IMB 895

HARITHA SARANGA AND SHASHIKANT KUMAWAT

STEP-BY-STEP TUTORIAL OF ANALYTICAL HIERARCHY PROCESS WITH AN

EXAMPLE

The analytical hierarchy process (AHP) is a method used for complex decision-making in which various subjective
components are quantified and prioritized as per the decision-maker’s own criteria. The final priority order can be
used objectively to arrive at the decision. This tool was developed by Prof. Thomas L. Saaty in the 1970s. It is one
of the techniques of multi-criteria decision making (MCDM).

When a decision is dependent on multiple criteria and/or involves subjectivity, AHP framework can be used to
arrive at an optimal decision in a rational and structured fashion. To reduce bias in decision making, AHP reduces
complex decisions involving several criteria and alternatives into pairwise comparisons and provides a quantitative
methodology that ensures consistency in the decision-maker’s evaluations.

Key elements of decision making using AHP include an objective (goal), criteria, sub-criteria, and alternatives (final
options to choose from). A hierarchical structure of the elements is created with a goal at the top, required criteria
and sub-criteria in subsequent layers, and alternatives as the last layer. Simple AHP models might not have sub-
criteria and the hierarchy may contain only three layers.

Five steps of AHP

1. Define objective (goal) and gather elements of decision making such as criteria, sub-criteria, and
alternatives.
2. Develop a hierarchical structure of the elements.
3. Make a pairwise comparison of elements in each group, to determine the relative importance of
different criteria with respect to the goal.
4. Calculate weighting and consistency ratio.
5. Evaluate alternatives according to weighting.

Input values can be

 Actual measurements, for e.g., price, weight, number of days, etc.

 Subjective opinions, for e.g., satisfaction feelings, preferences, etc.

Haritha Saranga, Professor of Production and Operations Management, and Shashikant Kumawat, Research Associate, Indian School of
Business, prepared this case for classroom discussion. This case was developed from available and permitted sources of information. No
funding was sought or received from any source for the development of this case. This case is not intended to serve as an endorsement, source
of primary data, or to show effective or inefficient handling of decision or business processes.”

Copyright © 2021 by the Indian Institute of Management Bangalore. No part of the publication may be reproduced or transmitted in any form or
by any means – electronic, mechanical, photocopying, recording, or otherwise (including internet) – without the permission of Indian Institute of
Management Bangalore.
This document is authorized for use only in Prof. Suvechcha Sengupta's MBA Business Analytics Minor Decision analytics 24.11.2023 at Somaiya Vidyavihar from Nov 2023 to May 2024.
 Step-by-Step Tutorial of Analytical Hierarchy Process with an Example

Output is in terms of

 Criteria weights
 Consistency index (intermediate parameter)
 Ratio scale (ranking of alternatives)

Illustrative example of a simple AHP: Buying a mobile

Let us walk you through an example to understand how exactly AHP works. Let us suppose that you (the decision
maker) want to buy a new mobile phone, as your old phone is getting really old. Let us follow the five steps
mentioned above to see how AHP can help you with your decision making.

Step 1: Define objective (goal) & gather elements of decision making, such as criteria, sub-criteria, alternatives

Of course, there are several criteria you must consider before buying a mobile phone such as price, aesthetics,
camera quality, phone’s RAM etc. Also, there usually are only limited number of alternatives you are interested in
based on other requirements such as features (because you are looking for a smart phone), price range (because
you are still a student and cannot afford too expensive a phone), etc. You need to create a list of these alternatives
and gather data on the various criteria you have shortlisted for each of these alternatives, to be able to use AHP
and arrive at an ideal mobile phone.

We next model our problem with the following hierarchy, starting with the general and going toward the detailed.

Decision maker: Individual

Goal: Buying a mobile
Criteria: Price, Aesthetics, Camera Quality, RAM
Alternatives: Mobile A, Mobile B, Mobile C, Mobile D, Mobile E

Suppose the following are the five alternatives and their corresponding data for each of the four criteria:

Mobile Name Price1 Aesthetics Camera Quality RAM

Mobile A
Rs. 31999 Matte Black 5.9” 12MP 6GB
Google Pixel 4a)
Mobile B (realme X3
Rs. 32999 Shiny Blue 6.2” 64MP 12GB
SuperZoom)
Mobile C (Samsung
Rs. 36999 Shiny Grey 6.0” 48MP 8GB
S10 Lite)
Mobile D (Mi 10) Rs. 48999 Golden Pink 6.2” 108MP 8GB
Mobile E (ViVO X50) Rs. 37999 Glaze Black 6.0” 48MP 8GB

Step 2: Develop a hierarchical structure of the elements

Next, we can develop a hierarchical structure of the various elements identified in step 1 in a layer by layer fashion,
starting with general (goal) to more specific (the criteria and the alternatives) in a pyramid shaped diagram. Figure
1 represents the hierarchical structure of our decision-making problem. This process of creating a hierarchical
structure helps you to understand not only the big picture, but also each of the individual parts and their
relationship to the big picture.

