Invest1 publisher3 min readPublished
The Predictive Index asked CEOs and managers questions too different to compare
The July survey behind the AI coaching pitch reports CEO impressions and manager self-reports as one gap. Its most useful figure is how few managers say they want to prepare for a hard conversation unaided.
The Investor · Invest desk

What happened
- The July survey behind the AI manager-coaching pitch was fielded by The Predictive Index, which sells Obi, a text-based chatbot for role-playing, coaching and preparing managers for difficult conversations.
- Only 20% of managers said they prefer to prepare for a difficult conversation on their own, without a framework, a coaching guide or input from HR.
- The survey named the top manager blind spots as delivering criticism constructively, anticipating an employee's reaction and staying objective.
- The stated goal of the tools is not to automate the people process but to let a manager practice delivering hard messages and simulate how a person might react in a specific scenario.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- contradiction A buyer who reads the gap as evidence that CEOs are overconfident about their bench is buying an interpretation the survey never tested, because the two figures come from separate populations answering separate questions.
- cost Round-the-clock access moves the cost of manager training onto the manager's own hours: the practice that used to happen in a scheduled session with a coach now happens at midnight on a Sunday, unpaid.
- exposure Where a tool ingests employee personality profiles to prime the simulation, preparation for a demotion or a layoff starts running through a third-party system with named-employee data attached.
- decision With free tools alongside paid ones, the first HR decision is not what to spend but whether to standardise on one tool or let 500 managers each rehearse against whatever they found.
The two percentages came from different questions put to different groups. CEOs and business leaders were asked whether their managers are "very confident" handling tough people conversations on their own, and 74% said they are [1]. Managers were asked something narrower, and 42% said they know what they want to say but struggle with how to say it [2]. Subtract the second from the first and you get 32 points; add them and you get 116 [20][21].
That spread sets an impression of other people's composure beside a self-report about wording, so it does not record two groups disagreeing about one fact. The figure underneath it is sturdier: 80% of managers want a framework, a coaching guide or input from HR before they walk into the room [3][22].
Of the six people CNBC quoted, four work for companies that sell AI coaching or role-play tools, including the firm that ran the survey [23]. Posh sells a role-play trainer to businesses [6], LHH offers one through a subsidiary [7], and Xactly's Penny gives managers tailored scripts and real-time coaching guidance [8]. The academic in the piece argued for preparation itself. "Spontaneity is not a goal of a high-stakes work conversation," said Emily DeJeu, a professor of business communication at Carnegie Mellon's Tepper School of Business [9]. She said the tools are better than talking in front of a mirror because you actually get feedback [24].
The economic case is one sentence, and it comes from someone who sells human coaching. "Most companies aren't hiring 500 coaches for 500 managers," said Casey Schaffer, founder of Definitions Coaching and Consulting [11]. A firm with 500 managers never carried a 500-coach cost line, so a licence is measured against nothing and clears the bar easily. CNBC did not report prices for the tools or outcome data for managers who used them [25].
The feature list is where the liability sits. Megan Barbier, chief human resources officer at Xactly, said a manager can ask the tool to grade the delivery, give critical feedback on the message, or flag when the conversation moves into risky topics from a legal perspective [16]. The drill is also only as good as the scenario the manager thinks to type, since these tools want specific parameters and context, down to a sensitive employee who is habitually late [26]. In my view what firms are buying is a cheap licence in place of manager training they never staffed. Two findings would change that. If managers were put the same "very confident" question the CEOs got and the gap survived, the 32 points would be a genuine disagreement [1]. If one vendor published results from managers who drilled, the strongest evidence in the file would stop being what managers say they prefer [25].
What to watch
- Whether LHH, Xactly or The Predictive Index publish per-seat pricing that can be set against a coaching engagement.
- Whether loading employee personality profiles into a vendor's role-play tool draws HR data-protection scrutiny.
- Whether the legal-risk flagging feature becomes a required step in termination workflows rather than an option.