AHP vs Weighted Scoring: Why the Math Matters for Real Decisions

Q: What is the Saaty 1-9 scale?

Thomas Saaty defined: 1 = equal importance, 3 = moderately more important, 5 = strongly more important, 7 = very strongly more important, 9 = extremely more important. Even numbers are intermediate. The reciprocal applies for the reverse direction.

Q: What's a good consistency ratio in AHP?

Below 0.10 is the standard cutoff. Above, your pairwise judgments are likely contradicting each other and the derived weights aren't reliable. Above 0.20 means you should reconsider — you may not have a coherent preference yet.

By Dmitrii Raev · May 2026 · 7 min read

TL;DR. Weighted scoring asks you to pick weights upfront ("salary is 40% important"). AHP asks you to compare criteria pairwise ("how much more does salary matter than commute?") and derives the weights mathematically. The pairwise approach catches a failure mode weighted scoring quietly buries: your own contradictory judgments. AHP flags this with a consistency ratio. Weighted scoring just gives you a number that looks right and isn't.

The two methods in one sentence each

Weighted scoring: assign each criterion a percentage (must sum to 100%), score each option 1-10 on each criterion, multiply, sum, pick the highest. This is what every "decision matrix template" on the internet does.

Analytic Hierarchy Process (AHP): compare each pair of criteria on a 1-9 scale ("salary is moderately more important than commute"), then do the same for options against each criterion. Eigenvector math turns the pairwise comparisons into priority weights. A consistency ratio measures whether your judgments contradict each other.

Where weighted scoring quietly fails

Picking percentages out of thin air is harder than it looks. Try this test: write down right now what percentage you'd assign to each of these for picking your next laptop:

Battery life
Performance
Weight
Screen quality
Price

Most people give answers like 20% / 25% / 15% / 15% / 25% — round numbers that sum to 100. The problem: those weights aren't reasoned, they're guessed to sum to 100. You didn't actually think "battery life is exactly 1.33× more important than weight." You thought "uhh, both matter, give them similar numbers."

  The hidden failure mode. Weighted scoring asks you
  to do too many things at once: rank importance, quantify it, and
  normalize to 100%. Your brain quietly defaults to "make the numbers
  look reasonable" and stops thinking about what each criterion is
  actually worth.

How AHP forces honesty

AHP breaks the same task into pairwise comparisons. You only ever decide between two things at a time:

How much more does battery life matter than performance? (1-9 scale)
How much more does battery life matter than weight?
How much more does performance matter than weight?
...and so on for each pair

This is harder per question but easier per decision: pairwise comparisons are how the brain naturally reasons about preferences. You're not performing arithmetic on percentages, you're answering "A or B, and by how much?"

The consistency ratio: why this is the killer feature

Here's where AHP earns its keep. Suppose you say:

Salary is moderately more important than commute (3 on the 1-9 scale)
Commute is moderately more important than office culture (3)
Office culture is strongly more important than salary (5)

These three judgments contradict each other. If salary > commute > culture, then salary should be even more important than culture, not less. AHP detects this with a consistency ratio (CR). Below 0.10 is considered acceptable; above means your pairwise judgments don't hang together and you should revisit them.

Weighted scoring has no equivalent. You write down 40% / 30% / 30% and the math obediently spits out a winner — even if those weights came from thin air or contradict your other reasoning.

Worked example: choosing between two job offers

Two offers, three criteria: Salary, Growth potential, Work-life balance.

Method 1: Weighted scoring

You guess weights and scores:

Criterion	Weight	Offer A score	Offer B score
Salary	40%	9	6
Growth potential	35%	5	9
Work-life balance	25%	7	6
Weighted total	100%	7.10	7.05

Offer A wins by 0.05. You take Offer A. Three months in you regret it because you're bored. What happened? Your "35%" for growth was a guess, not a reasoned weight. If growth is actually 50% important to you, Offer B wins by a mile.

Method 2: AHP pairwise

You compare criteria two at a time:

Salary vs Growth: growth is moderately more important (3 in growth's favor)
Salary vs Balance: equally important (1)
Growth vs Balance: growth is moderately more important (3 in growth's favor)

The eigenvector calculation produces these weights:

Criterion	AHP-derived weight
Growth potential	54%
Salary	27%
Work-life balance	19%

Consistency ratio: 0.04 (acceptable). Now apply the same option scores:

Criterion	AHP weight	Offer A	Offer B
Growth	54%	5	9
Salary	27%	9	6
Balance	19%	7	6
Weighted total	100%	6.46	7.59

Offer B wins by 1.13 points. Same data, opposite conclusion. The difference is the weights came from explicit pairwise reasoning, not gut numbers that "feel right."

When weighted scoring is good enough

Don't reach for AHP for everything. Use weighted scoring when:

You have just 2-3 criteria (pairwise barely adds anything)
The criteria are obviously dominant ("price is everything")
You're doing back-of-envelope analysis, not committing for years

Reach for AHP when:

You have 4+ criteria where importance isn't obvious
The decision is irreversible or high-stakes (job, home, $50k+ purchase)
You catch yourself rationalizing weights to match a preferred answer
Multiple stakeholders have different priorities (AHP makes the disagreement explicit)

FAQ

Is AHP harder than weighted scoring?

Per question, slightly. Per decision, often easier — you're answering "A or B, by how much" instead of inventing percentages. With an app like Decisio the eigenvector math is invisible; you just answer the pairwise questions.

What is the Saaty 1-9 scale?

Thomas Saaty (the inventor of AHP) defined: 1 = equal importance, 3 = moderately more important, 5 = strongly more important, 7 = very strongly more important, 9 = extremely more important. Even numbers (2, 4, 6, 8) are intermediate. The reciprocal applies for the reverse direction (if A is 5× more important than B, then B is 1/5 as important as A).

What's a good consistency ratio?

Below 0.10 is the standard cutoff. Above, your pairwise judgments are likely contradicting each other and the derived weights aren't reliable. Above 0.20 means you should seriously reconsider — you may not have a coherent preference yet.

Can I do AHP in Excel?

Yes, but the eigenvector calculation isn't a built-in function — you need either an iterative approximation, the LINEST workaround, or a macro. Most people who try this either give up or end up with weighted scoring in disguise. Decisio handles the math automatically and includes the consistency check, which is the part that's easiest to skip when rolling your own.

Who uses AHP in real life?

NASA (mission planning, supplier selection), Boeing (engineering trade studies), the World Bank (project prioritization), most Fortune 500 procurement teams, and academic researchers across operations research, supply chain, and policy analysis. It's been cited in 100,000+ papers since 1980.

Try AHP without the math headache

Decisio runs full AHP pairwise comparisons + consistency check. Free for 3 decisions.

Download on the App Store

iOS 17+ · iPhone & iPad