1
One US$ = 72.92 Indian Rupees (Rs.), as on June 3, 2021

Page 2 of 9

This document is authorized for use only in Prof. Suvechcha Sengupta's MBA Business Analytics Minor Decision analytics 24.11.2023 at Somaiya Vidyavihar from Nov 2023 to May 2024.
 Step-by-Step Tutorial of Analytical Hierarchy Process with an Example

Figure 1
Hierarchical structure of a mobile phone buying problem

Step 3: Make a pairwise comparison of elements in each group, to determine the relative importance of
different criteria with respect to the goal.

To reduce the complexity of dealing with multiple criteria simultaneously, the decision-maker (DM) takes two
criteria at a time and comes up with a pairwise comparison matrix by giving the relative preference on a scale of 1-
9.
Table 1 shows the nine-point scale of relative importance when comparing two criteria at a time, which can be
used for AHP pairwise comparison.

Table 1
Nine-point scale of relative importance

1 Equal importance
3 Moderate importance
5 Strong importance
7 Very strong importance
9 Extreme importance
2,4,6,8 Intermediate values
1/3 , 1/5, 1/7, 1/9 Values for inverse comparison

Table 2 shows the matrix format for assigning pairwise importance values.

Table 2
Matrix of criteria for pairwise comparison

Price Aesthetics Camera RAM

Price
Aesthetics
Camera
RAM

Page 3 of 9

This document is authorized for use only in Prof. Suvechcha Sengupta's MBA Business Analytics Minor Decision analytics 24.11.2023 at Somaiya Vidyavihar from Nov 2023 to May 2024.
 Step-by-Step Tutorial of Analytical Hierarchy Process with an Example

Table 3 is updated with sample values for illustration. To fill this table, the DM needs to figure out how important
is one criterion (say, Price) with respect to the other (say, Aesthetics). For instance, in our example, you look at
price vis-à-vis the remaining three criteria (Aesthetics, Camera and RAM) and find that Price is five times more
important than aesthetics, four times more important than camera and seven times more important than RAM.
This is where the subjectivity comes in and each individual DM is likely to give different levels of importance to
each criterion vis-à-vis the other. The pairwise comparison makes sure that one is always looking at only two
elements at a time for comparison purposes, which helps simplify the problem in one’s own mind and allows one
to quantify his/her subjective judgments into objective values.

Once the above exercise is done, one can make use of the following formula to fill in the pairwise comparison
matrix, as shown in Table 3.

Relative importance = Row element value / Column element value

Table 3
Relative importance values (pairwise comparison values) for illustration

Price Aesthetics Camera RAM

Price 1 5 4 7
Aesthetics 1/5 1 1/2 3
Camera 1/4 2 1 3
RAM 1/7 1/3 1/3 1

Step 4: Calculate weighting and consistency ratio

Once the pairwise comparison matrix is ready, the next step is to calculate the weights. For this, the DM needs to
first compute the normalized pairwise comparison matrix. Table 4A uses the values of Table 3 to calculate the sum
of column values. Table 4B normalizes the values by dividing each value with column sums.

Table 4A
Sum the column values of pairwise comparison matrix

Price Aesthetics Camera RAM

Price 1.00 5.00 4.00 7.00
Aesthetics 0.20 1.00 0.50 3.00
Camera 0.25 2.00 1.00 3.00
RAM 0.14 0.33 0.33 1.00
Sum 1.59 8.33 5.83 14.00

Table 4B
Computation of the normalized values

Price Aesthetics Camera RAM

Price 1.00/1.59 = 0.6278 5.00/8.33 4.00/5.83 7.00/14
= 0.6000 = 0.6857 = 0.5000
Aesthetics 0.20/1.59 = 0.1256 1.00/8.33 0.50/5.83 3.00/14
= 0.1200 = 0.0857 = 0.2143
Camera 0.25/1.59 = 0.1570 2.00/8.33 1.00/5.83 3.00/14
= 0.2400 = 0.1714 = 0.2143
RAM 0.14/1.59 = 0.0897 0.33/8.33 0.33/5.83 1.00/14
= 0.0400 = 0.0571 = 0.0714

