When AI Gets the Data Wrong: Why First-Party Sources Matter More Than Ever

This AI Overview is reporting a false metric because LinkedIn’s DAU is not public information (otherwise we’ll see a troubling stat + lots of automated bots and accounts).

Heading to the Social Shepherd article, it claims to be a 2025 study, but it quotes Statista.

The link to Statista is a 2021 research claiming the following:

“As of March 2021, 17.1 percent of Android LinkedIn app owners in the United States accessed the job searching app on a daily basis.”

(The chart is showing 16.2% instead of 17.1%, which is what the AI allegedly used to source and divide by the total reported users 4 years later).

Funny enough, the Statista piece is authored by Stacy Jo Dixon, and one of the topics she explores is “online disinformation”. Hallucinations become more of a problem nowadays as soon as data isn’t present in the top 1 layer.

The excessive fake accounts/bots + virality are the core reason I spend 90% less time on social over the past year.

And my focus on first-party data – newsletter + professional Slack groups – is further backed by all the fake news that even Google consistently provides today.

graphical user interface, text, application, email


Part of:

Mario Peshev is a 5x CEO and operator, founder of DevriX and Growth Shuttle, global value creation advisor, angel investor, and author of “MBA Disrupted.”

His original background in engineering rode the wave of IT entrepreneurship in the last 25 years, from product and service entrepreneurship through acquiring and selling businesses, to investing in global startups like beehiiv, doola, the Stacked Marketer, Alcatraz, SeedBlink.

Peshev spent over 10,000 hours in consulting and training contracts for mid-market and enterprise organizations like VMware, SAP, Software AG, CERN, Saudi Aramco since 2006. His books and guides are referenced in over 50 universities in North America, Europe, and Asia.


Follow Mario on social:

Latest Editions:

Latest Answers: