Making decisions on bad analytics data — here’s what that actually costs

The metric that convinced an entire industry to bet on the wrong thing

A single missed variable in marketing analytics can do more than distort a report. It can change budgets, reshape strategy, and push an entire industry toward decisions that the underlying audience data never justified.

For example, between 2015 and 2016, Facebook calculated one of its key video-advertising metrics – average time watching a video – using a method that excluded any view shorter than three seconds from the calculation entirely. That resulted in pushing the average view time up, making video audiences look far more engaging than they were.

Facebook acknowledged in 2016 that the metric had been overstated by 60% to 80%. Advertisers who later sued alleged the inflation was far higher — as much as 900% in some cases — and argued that the inflated numbers had already influenced buying decisions before the issue was fully disclosed.

The effects spread well beyond Facebook’s ad product. Facebook could charge more for video placements, while parts of the media industry reoriented around the audience the data appeared to show: cutting writing staff, hiring video producers, and rebuilding strategies around video growth. The results did not match the promise. Fox Sports’ video audience fell sharply over that period, and several publishers, including Vox, later reversed course and rehired writers after the video strategy failed to deliver. Advertisers who bought video placements based on the inflated figures filed a class action lawsuit, which Facebook agreed to settle for $40 million in 2019.

Nobody set out to make a bad decision. But one mismeasured metric was enough to reshape media strategy for years. That is what makes bad analytics data dangerous: it does not merely create a reporting problem; it creates confidence in the wrong decision. And the Facebook case was not an isolated exception.

That pattern matters because most bad data does not arrive as a scandal. It usually appears as a normal dashboard, a plausible report, or a metric no one has time to challenge. The cost begins when teams treat that incomplete picture as operational truth.

Adverity’s latest study found that 45% of the data marketers use to make business decisions is incomplete, inaccurate, or out of date. The same study found that 43% of CMOs believe less than half of their marketing data can be trusted.

What the numbers say: the baseline cost of bad marketing data

Adverity’s 2025 “Fixing the Foundation” research suggests this is not a niche problem. The 45% inaccuracy figure and the 43% distrust figure held up regardless of company size, sector, or region. When CMOs were asked what would most improve marketing performance, they ranked data quality above workflow automation and better data access.

The core problem is fragmentation. As MarTech’s 2025 survey found, the average marketing technology stack now runs 17 to 20 separate platforms – ad platforms, GA4, CRM, email tools, a CMS, and more. Each tool can define a “conversion” or a “lead” differently, and those definitions often do not agree with one another. That is not a glitch. It is the default state.

Five specific ways bad tracking data destroys marketing value

Five recurring tracking failures tend to damage marketing value by making the same basic question harder to answer: which activity actually created revenue?

1. Duplicate or missing conversion events make campaigns look better or worse than they really are, so budget gets moved to the wrong places.
2. Attribution blind spots hide which channels actually influenced a sale, so marketers may undervalue important touchpoints or overinvest in the last visible click.
3. GDPR consent mode and privacy-driven data loss reduce the amount of trackable user data, making reports less complete and decisions less certain.
4. “Unassigned” channel in analytics reports means traffic cannot be properly linked to a source, so performance by channel becomes unclear.
5. A site redesign, replatform, or CMS migration can break tracking setup, causing lost historical continuity and unreliable post-launch performance data.

Together, these issues lower marketing value because they make performance data unreliable. When conversions, channels, and customer journeys are measured incorrectly, marketers cannot clearly see what is working, what is wasting budget, or where growth is actually coming from. The damage then moves from the spreadsheet into the meeting room: teams spend the first few minutes arguing over what GA4 says versus what the ad platform says versus what the CRM says, before any strategy is discussed. Over time, that slows decisions, weakens confidence in reporting, and creates the sense that the analytics investment is not paying for itself — which, in a very real sense, it is not if nobody trusts what it is telling them.

Modern bidding algorithms vs. manual reporting – how can it turn into budget-wide problem?

In manual reporting environments, one bad data point usually corrupts one report. Someone can notice the issue, contain it, correct it, and move on. Modern bidding systems are different. They use the data they receive to make decisions continuously and automatically, replacing many of the manual judgments marketers used to make. That works well until the system starts optimizing against corrupt, incomplete, or restricted data — whether the cause is a tracking failure, a consent-mode gap, or another missing signal.

The system then learns from the incomplete picture it has been given. It keeps finding more of the events that appear valuable inside that flawed dataset, even when those events do not reliably drive revenue. The result is a false sense of optimization: the campaign may look smarter while the budget is drifting further away from real business outcomes.

Where this hits hardest: paid budgets, SEO priorities, and lead pipelines

Three parts of a business absorb most of the damage from bad tracking data, and each fails in a slightly different way.

• Paid media budgets take the most direct hit: misattribution reallocates spend toward whichever channel merely looks efficient on a flawed model, not the one that’s actually driving revenue.
• Technical SEO decisions take a subtler hit: Core Web Vitals has two different data sources — lab data generated instantly in a controlled test, and field data, drawn from real visitors and updated on a rolling 28-day basis through Google’s Chrome User Experience Report. Teams that chase a lab-data score can spend real development time “fixing” a page that field data shows isn’t actually a problem for real users, while a genuine field-data issue goes unaddressed for weeks because the lagged, page-group-level view in Search Console doesn’t surface it clearly.
• Lead and revenue pipelines take a third kind of hit: automation and sales follow-up built on leads that were double-counted, mis-tagged, or silently dropped at the tracking layer — covered in the automation section — quietly misdirect sales effort toward the wrong accounts and away from real ones.

All three of these frequently trace back to the same root: a technical implementation issue at the site level. A redesign, a new page builder, a platform migration — the moment the website itself changes is the moment tracking most often breaks, which is why analytics data quality and site build quality are rarely separate problems in practice.

Consider a rough example. A business spending €8,000 a month on paid media with even a conservative 10% attribution gap is misdirecting roughly €9,600 a year in ad spend alone. If one in ten real leads never fires a tracked conversion event, and each lead is worth even a modest amount in eventual revenue, that is pipeline quietly disappearing — not “waste” in the traditional sense, but revenue nobody ever knew existed. Add a few hours a month of reconciliation time across the team, and the total, even before any SEO effort is counted, can regularly reach five figures a year for a business with a modest marketing budget.

Where to start fixing it without a full analytics overhaul

When a tracking problem surfaces, the instinct is often to plan a full re-platform or ground-up analytics rebuild. That is usually the wrong first move. Most businesses need triage before they need a project: identify the handful of events that actually matter for revenue decisions — purchases, leads, bookings, and key form submits — and get those right before touching everything else.

The highest-leverage fixes are not complicated. Audit tag firing on the revenue events first, and fix duplicates or gaps before anything else. Agree on one definition of “a lead” or “a conversion” and enforce it across GA4, ad platforms, and the CRM, so the same word means the same thing in every system. Put basic monitoring on the handful of events that matter, so a broken tag gets caught the same week it breaks instead of eight months later. None of this requires a governance committee or a platform migration. It requires someone to look, a shared definition of what matters, and a plan to keep looking.

Contact us and set up analytics that show which channels actually drive sales.