Is Microsoft 365 Copilot too expensive?
Is Microsoft 365 Copilot too expensive? At $30/mo, it may more than double the cost of some existing Microsoft 365 licenses, and there seems to be a growing pool of advice out there as to how to be highly selective in who it is allocated to. But, if you're thinking about budgeting for Microsoft 365 Copilot, or ChatGPT Enterprise, or Duet AI for Google Workspace, what factors should you really be considering, what should you be concerned about?
Video transcript:
When Microsoft announced the pricing of Microsoft 365 Copilot back in July, a lot of people, including myself, were fairly taken aback, as that $30/mo price tag represents such a huge leap in cost for any of the base licenses it would be applied to.
However, as I have seen lots of content spring up with suggested approaches on how to decide who should get Copilot and who should not, I’m going to take a position that's somewhat contrary to all the advice I've so far seen. In my opinion, the right approach is to just buy it for everyone.
So you're probably thinking that I've gone crazy with such a suggestion, but keep watching and I'll tell you why I think it's the right answer, how businesses should really be thinking about the cost of these tools, and what, ultimately, the cost to your business might be if you decide to take a cheaper approach.
Before we dig in though, there's a little background we should be aware of.
The US Federal minimum wage is $7.25 per hour, so the very minimum any business is paying for a full time worker each month is about $1,250. And that’s just for the salary. If your business offers healthcare, retirement savings, or even with an allowance for paying employer taxes, this bill goes up. At $30/mo Microsoft 365 Copilot or its competitor Duet AI for Google Workspace represents less than 2.5% of pay or in terms of time 57 minutes per week.
But even jobs that were once firmly minimum wage jobs are no longer minimum wage. In many places you can make at least double this rate on your first day stacking shelves or flipping burgers, so you definitely don’t see many information focused office based jobs where you might be wondering about deploying Microsoft 365 Copilot advertised at $7.25 per hour.
According to ZipRecruiter the average US wage for an entry level office assistant role is $18.13 per hour. For this type of role a Microsoft 365 Copilot license represents slightly less than 1% of pay or 23 minutes of time per week.
Should any type of office-based worker be able to save 23 minutes per week using an AI tool like Microsoft 365 Copilot? Of course. And as you move through your company to look at more expensive workers, this number will come down radically. For a worker on the average US salary, according to the Bureau of Labor Statistics, of a little over $59,000 per year, a Microsoft 365 Copilot license costs a little over half of 1% of pay, and has its costs covered in less than 15 minutes of time savings per week.
But simply looking at time saving is problematic. According to the ADP Research Institute, workers on average work an extra between just over 5 and about 8.5 hours of unpaid overtime per week depending on age. This represents all those extra five minute tasks, evening emails, early morning phone calls, that project that just has to get finished before you can leave for the day. Companies are getting the benefit of all this extra time and there’s a strong likelihood any AI efficiency gains will accrue back to workers by reducing this time, not actually adding to the bottom line.
So, in a work culture where workers saving time isn’t going to automatically generate more output, because they are generally expected to deliver off-the-clock output for free, what measure might appeal in board rooms across the country?
Productivity is the measure of outputs versus inputs. So if we put $30 into a Microsoft 365 Copilot license, and that input generates $300 in outputs, that’s not a bad deal. But if it generates $30 or $0, that’s really not so good. The problem is that while in a macro sense we are great at measuring productivity, on a micro or individual worker level we are horrible at it. For some types of roles it’s more clear cut, but for many it’s really tricky.
Last year, the New York Times ran an interesting article looking at the issue of the modern productivity score - there's a link below if you want to dig into this. The examples given were scary. A senior vice president of a company making more than $200 an hour who wasn't being paid for anything that couldn't be tracked by her computer, so talking, writing on paper, even thinking on company time wasn't automatically compensated. Anyone who would argue this is an accurate reflection of productivity is crazy.
As workers our total contribution to our companies is far more nuanced than the hard work outputs that are tracked or even trackable. How much effort do you put into your team succeeding? Or being a positive influence in the workplace? Or helping with the training and coaching of newer employees? Or being the person who's accessible to someone when they're having a tricky day? In your average office environment, there are dozens of things going on all the time that on one hand can be viewed as distractions to your core business but on the other hand are a vital part of running a successful business where team members are engaged.
For every position where there are really strongly measurable outputs like an outbound salesperson, there is another where those outputs are always going to be rather more fluffy, like someone working in HR. And it's been well evidenced that the best salesperson doesn't necessarily make the best sales manager, so there's definitely something more complex going on here than just the people with the best metrics are stars and everyone else is a lazy slacker.
What does this have to do with Microsoft 365 Copilot though? Well I'm getting to that.
