On-Demand Webinar · 1 hr 2 min

DataOps Therapy with DataKitchen's Founders

DataKitchen's three founders, Eric Estabrooks, Gil Benghiat, and Chris Bergh, on the results of their 2021 data engineering survey, why data engineers burn out, and what DataOps can do about it. Recorded December 2021; updated August 2026.

Presented by Eric Estabrooks, Gil Benghiat, Chris Bergh

What you'll learn 6 points
  • A survey of 600 data engineers found 97 percent burnt out, and 78 percent wishing their job came with a therapist to help manage the stress.
  • The named causes are mostly relational rather than technical: 63 percent blame and finger pointing, 50 percent the relentless flow of errors, 50 percent manual processes crowding out innovation, 50 percent disruptions to work-life balance, and 49 percent constantly playing catch-up with stakeholder requests.
  • The surrounding numbers are worse than the mood: Gartner puts data analytic project failure at 85 percent, and New Vantage Partners recorded data-driven organizations falling from 37 percent to 24 percent, against an average CDO tenure of 2.5 years.
  • The founders' answer is to treat stress as an output of the system: don't suffer, build a system, and automate rather than agonize.
  • 'Done' is defined as live in the customer's hands, which removes the ambiguity that generates most rework and most late-stage surprises.
  • Heroism is explicitly discouraged. Making heroism a rare event is the goal, because a process that depends on individual rescue cannot be staffed sustainably.

Slides

14 slides

Questions from this session

Why frame a DataOps discussion as therapy?

Because the survey findings are about distress rather than tooling. 97 percent of 600 data engineers reported burnout and 78 percent wished their job came with a therapist. When the leading causes are blame, error floods, and constant catch-up, the useful conversation is about what the working system is doing to people, not which product to buy.

What does the data say about data analytics projects overall?

Gartner puts outright failure at 85 percent. New Vantage Partners recorded the share of firms describing themselves as data-driven falling from 37 percent to 24 percent, while average CDO tenure sits at 2.5 years. The pattern is short-tenured leadership attempting long-cycle change.

What are the core DataOps principles named here?

Don't suffer, build a system. Don't agonize, automate. Run toward errors. 'Done' means live in your customer's hands. Don't be a hero; make heroism a rare event. Don't put your head in the sand — focus on value delivery.

Why is 'don't be a hero' a principle rather than a criticism?

Because heroism is a symptom of an unreliable process, and rewarding it entrenches the unreliability. If a release only ships because someone stayed until 2am, the organization has learned that the process works, when what actually worked was one person's unpaid overtime. Making heroism rare means fixing what required it.

Does automation solve burnout on its own?

No. Two of the five leading causes — blame and finger pointing at 63 percent, and restrictive policy — are cultural, and automation does not touch them. Automation removes the manual toil and the error flood; the blame has to be addressed by changing how the team responds when something breaks.

Where to go next