Students are not graded on the number of significant figures in numeric answers (unless the question is specifically asking about significant figures). In general, they just need to be within a certain tolerance (usually 2%) of the correct answer. To prevent errors, they should not round off answers at each intermediate step of calculations. Instead, they should keep all the digits until the very end. When in doubt, it is better for students to keep too many digits than too few. For more information about significant figures, see Significant Figures.
Sapling Learning has been designed for science education with a sophisticated engine for handling significant figures. Many competing systems simply round off numbers to a set number of digits, whereas Sapling Learning displays the proper number of significant figures in all solutions.
One area in which this engine excels is the handling of multistep calculations. Many textbooks recommend that students use unrounded values throughout a multistep calculation and apply significant figure rules only to the final answer. Sapling Learning does exactly this by only rounding the answer at the last step of the calculation.
Students shouldn't be concerned if their intermediate values don't match exactly what is on the screen. We want to show students proper significant figure rules at each step. However, our final answer must match the answer that students get when they perform the calculation the proper way (by carrying all significant figures until the end)—otherwise students would be scored as incorrect for following the proper procedure—so we are careful to deliver the best of both worlds. Sapling Learning rounds numbers displayed in intermediate calculations to proper significant figures, but "remembers" the unrounded number for use in the final calculation, giving the student the same answer they find if they calculate the answer as their book recommends.
Your textbook may show rounding for each intermediate step, as does Sapling Learning. However, the final answer (to which a tolerance is applied and the student's answer is compared) is obtained by performing the calculation the proper way by carrying all digits throughout a calculation and then applying significant figure rules at the end.