What is the Analytic Hierarchy Process?

Where it comes from

AHP was developed in the 1970s by Thomas L. Saaty, a mathematician and professor at the Wharton School of the University of Pennsylvania. Saaty was looking for a practical way to help people deal with decisions involving many competing criteria — something rigorous enough for analysts, but simple enough for anyone to answer.

The method spread quickly. In 1983 Saaty partnered with Ernest Forman to build Expert Choice, one of the first decision-support software packages, and in 2008 the Institute for Operations Research and the Management Sciences (INFORMS) recognized AHP’s broad impact on the field. An international research community still meets at the biennial International Symposium on the Analytic Hierarchy Process.

The five steps of AHP

Every AHP analysis follows the same backbone:

1. 1Model the decision as a hierarchy — a goal, the criteria that matter, and the alternatives.
2. 2Judge the elements in pairs to establish priorities.
3. 3Turn those judgments into numerical weights.
4. 4Check the judgments for consistency.
5. 5Synthesize everything into a final ranking and decide.

Step 1 — Model the decision as a hierarchy

AHP starts by giving the decision a shape. At the top sits the goal — the thing you’re trying to decide. Below it are the criteria: the factors that matter, which can be split further into sub-criteria for large decisions. At the bottom are the alternatives — the options you’re choosing between.

The hierarchy matters because it lets you deal with one small piece of the decision at a time, instead of holding everything in your head at once.

Step 2 — Compare elements in pairs

This is the heart of the method. Rather than rating each criterion on its own, you compare them two at a time: which of these matters more for the goal, and how much more? Then you do the same for the alternatives: under each criterion, which option is better, and by how much?

Each answer is expressed on Saaty’s fundamental scale, which runs from 1 (the two elements are equally important) to 9 (one is extremely more important than the other):

Step 3 — From judgments to priorities

Behind the scenes, the pairwise answers fill a comparison matrix. If you said cost is twice as important as commute, the matrix records 2 in one cell and its reciprocal, ½, in the mirrored cell.

From this matrix, AHP derives a priority weight for each element — classically by computing the matrix’s principal eigenvector. You don’t need to remember the linear algebra; what matters is the result: a set of weights between 0 and 1 that sum to exactly 1 and capture the relative importance of every element. In the example above, cost ends up with 57% of the weight, commute 29%, and neighborhood 14%.

Step 4 — Check the consistency

Human judgments wobble. You might say cost matters twice as much as commute, commute twice as much as neighborhood — but then judge cost only equal to neighborhood. AHP quantifies that wobble with a consistency index (CI), which it compares against the average index of purely random judgments (the random index) to produce a consistency ratio (CR).

The conventional rule of thumb, proposed by Saaty: if CR exceeds 0.1, the judgments are too contradictory to be trusted, and it’s worth revisiting the comparisons that clash the most. A little inconsistency is normal and even expected — the check exists to catch reasoning that has genuinely lost its thread.

Step 5 — Synthesize and decide

The final step multiplies everything together. Each alternative’s score under each criterion is weighted by that criterion’s priority, and the results are summed into a global priority for every alternative. The output is a ranking in which the numbers have real meaning: an option with priority 0.58 isn’t just first — it carries roughly twice the weight of one at 0.27.

Crucially, AHP doesn’t decide for you. It reflects your own judgments back in a structured form — the final call, including whether to trust the result at all, stays with you.

AHP doesn’t replace your judgment — it organizes it.

Where AHP is used

Fifty years on, AHP is one of the most widely used multi-criteria decision-making methods in the world. Governments apply it to resource allocation and infrastructure planning; companies to supplier selection, project prioritization, and strategy. Documented cases range from Microsoft assessing software quality and the U.S. Department of Agriculture managing watersheds to risk assessment for petroleum pipelines and choosing bridge construction methods.

It’s also a staple in university courses on operations research and quality management, and shows up inside frameworks like Six Sigma and quality function deployment.

Criticisms and limits

No method this popular escapes scrutiny. The best-known criticism is rank reversal: in the classic form of AHP, adding or removing an alternative can sometimes change the order of the others — behavior some decision theorists consider a flaw. Variants such as the ideal synthesis mode (1993) address this. Researchers have also debated whether translating verbal judgments like “moderately more important” into fixed numbers can introduce precision that isn’t really there.

The practical takeaway isn’t that AHP is broken — it’s that it works best as a thinking aid rather than an oracle. It excels at surfacing your priorities and keeping your judgments honest; the interpretation of the final numbers remains a human job.

Further reading

This post covers the essentials, but there’s much more — sub-criteria, group decision-making, the mathematics of eigenvectors and random indices. The Wikipedia article on the Analytic Hierarchy Process is a good place to continue, including fully worked examples.