Measurement is only useful if the data is right

Every conversation about activating data, feeding it into smart bidding, building reports from it, or putting AI on top of it rests on one assumption: that the data is accurate. In our experience, that assumption needs to be tested. Data quality is earned, never inferred. So we built a way to check it, regularly.

The Measurement Grader is our proprietary measurement health check platform. It runs 291 checks across six areas of a measurement setup: Analytics Health, Tag Management, Consent & Compliance, Signal Quality, Data Enrichment and Measurement Durability. It connects to the GA4 and Google Tag Manager APIs to read the property and container directly, and it dispatches browser bots to the site's live pages to inspect every network call and test how the consent setup behaves when a visitor accepts, rejects or ignores the banner. It then grades what it finds. We built the decision logic behind those checks from thousands of audits our consultants carried out by hand over the past decade. What used to take a senior analyst a week of discovery now takes minutes, so the senior analyst can start with diagnosis instead. 

We run the Measurement Grader when we engage with a prospective client, and we run it systematically across every client under our fractional data team SLAs, so that data quality and hygiene are maintained continuously. That is what has changed the conversation with our clients. When we continuously verify tracking, we spend our time focusing on data utilization, not on finding out why it is wrong.

Since June, we have run hundreds of Health Checks. This report pools the first 100 Checks run on organizations we hadn't worked with before, so it presents the measurements as we find them, not after our team has reviewed them. It answers the question we hear most often: Can we trust our data, and are we meeting industry best practices? 

The six issues we find most often

The same six problems appear across almost every sector. Each percentage represents the share of organizations to which the check applied.

1. Consent that doesn't hold (72% had at least one failure). On half of sites (52%), tracking fired before the visitor answered the banner. On 61%, marketing cookies were set after the visitor clicked Reject. 17% had no consent platform at all. Where one was present, 29% collected the visitor's choice but never passed it to the tags.
2. Conversions with no value (85%). GA4 recorded that a conversion occurred but not what it was worth. Bidding and ROI reporting then treat a $20 sale and a $2,000 sale the same way.
3. Key events that never reach the ad platforms (92%). The conversions marked in GA4 weren't what each ad platform was optimizing toward, so each platform chased its own definition of success.
4. Enhanced Conversions switched off (94%). Google Ads conversion tags ran without hashed first-party data, so a conversion was lost whenever the cookie was gone. Safari and most privacy tools routinely clear cookies.
5. Everything sent from the browser (73% had no server-side tagging). Ad blockers, browser privacy features and short cookie lifetimes erode browser-only data, and nearly every Google Ads setup we reviewed relied on it entirely.
6. Tracking that slows the site (87% exceeded our page-load threshold). On 83% of sites, third-party scripts blocked the page from rendering. The tags intended to measure conversions were costing some of them.

Missing tags are rare. The tags, the GA4 property and the container are almost always in place. The problems are in the configuration: tags firing without consent, conversions sent without values, and signals that leave the browser and never arrive in your platforms. Configuration problems can usually be fixed in a few weeks.

The headline numbers

Two in three organizations (68%) had at least one critical issue, and for 95% of those, the critical issue was consent: tracking that fires before the visitor agrees, or continues after the visitor says no.

A single critical issue caps a site's overall score at 65, which is why scores cluster at 65, 55 and 45. The typical organization scored 65 out of 100, with a range from 45 to 86. It also had six critical or high-priority issues across several parts of its setup.

By area, Google Analytics Health was weakest (median 55), followed by Google Tag Management (57), Consent & Compliance (65) and Measurement Durability (67).

What this looks like in your sector

Non-profits, B2B SaaS and financial services scored highest. Higher education and hospitality scored lowest. Every sector shares the same six weak spots, though in a slightly different order, so the sections below cover what is specific to each sector.

