Most people begin with the beginning. That's why they never finish. Defining the end first is what creates direction. Today we go slowly through the simple tool called backward design.
Three years ago I had a huge question for myself. "Why do I start dozens of books and never finish any?" I opened my laptop — about 47 started writings. Finished: 0. Strange. I was working hard with no output. A friend threw me one line. "You never defined the end." That one line changed my working method. Today I'll share it. Start from the end. Let me go slowly.
This essay covers three things: why starting from the start doesn't work, how to define the end, and how to work backward.
Let me lay down the principle. Very simple. If you start without defining the end, it won't end.
Sounds obvious. Yet most people break this rule. Because starting is easy. Open a blank doc and write the first sentence — 30 minutes. But deciding "what should this look like when done" takes 3 hours. Hard. So people run to the easy side and push the hard side to later.
Result? It never ends. Or more precisely, because you don't know where the end is, it can't end. Why I had 47 unfinished starts — no end defined, so no idea where to stop, endless middle-wandering. Wander too long and start a new one. 47 starts.
Imagine a road trip.
You start driving but never typed a destination into the nav. What happens? You're pressing the gas. The wheels turn. But you don't know which way. Right, left, or straight. Any distance you drive, you don't know where you arrived.
Now type in the destination first. The nav calculates the route. "Turn right in 1 km, go straight for 2 km, then left." Every moment of judgment gets easy. At this junction, what to do — the destination decides.
Work is the same. Mark the end first and every choice along the way becomes easy. Without an end, every choice is vague.
A start without an end doesn't end.
That's the one-line summary.
How do you actually define the end? Three questions are enough.
One — "Who receives this?" The end always has a receiver. Writing = reader. Product = user. Video = viewer. No receiver, no end. For this essay, the end is "a creator who can't finish their work understands backward design." Make the receiver specific and the end sharpens.
Two — "If the receiver takes only one thing, what is it?" If they walk away carrying one thing, what's the most important? For this essay, it's the single line "define the end first." The rest is scaffolding to make that line land.
Three — "How will you know they took it?" Most important question. Define a measurable signal. For me: "Does the reader, tonight, try to write their project's end in one line?" When these three questions have answers, the end is concretely defined.
With the end defined, the rest is backward arithmetic. Easy. Trace back from the end.
Take one of my book projects. End: "a completed 200-page manuscript in 3 months."
Backward:
From the end back to the start, it reduces to "2 hours of focused work today." That's the power of backward design. A far-away end translates into this moment's action. Starting from the start, you never get these numbers.
With this method I finished that book in 78 days. After abandoning 47 unfinished starts, it was my first completed book. The only change? Defining the end first.
The same principle works when giving tasks to AI. Most people start like this. "Write something." "Write code." "Give me ideas." That's a start without an end. The AI wanders.
Instead try this. "The goal is that a 30-something office worker, after reading this, writes their project's end in one line tonight. Write a 5000-character essay that achieves that goal." Say the end first and AI calculates a path. The output gets dramatically better.
After I internalized this, my AI output quality went up roughly 3×. Same model, same tool — the only change is the opening sentence of the prompt.
One misconception. "Doesn't defining the end take away freedom?" Common question. No. You can change the end anytime.
What matters is that in this moment, an end exists. If a better end occurs to you in three weeks, change it then. Change the end and the route recalculates automatically. Like a nav rerouting. Having an end is not the same as an end being unchangeable forever.
I changed the end of my book project twice. Originally "AI tool usage guide" was the end, but mid-work I realized "a work philosophy for the AI era" was the right end. Once I swapped it, the path from that point recalculated and the result was much better. An end is a compass, not a cage.
One caution. If the end you defined is too big, it kills momentum. "A bestseller in 2 years" sounds great but doesn't translate to today's action. Too far.
When that happens, slice the end. I cut big ends into 3-month chunks. "Bestseller in 2 years" → "First draft done in 3 months" → start the backward math. 3 months is the unit that actually back-calculates well. Shorter is too narrow. Longer loses the sense of urgency.
Get in the habit of defining a new end every 3 months. Four backward-design cycles a year. I've been working on this rhythm for five years.
Defining the end is different from finishing what you start. People mix these two, but I separate them clearly.
"Finish what you start" is about the will to complete. Don't quit midway, keep going when you're tired. Great advice. But it doesn't tell you where the end is. Will without a map.
"Define the end first" is about designing direction. Fixing where "complete" is, before you depart. Only then does completion become measurable. The anxiety of "is this enough?" disappears.
I need both. Define the end, then finish to that end. Reverse the order and nothing works.
Start without an end and it won't end. Define the end first and direction appears; direction lets this moment's action be determined. Use three questions (receiver, core, signal) to define the end, then work backward to today's task.
Tonight, do one thing. Pick one active project and write its end in one line. If that's hard, that's exactly how vague your direction is right now. Spend 30 minutes on it and the next 3 months sharpen.
A start without an end doesn't end. Start from the end.
Remember — receiver, core, signal.