Nobody decides to believe a lie about their data. The lies arrive pre-installed: they ship with the tools, the dashboards repeat them, and everyone in the meeting nods because everyone else is nodding. I spend my working life inside companies' analytics, and the same five beliefs cause more bad decisions than every tracking bug combined.
Lie 1: "Our numbers are accurate"
Your analytics undercounts. Not might; does. Ad blockers stop the tracking script for a chunk of your visitors. Consent banners remove another slice, legitimately and permanently. Browser privacy features trim cookies and shorten memory. Then implementation drift, the quiet bugs of every redesign, subtracts its own unpredictable share.
None of this makes the data useless. It makes the data a sample, and samples are fine when you know their bias. The failure mode is not the gap; it is running a business as if the gap were zero, then reorganizing the marketing budget over a 4% "drop" that was actually a consent banner update. The full argument, including what accuracy is achievable, is in is 100% data accuracy possible?
Lie 2: "The ad platforms report real conversions"
Meta says 120 conversions. Google says 90. Your shop processed 130 orders total, and some had nothing to do with ads. Everyone has seen this table, and most companies resolve it by quietly trusting whichever platform's number justifies the current strategy.
The mechanics are not mysterious: each platform grades its own homework, using its own attribution window, counting view-through conversions the others cannot see, and each takes full credit for shared customers. Adding platform-reported conversions together counts the same order multiple times, by design. I broke the whole reconciliation down in why GA4, Facebook and your store disagree on purchases. The short version: platform numbers are optimization signals, not accounting.
Lie 3: "More dashboards means more clarity"
Clarity is a property of decisions, not of screens. A company with forty dashboards and no metric dictionary does not know more than a company with five defined numbers; it knows less, because contradictions multiply with surface area and each contradiction burns trust and meeting time.
The tell: when a number looks wrong, does anyone know which of the forty views is authoritative? If the answer is a shrug, the dashboards are scenery. That pathology, and the definitions-first fix, is its own essay.
Lie 4: "Attribution tells us what caused the sale"
Attribution models distribute credit; they do not discover causation. Last-click hands the trophy to whoever touched the customer last, usually brand search, which would have happened anyway. Data-driven models are better but remain redistribution formulas over observed touchpoints, blind to word of mouth, podcasts, and the colleague who recommended you at lunch.
The dangerous version of this lie is spending as if the model were physics: cutting a channel because "attribution shows nothing," when the model simply cannot see what the channel does. Attribution is a map drawn by someone who has never left the road. Useful. Not the territory.
Lie 5: "Industry benchmarks say we should be at X"
That "average conversion rate for e-commerce" figure circulating in your strategy deck: trace its lineage sometime. It is typically a vendor's blog post aggregating self-selected customers of one tool, across countries, price points, and traffic mixes that have nothing to do with yours. Benchmarks flatten every variable that makes your business your business, and then executives set targets against the flattened fiction.
The honest replacement costs nothing: your own trend. Your conversion rate this quarter versus last, same definition, same measurement. You versus you is the only benchmark with clean provenance.
What believing the truth looks like
None of the five corrections requires new software. Know your measurement gap and its direction. Treat platform numbers as biased instruments. Define ten metrics and delete the rest. Read attribution as allocation, not causation. Benchmark against yourself.
Companies that internalize these five run on fewer numbers than their competitors and trust them more. If you want to know which of the five your company currently believes, the audit is one uncomfortable meeting long: put your shop's order count, GA4's purchases, and the platforms' claimed conversions on one slide, and watch the room discover they describe three different realities. What happens in the ten minutes after that slide is, in my experience, the most honest analytics conversation a company has all year.