Ecommerce

Typical score: 65  |  Weakest area: Analytics Health  |  Critical issue: four in five

Ecommerce has the most at stake when it comes to accurate data, and its weakest point is the link between GA4 and the ad platforms. Wherever we checked, purchase events were not reaching every ad platform the store pays for. Nearly every store also had product IDs and names that did not match one-to-one in GA4, which breaks product-level reporting. Smart bidding on value, product feed optimization and margin reporting all depend on that link working.

Lead generation

Typical score: 65  |  Weakest area: Analytics Health  |  Critical issue: most

Lead generation sites most often lose track of the source of their leads. Referral visits were recorded as direct, so the channels that brought qualified leads were under-credited, while those that brought noise were not. For businesses that close offline, the second gap is that the CRM outcome never flows back to the platform that generated the lead. Close that loop, and the platforms bid on qualified leads, not form fills.

B2B SaaS

Typical score: 75  |  Weakest area: Tag Management  |  Critical issue: a minority

SaaS setups were among the cleanest we checked, and the gap was further down the funnel. Offline conversions, including booked demos, trial conversions, and closed deals, never reached Google Ads. The platform optimized for form fills, while the business measured pipeline. Send closed-won back to the platform and it optimizes toward pipeline.

Financial services

Typical score: 74  |  Weakest area: Measurement Durability  |  Critical issue: a minority

A recurring pattern was that direct traffic converted better than paid. That is usually a sign of lost attribution rather than a loyal customer base. Paid clicks that convert after the cookie has expired are recorded as direct, and paid media is judged on an incomplete picture. Durable measurement restores that attribution, so paid media is judged on the conversions it drove.

Healthcare

Typical score: 65  |  Weakest area: Consent & Compliance  |  Critical issue: all

Healthcare had the joint-lowest Consent & Compliance score of any sector, at a typical 49. In a sector where data is most sensitive, consent is the first thing to fix. Done properly, it also improves data quality, because Consent Mode can model what declined visitors would have done instead of discarding them. Once consent is right, the rest of the measurement can be trusted.

Media and publishing

Typical score: 65  |  Weakest area: Analytics Health  |  Critical issue: all

Media sites carried the heaviest third-party script load we saw, slowing every page, and shared healthcare's consent score of 49. For publishers, the two are linked. Ad tech added over the years without governance causes both the page weight and the consent failures, so bringing that stack under control fixes both.

Hospitality and travel

Typical score: 64  |  Weakest area: Analytics Health  |  Critical issue: all but one

Hospitality sites were among the slowest we tested. Every one exceeded our page-load threshold, largely due to third-party scripts. Their Google Ads conversion tags ran without Enhanced Conversions, so bookings made after a cookie was lost went unmatched. In a sector with long consideration windows, that is a large share of bookings. Enhanced Conversions and durable tagging attribute those bookings to the campaigns that drove them.

Higher education

Typical score: 60  |  Weakest area: Measurement Durability  |  Critical issue: most

Higher education scored lowest overall. Several universities had invested in server-side tagging, but their Google tags still sent data directly from the browser, bypassing the infrastructure. They paid for the infrastructure, yet the data didn't use it. Routing the Google tags through it is a configuration change.

Non-profit

Typical score: 81  |  Weakest area: Tag Management  |  Critical issue: a minority

Non-profits scored best, with critical issues in a minority of organizations. What stood out was page speed and the absence of server-side tagging, which matters for sectors that rely on grant-funded Google Ads and need every conversion counted. Fix those two and the grant budget is optimized on complete conversion data.

Measurement Durability: what Safari does to your data

One of the six areas we grade is Measurement Durability: whether your data survives in the browser. Safari deletes cookies set by scripts after seven days, so a customer who returns after a week appears brand new, and the link to the ad they first clicked is gone. We compared Safari with Chrome across the same device types, iPhone vs Android and Mac vs Windows, so the gap reflects the browser, not the user.

