As a consultant myself, I can say you're reasonably accurate with your assessment. Often times it works by a consulting firm beautifully presenting to non-technical execs at a company about how they can turn the company's data into money. Lots of over-confidence and some nicely "massaged" benchmark figures.

If said company hires the consulting firm, then the firm staffs the project with people who basically have to figure out how to build what was sold...typically in a short time, with a new team, none of whom understand the company's data landscape and the nuances associated with the company. You might be thinking that this is a recipe for disaster. You'd be right. It's why job satisfaction for data scientists and ML engineers in consulting is super low. But success stories do exist when the stars align and a team is able to build things like recommendation engines, supply chain optimization models, conduct a good segmentation analysis, etc. for companies that don't have sufficient talent to execute these internally.