An interesting conversation came up a few days ago on the Pandas Arms1:
as an independent contractor working in data science, what day-rate should you charge clients, and how should you compare this rate to an annual salary?
Now, before we go further, please note that I’m absolutely not qualified, regulated or approved to give any kind of financial advice to anyone about anything. The content of this post is pure personal opinion and rank speculation: please do your homework elsewhere. Don’t listen to me(!)
I spent 2005-2009 as a software and decision sciences consultant in PA Consulting Group in London & NYC, and spent 2013-2019 selling and executing work in London & Dublin via my own consulting company Applied AI.
As such, I’m inclined to think of work in terms of projects, deliverables, deadlines, and day rates. I shared the following back-of-the-envelope calculations for day rates and utilisation on the Pandas Arms, and why not share here too: it might be interesting to someone.
TLDR: day rates sound high compared to salaried work, but there’s a lot of risk factors they need to cover (utilization carries the most variance). My rule of thumb is a 1/144 multiplier from annual pre-tax salary to roughly equivalent day rate.
This is a pure day-rate calculation ignoring equity stakes, risk/reward, bulk sales discounts, IP licensing, profit-maximisation etc.2
You sustainably have approx. 225 days per year available for chargeable work:
365 days - 104 (weekends) - 8 (UK bank holidays) - 25 (vacation) - 3 (sick / discretionary) = 225 days
… this is a nice round number.
Aim for an 80% utilisation:
This is classic explore / exploit and should let you balance your current client work with the personal / community / technical / business development that’s essential to get your next contract
Without this active management, your worklife will become a cycle boom and bust: times of simultaneous unfullfillable offers vs times of total drought
If you’re doing niche / bleeding edge work, or making a new market, or running a big team, you might need to aim for 60% utilisation so you can keep on top of developments
Much lower than 60% and you’re probably more of a classic expert consultant and charge clients according to outcome / profit share / blackmail, so this entire blogpost is irrelevant
Note these are also nice round numbers 80% = 180 days, 60% = 135 days, and 180 / 12 = 15 days = 3 weeks per month; easy to play around with and keep track of in your head
If you get 100% utilization for a 12 month contract, then well done! Treat this as an annual bonus payment (+25% woot), and try to allocate ~8 hrs/wk on technical and/or business development for the next contract, and be careful about IR35.
Hire an accountant, a bookkeeper, use modern software and pay your taxes. This post is for UK readers, I’ve no idea how it works in USA.
Assume approx 20% corp tax, assume minimal VAT offsetting benefits:
Corp tax at 19%, round that up to 20% to include costs of doing business and any of your time spent on business admin
Flat rate VAT used to be cool, but HMRC mostly killed that off, so assume no meaningful savings
Note you can choose to skip employers and employees national insurance by paying yourself via dividends
Assume dividend marginal tax rate roughly the same as salaried employment:
Importantly, it actually isn’t similar at all, but at first approximation, use this delta to cover employer benefits (pension plans, health care, life cover etc)
Remember business expenses for IT equipment, trips to conferences etc
Remember you don’t have to take all the money out of the company each year, rolling over can be advantageous.
Using nice round numbers for example:
A 90k salary pretax earned over a full 225 days, becomes / 225 = 400 ('salaried' day rate) / (1 - 0.2) = 500 (adjust for corp tax (approximated)) / (180/225) = 625 (adjust for utilisation) Collapse those modifiers into 225 * 0.8 * 0.8 = 144 90000 / 144 = 625
… so take your desired annual salary, divide by 144 and there’s your approximately equivalent day rate ex VAT.
Naturally adjust that:
Perhaps food for thought.
The Pandas Arms is a long-standing Slack related to PyData London and attracts a very knowledgeable and engaged group of mainly Londoners, mainly discussing Python-based data science ↩︎
This also ignores costs of business: the big consulting corps have a considerable cost-base, e.g. a 3:1 ratio of billable consultants to back-office staff, which goes some of the way to explain their fees ↩︎