Getting The Slope Right Is Where People Mess Up
The equation of a tangent line is y - y = m(x - x), where m is the derivative evaluated at your point and (x, y) is the point of tangency. That's it. Most of the work isn't the formula itself, it's correctly computing the derivative and plugging in the right numbers without making arithmetic mistakes. I've seen students lose points not because they didn't know the concept but because they took the derivative of x³ as 3x instead of 3x². Standard stuff. Start with a function and a specific x-value. Take the derivative of the function using whatever rule applies — power rule, product rule, chain rule, quotient rule — and evaluate it at that x-value. That gives you the slope. Then find the y-value by plugging the x back into the original function. You now have a point and a slope, and the point-slope form handles the rest. Rearrange to slope-intercept if your instructor wants it that way. Here's a concrete example. Find the equation of the tangent line to f(x) = 2x³ - 5x + 3 at x = 2. The derivative is f'(x) = 6x² - 5. At x = 2, that's 6(4) - 5 = 19. So the slope is 19. The point is (2, f(2)) = (2, 16 - 10 + 3) = (2, 9). The equation is y - 9 = 19(x - 2), which simplifies to y = 19x - 29. Check it: at x = 2, y = 38 - 29 = 9. Correct.
Now here's something most textbooks don't emphasize enough. The derivative gives you the slope of the tangent line, but only if the function is differentiable at that point. If there's a cusp, a vertical tangent, or a jump discontinuity, the whole method falls apart and you need to state that explicitly rather than pretending the derivative exists. I remember working through a problem set where the function was defined piecewise, and at the boundary point the left-hand derivative was 4 while the right-hand derivative was -2. The function had a corner there, not a smooth curve. Anyone who just blindly applied the derivative rule would get the wrong answer. I had to compute both one-sided limits separately and conclude the tangent line doesn't exist at that point. Took me about ten minutes to figure out what was going on instead of twenty minutes of writing nonsense. Another thing people miss: the tangent line is a local approximation. It matches the function's value and its first derivative at the point of contact, but it can diverge pretty quickly as you move away. For f(x) = sin(x) at x = /2, the tangent line is y = 1, which is flat. That's correct at the point, but it's a terrible approximation anywhere near x = or x = 0. Don't confuse the tangent line with the function itself beyond an immediate neighborhood. The linearization error grows roughly with the square of the distance from the point, so if you're working more than a few units away you should be using higher-order approximations or just evaluating the original function directly. There are also cases where the derivative is zero — horizontal tangents — and students sometimes think that means the tangent line is the x-axis. It doesn't. If f(x) = x² at x = 0, the tangent line is y = 0, which happens to be the x-axis, but if f(x) = x² + 3 at x = 0, the tangent line is y = 3. The slope is zero, but the y-intercept comes from the original function, not from the origin. Easy to overlook.
If you're working with implicit functions where y isn't isolated, you use implicit differentiation. Take the derivative of both sides with respect to x, treat y as a function of x, solve for dy/dx, then proceed exactly as before. The algebra gets messier but the logic is identical. I usually spend extra time checking my implicit differentiation work because it's where sign errors and forgot-the-chain-rule mistakes hide.
Get the Full Details
When The Method Breaks Down
The main failure modes are non-differentiable points, vertical tangents where the derivative blows up to infinity, and parametric or polar curves where you need to convert things first. For a vertical tangent, the slope is undefined, so the point-slope form doesn't work. You end up with an equation like x = a instead. For parametric curves, dx/dt and dy/dt both need to exist, and dy/dt can't be zero when dx/dt is zero unless you take a limit. It's straightforward if you've done it a few times and annoying the first couple of times you encounter it. The one workaround I found useful when dealing with messy rational functions is to simplify the expression before differentiating whenever possible. A complicated quotient that reduces to something simple after canceling common factors will save you a significant amount of algebra. I cut my computation time on one particularly ugly problem from about twenty minutes down to five just by noticing a common binomial factor that canceled out. Always check for simplification opportunities first.
