The point of a ballpark estimate is to answer the question, "Is this in the ballpark?" Namely, can I afford it? Is it even worth paying you $1,000 for a discovery session if the implementation is going to cost me 10x my budget in any case? Is it worth taking the next meeting with you?
The author is right about anchoring to value, and the author is right that it's impossible to fully-scope a project in a minimal amount of time.
But you're the expert! So the right answer is: "We'd look to get a lot more accurate on cost, and a Discovery Session might be a good way to get you an accurate estimate on the cheap. But to get you a sense of magnitude, similar clients have paid $x-(x+20%) and have seen 10x returns. How does that fit into both your budget and the business case you had in mind?"
That's correct. This makes me sound bitter, but here's how it really works:
They ask for a ballpark estimate, and you quickly throw out a number that's believable but not too high so as to scare them off. This is usually done by a sales person who IM'd a free developer and gave a two or three line summary of what the client is suggesting.
Then there is some back and forth, where they customer adds or removes requirements, and the number changes somewhat accordingly, but really, in the context of sales, not technical challenge. "Oh we wrote X for some other client a while back, we'll throw that in for free."
Then it might get complicated, and you get nervous, and you suggest having the client pay to do a detailed analysis. The customer says no, cause, you know, those ballparks are awesome and free.
You agree on an approximate number and contracts are written up. Because really, the ballpark is the actual estimate. We were fools to think otherwise. But of course there is lots of wiggle room and vagueness written into the contract because you nor the client didn't really want to spend much time fleshing it out.
Customer signs. Contrary to what you might think, you just won! You hooked 'em!
Now, you work on the project, and maybe you hit the ballpark, and maybe you don't. If you don't you go back to the client and say, "Well, you know it was just a ballpark estimate, we had lots of new things come up, and you added this and that, and now it's going to cost more money."
Customer can't argue too much, because they didn't want to pay for a detailed analysis, and they did ask for a ballpark. And now they're already in it for some money, and to bail now would mean losing that investment. So they spend more.
Lather, rinse, repeat until the project is done or the customer is out of money.
Now don't think I don't find this system awful - I do. But that's how it works. You'll never get it to change because it will always involve ignorant sales people and ignorant business owners, and they are playing a social game, not detailing a technical endeavor.
For a real example from ad tech, "We need access to a DMP (Data Management Platform) to learn more about our visitors." "That will be about $5k/month for a license." "That's fine."
Oops, you expected to be able to get what the DMP thinks about each individual customer and do the analytics on your side? That license only lets you take a pool of people, and get back aggregate statistics. Not collect per person data...
You have to communicate orders of magnitude, and keep stressing that you'll work with them to figure out the cheapest thing that makes sense, but costs can jump very, very quickly.
Otherwise I completely agree with your 'right answer' and how you framed it: "You're the expert". An estimate by comparison is no bad thing, and demonstrates experience.
At the end of project discovery, you’ll find that 3 exciting things will happen:
1. Now you’re allowed to ask how much the project will cost! You thought I had forgotten about that, didn’t you? It’s OK – I know you’ve been dying to know when you can ask, and the answer is, “now”. And the best part is, I’ll be able to give you a real answer, not a ballpark estimate.
2. You’ll have a more in-depth understanding of your project. From pros & cons to potential profitability, you’ll have a newfound sense of x-ray vision for how your business is currently operating vs. how the new project will enable it to operate. You’ll have a rock-solid grasp of why you want to do this project and what you’ll get from it.
3. I will hand you a complete project plan. This project plan covers all the particulars we uncovered during discovery; this plan is yours, bought and paid for. If you want, you can use this plan to execute the project yourself, or even shop the plan to other consultants for an estimate.
That's key. A small business like his is probably not all over the place, they must be hired for a limited categories of problems. After 12+ years doing it, the author should be able to ballpark something within a set of assumptions that will be confirmed during the discovery phase.
I have seen companies with vastly different price point. There is a huge difference hiring a company that charge $2000/manday vs one that charge $200. ( not to say that you don't get your money worth with the $2000 guys but you don't need the Apple vs Samsung lawyers to represent you against the plumber that fucked up your office kitchen tiles when fixing the sink )
And being both sides of that table at some point, it is always possible to give an estimate.
The difference is that the estimate should be a range.