• Returning visitors counted as new (one in four organizations). Safari records roughly 3% to 6% of returning visitors as new, and up to 10% in the worst case.
• Paid clicks recorded as direct (half of organizations where we could measure paid traffic). Safari sessions show 6 to 9 percentage points more direct traffic than Chrome, and up to 17 points. Paid clicks that convert after seven days lose their link to the ad, so Google Ads undercounts them and bids against an incomplete picture.
• Returning buyers: a smaller effect than expected. Where sites had sufficient purchases to measure, one in ten showed returning Safari buyers recorded as new due to cookie loss. On most other sites with a gap, the difference stemmed from how Safari users shop.

‍

These are estimates. About half of the affected sites use Advanced Consent Mode, which can produce a similar pattern, so we report a wider range for those sites. Two fixes close most of the gap: server-side tagging sets cookies from your own infrastructure, which Safari retains for longer, and Enhanced Conversions matches conversions without relying on the cookie.

From broken tracking to data utilization

This is the order our team follows for every SLA, and the order the Health Check prioritizes its findings. Each step makes the next more valuable, and the aim is data you can use, not a clean report.

1. Fix consent first. It carries compliance risk, so it comes first, and when done properly, it improves the data. When consent signals reach the tags correctly, Consent Mode models conversions from visitors who decline, rather than losing them. Google reports that this modeling recovers, on average, more than 70% of ad-click-to-conversion journeys lost to consent choices.
2. Give every conversion a value and send the same key events to every platform. Smart bidding can only optimize for what it is told a conversion is worth. This is usually the quickest win for media performance. For businesses that close offline, Google reports a 26% average increase in incremental ROAS from connecting offline and app data through Data Manager for Search campaigns bidding to value.
3. Turn on Enhanced Conversions and move to server-side tagging. This protects the conversions you already record from ad blockers and cookie loss. Google reports an 8% incremental ROAS from Enhanced Conversions for advertisers bidding to value on Search, and a 14% average conversion uplift from adding Google tag gateway. Pinterest reports a 24% average increase in attributed conversions when its Conversions API runs alongside the Pinterest tag.
4. Cut the scripts slowing your pages. Fewer render-blocking third-party scripts mean faster pages, and faster pages convert better.

The figures above are the platforms' own published averages, not results from our data, and each applies only under the stated conditions. We do not assign a dollar figure to any of this ourselves, because the gain depends on your spend, your traffic and which of these issues you have. That is what a Measurement Health Check answers for a specific site.

Once those four steps hold and continue to do so, as the Check is rerun, the work changes. Reporting can be trusted, bidding has an accurate signal, and the conversation shifts to what the data can do.

Methodology

This report aggregates the first 100 Full Measurement Health Checks completed between June 1 and October 13, 2026, for organizations' first Check. A Full Check connects to the organization's GA4 property and Google Tag Manager container through their APIs, and runs browser bots against the site's live pages to record network calls and test consent compliance under each consent state.

• First Check only. Organizations already on an Adaptive SLA are excluded, so the figures reflect the measurements as we find them.
• One result per organization. Where an organization has several sites, we used its most recent Check, so no organization is counted twice in the same sector.
• Complete Checks only. Organizations whose Check could not fully crawl the site are excluded because their consent, tag and durability results would be missing.
• Scores recalculated with our current scoring rules, leaving out a small number of checks still in validation.
• Sectors assigned by us, based on what each organization does. Sectors with only a few organizations are described in words rather than exact figures. A small number of construction and professional services firms appear in the overall figures but not in the sector sections.
• Percentages are the share of organizations where a check applied. A purchase-tracking check counts only sites that sell online.
• Who is in the sample: Organizations that requested a Full Check. They chose to look closely at their measurement, so read this as a picture of organizations that take measurement seriously, not of the web as a whole.
• Anonymous by design. No result in this report can be attributed to an individual organization.

See where your setup stands

A Measurement Health Check grades your site across the same six areas, lists every issue by severity, and tells you what to fix first. A Full Check includes your GA4 and Tag Manager setup and takes a few minutes to connect. Request one, and we will run it with you and walk through what it finds. What you get is a prioritized fix list for your own setup. If you want it fixed, and kept fixed, that is what our fractional data team does.

Request your free Measurement Health Check