Giving your most valuable, and probably most expensive, employees the very best tools for their work is vital to any business' overall level of success. This is the foundational argument I have seen from those who are advocating being highly selective in the deployment of AI tools, where the cost of those tools must be balanced against the cost and pay-back associated with the employees - or level or type of employee - you are considering equipping with Microsoft 365 Copilot.
This is a sensible argument, but I think it overestimates the ability of top-level team members or star employees to overcome factors that cause organizational inertia in the opposite direction. Poorly engaged employees, those who are doing the minimum possible - or quiet quitting in today's parlance, churn caused by staff turnover, or other factors that cause your productivity bucket to become leaky - these are issues that must be addressed in the long-term if you want to succeed, and can even make it difficult to hang onto those highly compensated, highly productive, top performers who were lucky enough to be given those Microsoft 365 Copilot licenses.
The promise of AI technology isn't just about getting more done, it's about doing work better. The 9pm pings of emails arriving have just as much impact on the work-life balance of the $18.13/hour office assistant as they do on the $200 per hour senior vice president. Mundane, repetitive, low-engagement tasks, can tank motivation and morale at any level of your hierarchy. Having greater flexibility to keep on top of complex topics and even meetings, can be a benefit that is felt by someone involved in one project at a time or twenty projects at a time.
According to the Society for Human Resource Management, on average it costs just over $4,000 to fill a vacancy in a business. By the way, this would pay for Microsoft 365 Copilot licenses for 11 employees for an entire year. It also takes 42 days to fill a position, and this doesn't factor in all the soft costs of getting that new person up to speed when they arrive. In my experience, every vacancy puts pressure on managers and on other team members and tanks progress on issues that might be important but are not urgent; and vacancies also mean that HR teams have to focus on recruitment and onboarding rather than on things that add greater value like employee engagement.
The average company will see voluntary turnover in about 1/40 employees per year, so a company of 200 might expect to spend $20,000 in hard recruitment costs, and lose 210 days or nearly 81% of a work year in unfilled positions each year. But employee engagement is a big lever in this, according to Gallup research, highly engaged workplaces can see up to a 67% reduction in employee turnover.
Put all this together and it becomes obvious that the potential of AI to have positive or negative impacts on any business stretches far beyond the immediate productivity gains of any one worker using those tools. Over the last few years attitudes to work have changed radically, the acceptance of different types of treatment for different hierarchical levels has reduced, and whereas once the work day extending at the whim of your boss or your colleagues was just part of the job, it is now seen as a big problem by many.
No one sees their Outlook email account as a job perk, but if a tool like Microsoft 365 Copilot truly does give any worker the ability to get time back, it's more likely to be seen in the same light as a benefit like flexible working rather than tooling that is just there to get more done like a QuickBooks license. The entry-level associate who up until now has tolerated their boss' Sunday evening emails, may find them rather more frustrating when she has to wonder whether they were generated using that prized Microsoft 365 Copilot license that has been bestowed to the bosses but not to her.
Say you're one of those workers who up until now has been used to working unpaid for 8 hours each week and your business gives you Copilot and that saves you a few of those hours. You're still working on your own time a bit, and the whole benefit is to you, so you're not actually getting any more done, but when you receive a job offer from a competitor that doesn't use AI, how much would it cost them to get you to go back to giving up more of your free time again? For a per worker investment that is at maximum 2.5% of pay, that seems like a pretty attractive benefit to offer, and one that will help to boost engagement and maximize retention in the long term.
Now, you clearly don't have to buy it for EVERYONE, but you should buy it for everyone whose role includes a reasonable proportion of the types of work that can be supported by a product like Microsoft 365 Copilot. And if that is found to have benefits in your business, you should definitely then look at options for other types of roles to be supported by AI - for example your service technicians probably don't need Microsoft 365 Copilot, but they may well benefit from some of the AI powered resource scheduling options coming online in Dynamics 365. The point is that you are using AI as a catalyst to improve work which in turn improves your business, rather than just seeing AI as a tool to directly increase revenue without all the steps to do with your people that are in between.
The fact is, if you are solely thinking about the cost of Microsoft 365 Copilot, or Duet AI for Workspace, or ChatGPT Enterprise, in the direct dollar for dollar work outputs you'll get in your business, you might be missing the point. These tools will definitely deliver benefits here, but their impact will be far wider, and transcend the types of measures you have thought about before from your Microsoft 365 resources. I wouldn't be surprised if in years to come you need to state whether a role gets access to the best AI on job ads to recruit the best talent, or if upgraded AI allowances become part of retention negotiations.
One thing is clear. If you don't deploy Microsoft 365 Copilot, there's probably a competitor down the street who will. And remember, you're not just competing for the highest paying customers, you're also competing for the best talent, and your decisions on how to use AI in your business may have a big impact in both these areas.
What do you think? Have I convinced you that these expensive tools deserve mass deployment? Or are you still comfortable with a cautious and limited approach? Let me know down in the comments.