How tall is the Eiffel Tower? If you didn't know and gave an estimate you might say, 172m is quite high, maybe that's high enough to be the answer without over-shooting.
You could give a range as the answer: Between 100m and 500m tall. It's definitely more than 100m high, for why else would there be lifts, such a wide base, etc, etc... and it's definitely less than 500m, because it certainly doesn't rise half a KM into the sky.
That first estimate... wrong. Wildly wrong, but because it is a fixed and precise number it gives the illusion of accuracy and knowledge on my part where none was applied in coming to the number. As an estimate, the client will be upset.
That second estimate... well now I've Googled and the answer is 300m and so it turns out the estimate was correct.
That's all a client wants... the best and worst case, and when there is a huge variance between the two, what is the basis for that variance? How much of it comes back to the clients as requirements that could be made simpler? How much is the tech choice? The features?
Always give an estimate, just make it a range.
(The reverse is true for, say, salaries. If you tell me you pay $100-$200k/yr for a job, I'm already thinking about how to spend that $200k.)
So your ranges should be within your negotiating contingency (say, 20%).
$1m - we're only doing x, y and z, within these basic constraints, the scope does not slip without renegotiation.
$10m - you get m-through-z, we'll build in capacity for scope changes and discovery on the way
We should be lowest quartile ($1m-$3m) of that range if we can agree through a discovery phase what is in scope and basic constraints/assumptions.
I am looking for a logarithmic bucket to gauge if the approach is within my boundaries (time/money) as quickly as possible.
The question that a ballpark answers is simply this: can I afford to work with these people?
When I finally took a sales class the most insightful thing that I picked up was to ask the question "is that a lot or a little" because it truly reflects what ballpark estimates are about. There are some people who will be totally at their limit with $1,000 budget and others who won't take you seriously if you say $50,000 because it will be too low.
That's what a ballpark does and it's beneficial for both sides because it essentially allows you both to fail fast and not waste each others time. If you say $10,000 and a client balks at the cost, no amount of invested sales time is going to win you that client and it's not going to be worth it to you anyway. They either can't afford what they want or they need an off the shelf solution.
If you say $10,000-$20,000 and the conversation continues then you're good to continue.
This is all very reasonable and we are happy.
Communicate, be reasonable, and people will generally understand.
Large projects need the discovery/consultation process to help arrive at the estimate so that at least there are defined parameters and resources committed to each one.
But this isn't asking about fundamental physics or the development of a new operating system or something. There are unknowns and more importantly there is tailoring to the clients needs and budget, but the unknown unknowns should be minimal, that's why they're interested in hiring you. There is a big difference between the author's "well $10k vs $50k and what about ROI" versus "What you just asked for sounds like what $MEGACORP started two years ago with an initial budget of $12 million and last I heard it is now up to $30 million". Most development jobs do have a range of budgets they can be tailored to but that range isn't infinite; some requests are just clearly going to be flat out impossible for a given business and they shouldn't have to spend any capital, economic or human, chasing a total dead end.
I also think the author very much overstates software being some inherently different, special thing. He wrote:
"you’re asking me to figure out what it will cost to build something that, by definition, has never existed before."
That's silly, unless it's something extremely cutting edge or aimed with vast scalability requirements, "websites" have absolutely existed before. Not this specific one sure, but lots of others similar to it, made for businesses with needs that also share similarities. The real world is also full of projects that are even more potentially unbounded, like real estate or construction. That sort of place is exactly where the whole question of "ballpark estimates" originates, because it's all too easy for a concept guy to design out some "beautiful and organic structure" that in terms of architecture and engineering is obviously going to be budget busting. A basic reality check matters. Precise estimates are what comes when a project goes out to bid, but any decent professional in the business should be able to eye a bridge or building design and feature set and quickly come up with a rough idea of whether it's even worth pursuing or is just silly.Edit: Tried to use typical asterisk as multiply symbol, forgot that added emphasis.
Any decent client understands that you can't give them a fully spec'd estimate before you've even begun the engagement. But they want to understand what the process will be, and yes--that includes what the ticket price might look like.
Having been on both sides of this exchange, I really, truly understand why ballpark estimates are hard. But I also understand that approximately nobody is going to approve an expense without having any idea of how large or small it might be.