EVIDENCE BASE loading…

Change Advisor · A living, versioned instrument

What is AI worth to your company — and what makes it true?

Most ROI calculators give you one flattering number and no provenance. This one gives you a defensible range, the published evidence behind every figure, and a quarterly versioned history you can audit. Two lenses, AI investment and people investment, because the research says the returns live in the pairing.

Your company

Defaults reflect a 100-person mid-market company, the worked example from the What Actually Returns series. Every output is a range, on purpose.

Global baseline: only 20% of employees are engaged (Gallup, 2026). Mid-market firms routinely have 60+ points of headroom.

Capped at 30% on principle: no peer-reviewed causal return exists yet for financial-wellbeing programs. The output labels this slider directional, and that restraint is the point.

Want to present this?

Your numbers generate a board-ready briefing: the investment case, the architecture gap, and every figure cited to its source.

The link reproduces your exact settings, so you can share it with a colleague and build consensus before the meeting. Nothing is stored on our servers.

The defensible number

Uplift is applied only to AI-addressable work, never to total payroll.

Show the math & sources

      
This model estimates productivity-denominated value and selected human-capital effects. It does not capture revenue innovation, risk reduction, customer-experience gains, or implementation-quality variance. That is precisely why the number is a conversation starter, not a guarantee. Methodology and full citations: see the evidence ledger below.

    

  


  
  

    
    

      
Help answer what AI and people investment actually return

      
I'm building a study from the organizations doing the work, so this calculator keeps pace with what they're seeing on the ground. Take part and you'll see the findings first, with early access to the research report when it launches. Add your email and I'll reach out when the first survey opens.

      

        

          

            
            
          

          

            
            
          

        

        
        
        
        

          
          
          
        

        
        
        

        
Your email is only used to reach you about this research and the report. No marketing lists, and you can opt out anytime. Nothing you enter in the calculator is tied to it.

      

    


    
    

      
Seen research we should know about?

      
This evidence base is refreshed quarterly, and good sources come from everywhere. Submit a study, article, or dataset for consideration. Every submission is human-reviewed and held to the same verification standard as everything else on this page. Peer-reviewed work moves fastest.

      

        

          

            
            
          

          

            
            
          

        

        
        
        
        

          
          
        

        
        
        
        
        
        
Your name and email are used only to follow up about this submission. No lists, no marketing. Submissions do not enter the evidence base until independently verified against primary sources.

      

    

  


  
  

    
The next step

    
See where your organization actually stands

    
The calculator shows the size of the prize. The AI Transformation Diagnostic shows you where the architecture gap sits in your own organization, and what to do first.

    Decide what to change first →
    Take the AI Transformation Diagnostic →
  


  
  

    
The evidence ledger

    
Every figure in the calculator traces to a published source, graded by tier: T1 peer-reviewed experiments · T2 large-N institutional research · T3 secondary, flagged. Contested findings stay in, and stay labeled. Organized by the five pillars of the AI Transition Maturity Model.