monitor for these changes after 24 hours)
  • Hawke AI contains a subset of your LinkedIn Ads campaigns due to selective importing of campaigns during Hawke AI setup

Microsoft Advertising:

• In Hawke AI, navigate to Reporting –> Microsoft Advertising
• In Microsoft Advertising, navigate to All Campaigns –> Campaigns and ensure that there are no active filters
• Ensure the same date ranges are selected
• Compare the metrics from the bottom of the Microsoft Advertising page (see “Overall total – # campaigns” row) to the values shown at the top of the Hawke AI page
• Potential issues:
  • Conversion-based metrics (Conversions, Conversion Rate, CPA) do not match – this is likely conversion lag due to the conversion window set in Microsoft Advertising (conversions may continue to increase in Microsoft Advertising, while Hawke AI does not monitor for these changes after 24 hours)
  • Hawke AI contains a subset of your Microsoft Advertising campaigns due to selective importing of campaigns during Hawke AI setup

TikTok Ads:

• In Hawke AI, navigate to Reporting –> TikTok Ads
• In TikTok Ads, navigate to Campaigns and ensure that there are no active filters
• Ensure the same date ranges are selected
• Compare the metrics from the bottom of the TikTok Ads table to the values shown at the top of the Hawke AI page
• Potential issues:
  • Hawke AI contains a subset of your TikTok Ads campaigns due to selective importing of campaigns during Hawke AI setup

Pinterest Ads:

• In Hawke AI, navigate to Reporting –> Pinterest Ads
• In Pinterest, navigate to Campaigns and ensure that there are no active filters
• Ensure the same date ranges are selected
• Compare the metrics from the bottom of the Pinterest Ads table (see “Results from # campaigns” row) to the values shown at the top of the Hawke AI page
• Potential issues:
  • Hawke AI contains a subset of your Pinterest campaigns due to selective importing of campaigns during Hawke AI setup

FAQ

Q: All of my data matches, except for conversions. Why is that?
A: While the ad platforms that Hawke AI connects with finalize almost all metrics within 24 hours (72 hours with Google Ads), conversions are often the exception due to conversion window settings within the connected platform. The number of conversions shown in the ad platform can, and often do, increase over time as a result of the conversion window set for the individual campaign.

Q: Why does my Google Ads Search Impression Share not match between Hawke AI and Google Ads?
A: In Google Ads, search impression share is noted as “the impressions you’ve received on Google search sites divided by the estimated number of impressions you were eligible to receive.” Given the estimated nature of eligible impressions, and the need to aggregate this daily, it is possible that Hawke AI may report data that is slightly different from Google Ads. Given the nature of this metric, as it relates to strategic decision making, this difference is negligible.

Q: Why does real-time (or live) data not match?
A: In some instances, the real-time components of Hawke AI may not exactly match what you see as live data in the connected platform. This is a result of the lag in the connected platform updating the data available via their API. The notion of ‘real-time’ connectivity can often more accurately be thought of as ‘as close to real-time as possible’.

Q: Does Hawke AI factor in credit notes or invalid activity refunds?
A: It is common for ad platforms to credit business back when they identify invalid or potentially-fraudulent ad clicks. Ad platforms consciously make these adjustments at the billing level and do not adjust already-finalized campaign metrics. As such, Hawke AI does not factor in these credit notes in any report or feature.

Q: Where and how often is information updated for each feature in Hawke AI?
A: Each feature within Hawke AI contains a subheading (see example below) that shares information relating to when the data/insights were last refreshed, how it is updated, and links to helpful resources (resource guide article and feature video).