# Using AI in marketing and advertising: Check the data your tools receive first

AI tools base their advice on your analytics and ad platform data, which is often incomplete. Run these 5 data checks before you act on AI recommendations.

![Illustration of orange and green data streams flowing from a laptop across a dark blue background](https://taggrs.io/_astro/hero.D10TLQRe_ZMXkXE.webp)

Source: https://taggrs.io/give-ai-better-data/
Author: Małgorzata Poddębniak, Content Marketer · [LinkedIn](https://www.linkedin.com/in/malgorzatapoddebniak/)
Published: 2026-11-20

---

Marketers increasingly connect AI tools to the platforms they work with to find out which campaigns perform best and where to add budget. Since every team can use the same AI, the advantage lies in the data it works with. Your website sends that data to every tool you connect, from GA4 to Google Ads, Meta, TikTok and LinkedIn, and part of it often gets lost along the way.

## Why ad platforms miss conversions: Ad blockers, ITP and cookie banners

Your shop system records every order, but tools such as GA4, Google Ads and Meta often receive data on only around 60% of them. For the other orders, the tracking data gets blocked or lost in the visitor’s browser, so those conversions are missing from the platforms’ reports. There are three main causes:

- **Ad blockers** stop tracking scripts and requests. [About 30% of internet users worldwide use some form of ad blocker](https://backlinko.com/ad-blockers-users), and many of these tools block analytics and conversion requests along with ads. When an ad blocker prevents your Google Tag Manager container from loading, every tag inside it fails as well, including GA4, the Meta pixel and the Google Ads conversion tag.
- **Browser restrictions**, such as Safari's Intelligent Tracking Prevention (ITP), shorten the lifetime of cookies set by scripts, in many cases to as little as 7 days. If more than 7 days pass between the ad click and the purchase, the platform may be unable to connect the order to that click.
- **Consent** decides whether client-side tags fire. A visitor who declines has made a choice you need to respect. A separate problem appears when an ad blocker hides the consent banner itself, since most consent platforms load from well-known third-party domains on filter lists. The visitor sees no prompt and the site doesn’t record their consent choice, which leaves a gap in both measurement and privacy compliance.

These losses are larger for some audiences than for others, which affects how an AI tool compares your campaigns. TAGGRS tests show that requests are blocked more often for technical audiences, such as developers, marketers and analytics professionals. As a result, campaigns aimed at those groups appear to perform worse in your reports, and an AI tool analyzing your account works from those same numbers. Automated bidding also tends to spend less on groups with fewer recorded sales, which can decrease their measured results even further. For more on this, read [why better measurement foundations matter in an AI-driven world](/taggrs-at-the-martech-summit-amsterdam-2026/).

## Modeled conversions in Google Ads and Meta

Conversions from visitors who decline cookies or from iOS users who opt out of tracking under App Tracking Transparency (ATT) are difficult for the platforms to measure directly. Google estimates part of them through [Consent Mode](/google-consent-mode-v2/), and Meta does it through its own conversion modeling.

The platforms base this on the conversions they could observe. When that number drops, a larger part of your reported total is estimated. In Google Ads, this only kicks in with advanced Consent Mode and at least 700 ad clicks over 7 days per country and domain. Modeled conversions are part of the totals in your reports, so an AI tool reading your account works with observed and modeled figures combined. Read more about [what conversion modeling is and whether you can trust it](/what-conversion-modeling-is-and-can-you-really-trust-it/).

Each platform also uses different attribution rules: the model, the time window and what counts as a conversion all vary. Meta credits purchases within about 7 days of a click and also counts view-through results, while Google Ads commonly uses a 30-day window. When you ask an AI tool to compare the platforms, part of the difference comes from these rules.

## 5 data checks before you act on AI recommendations

1. **Set a baseline from your order data.** Pull purchases and revenue from [Shopify](/shopify-server-side-tracking/), [WooCommerce](/woocommerce-server-side-tracking/), Stripe or your CRM for a fixed period. These are paid orders, so compare every platform number to them.
2. **Align your settings.** Use the same time zone and currency in every platform, and note each platform's attribution window. Google Ads records a conversion on the click date and GA4 on the purchase date, so the same sale can appear in different weeks.
3. **Check whether your consent banner reaches every visitor.** Compare the number of consent records in your consent platform with the number of sessions in your analytics for the same period. A consistent gap suggests the banner is being blocked. Serving it from a first-party subdomain keeps filter lists from matching it.
4. **Check Meta deduplication.** In Meta Events Manager, open your Purchase event and check how pixel and Conversions API events are deduplicated. A low rate means double-counted purchases or missing server events.
5. **Check the "Unassigned" share in GA4.** A growing share usually points to inconsistent UTMs, missing referral exclusions or consent gaps.

If your platform numbers are still well below your order data after these checks, it means that [ad blockers](/tracking-beyond-ad-blockers/), browser restrictions or a blocked consent banner are preventing some of your conversions from being tracked. Fixing it usually requires involving several teams: marketing uses the data, developers manage the website, the legal team owns the consent banner and an agency controls the tags. Naming one person or team that’s in charge of the full setup, from the website to the ad platforms, makes it easier to spot problems early and decide who fixes them.

## How to ask AI better questions about your marketing data

You can give your AI tool the context it needs to read your data correctly:

- Share your order data for the same period and ask it to compare conversions per campaign with real orders.
- Ask it to state which conversion action, attribution model and attribution window each number is based on.
- Ask it to flag sudden changes in conversions that line up with a website update, a new consent banner or a tracking change before it suggests budget shifts.

## How Server-side Tracking sends more data to the tools you connect

How many events reach your analytics and ad platforms through Server-side Tracking depends partly on where your server runs. If you run it on a subdomain of your own website, requests from the browser count as first-party. A TAGGRS-provided subdomain is also an option, but it makes the requests third-party, so not all benefits from the list below apply.

TAGGRS offers hosting for [server-side Google Tag Manager](/gtm-server-side-tagging-hosting/) and the [Meta Conversions API Gateway](/meta-capi-gateway-hosting/) and manages the servers for you. From the server, events go to platforms such as GA4, Google Ads, Meta, TikTok, LinkedIn, Pinterest, Snapchat, your CRM and other tools over a direct server-to-server connection.

- Requests to your own subdomain count as first-party, which ad blockers block less often than requests to third-party domains.
- Cookies set by your server avoid the 7-day cap Safari applies to cookies written by JavaScript.
- One consistent event ID per purchase lets Meta deduplicate browser and server events.
- On your own server, you can anonymize IP addresses, remove personal data before it reaches third-party tools and control what data each tool receives.

The request from the browser to your server often contains recognizable patterns, such as /collect in the path, and ad blockers can block it even when the script loads from your own domain. The TAGGRS [Enhanced Tracking Script](/enhanced-tracking-script-against-ad-blockers/) encrypts the full request from the browser to your server, so it no longer contains the patterns ad blockers look for. In tests with 30 partner organizations and on the TAGGRS website, this increased measured page views by 9.1% compared with standard client-side tracking, and by close to 30% for audiences of developers and marketers.

In the TAGGRS dashboard, the Tracking prevention impact metric shows the share of events that Safari's ITP, Firefox's ETP and similar browser features would have blocked in a client-side-only setup and that your server container recorded.

Server-side Tracking leaves each platform's attribution rules as they are, so Google Ads, Meta and GA4 will still report different numbers. What changes is how much of your data reaches each tool, including the bidding algorithms and AI tools that rely on it. Learn [why your Meta, GA4 and Google Ads conversions don't match](/meta-ga4-and-google-ads-conversions-dont-match/).

## Next steps: Free tracking audit and guide

To see where your website loses data, run the free [website tracking checker](/website-tracking-checker/). For a step-by-step look at where tracking breaks down and what to fix first, download the free guide [How resilient is your tracking setup?](/server-side-tracking-how-resilient-is-your-setup/)

When you're ready, [create a free TAGGRS account](https://dashboard.taggrs.io/register). Try Server-side Tracking on your own website and see how much more of your data reaches your analytics and advertising platforms.

An AI tool can point you to the campaigns that look strongest and save you hours of analysis. Before you change your campaign budgets based on its advice, make sure the conversions behind it reflect your actual orders. Using AI is now standard practice. Using it on complete data is what makes the difference.

## FAQ

**Can an AI tool tell modeled conversions apart from observed ones?**

Usually not. In Google Ads, modeled conversions are reported in the same Conversions column as observed ones. Consent Mode diagnostics show the modeled share only for a limited period after setup. An AI tool reading your account therefore sees one combined number. Your order data stays the most reliable baseline.

**What share of visitors accept cookies?**

In the EU, the average cookie consent rate was 68.8% between January and September 2026, according to the [TAGGRS cookie consent benchmark](/cookie-consent-rate-benchmark/). That means roughly one in three visitors declines, and their conversions can only be estimated.

**Does Server-side Tracking work around cookie consent?**

No. Consent choices apply in a server-side setup just as they do client-side. The server container can read the consent state attached to each event and forward only the data a visitor agreed to share.

**How much more conversion data does Server-side Tracking capture?**

On average, TAGGRS users measure 16–21% more conversions after implementation. The exact increase depends on your audience, their browsers and how many of them use ad blockers.

**How long does it take to see the effect of Server-side Tracking?**

Conversions can look lower in the first week or two, because the new setup starts without historical data. For a reliable comparison, check platform conversions against your order data 30 and 60 days after implementation.