Page 4 of 9

This document is authorized for use only in Prof. Suvechcha Sengupta's MBA Business Analytics Minor Decision analytics 24.11.2023 at Somaiya Vidyavihar from Nov 2023 to May 2024.
 Step-by-Step Tutorial of Analytical Hierarchy Process with an Example

The criteria weights for each criterion in the rows can next be computed by taking the average of normalized row
values, as shown in Table 4C. For example,

Criteria weight of “Price” = (0.6278+0.6000+0.6857+0.5000+0.6034)/4 = 0.6034

Table 4C
Computation of the criteria weights

Price Aesthetics Camera RAM Criteria weights

Price 0.6278 0.6000 0.6857 0.5000 0.6034
Aesthetics 0.1256 0.1200 0.0857 0.2143 0.1364
Camera 0.1570 0.2400 0.1714 0.2143 0.1957
RAM 0.0897 0.0400 0.0571 0.0714 0.0646

Please note that, while assigning values in the pairwise comparison matrix, there are chances of inconsistent
feedback by the decision-maker. For instance, in our example, the DM deems price to be a more important
criterion than camera and camera to be more important than aesthetics. However, if during the pairwise
evaluation, the DM assigns same values to price and aesthetics, it would be considered as an inconsistent
evaluation.

To address this concern, AHP provides a measure called consistency ratio (CR).

The following steps are used for calculating CR:

Step 4.1: Take the original pairwise comparison values that are not normalized (Table 4A). Shift the criteria
weights to columns against the respective criteria, i.e., price against price, as shown in Table 5A.

Table 5A
Re-arrangement of criteria weights

Criteria weights 0.6034 0.1364 0.1957 0.0646

Price Aesthetics Camera RAM
Price 1.00 5.00 4.00 7.00
Aesthetics 0.20 1.00 0.50 3.00
Camera 0.25 2.00 1.00 3.00
RAM 0.14 0.33 0.33 1.00

Step 4.2: Multiply each value in the column with the respective criteria weights, as shown in Table 5B. Take the
sum of each row to calculate weighted sum value.

Table 5B
Derivation of weighted pairwise comparison values and weighted sum value

Weighted sum
Criteria weights 0.6034 0.1364 0.1957 0.0646
value
Price Aesthetics Camera RAM Sum of row values
Price 1 * 0.6034 5 * 0.1364 4 * 0.1957 7 * 0.0646 2.5204
Aesthetics 0.2 * 0.6034 1 * 0.1364 0.5 * 0.1957 3 * 0.0646 0.5487
Camera 0.25 * 0.6034 2 * 0.1364 1 * 0.1957 3 * 0.0646 0.8132
RAM 0.14 * 0.6034 0.33 * 0.1364 0.33 * 0.1957 1 * 0.0646 0.2587

Page 5 of 9

This document is authorized for use only in Prof. Suvechcha Sengupta's MBA Business Analytics Minor Decision analytics 24.11.2023 at Somaiya Vidyavihar from Nov 2023 to May 2024.
 Step-by-Step Tutorial of Analytical Hierarchy Process with an Example

Step 4.3: Calculate the ratio of weighted sum and criteria weights for each row as shown in Table 5C.

Table 5C
Computation of the ratio of weighted sum to criteria weights

Ratio of weighted
Weighted Criteria
Price Aesthetics Camera RAM sum and criteria
sum value weights
weights
= 2.5204 / 0.6034
Price 0.6034 0.6820 0.7828 0.4522 2.5204 0.6034
= 4.1770
= 0.5487 / 0.1364
Aesthetics 0.1207 0.1364 0.0979 0.1938 0.5487 0.1364
= 4.0229
= 0.8132 / 0.1957
Camera 0.1509 0.2728 0.1957 0.1938 0.8132 0.1957
= 4.1551
= 0.2587 / 0.0646
RAM 0.0845 0.0450 0.0646 0.0646 0.2587 0.0646
= 4.0042

Step 4.4: Now, calculate eigenvalue of the matrix (λmax) by averaging the ratio values

Formula - Eigen value of the matrix:

λmax = Average of (weighted sum/criteria weight)

λmax = (4.1770 + 4.0229 + 4.1551 + 4.0042)/4 = 4.0900

Step 4.5: Calculate the consistency index using the formula:

Consistency Index (CI) = (λmax-n)/(n-1), where n is the number of criterion

Since the number of criterion in our case is 4, n=4.

Consistency Index (CI) = (λmax - n)/(n - 1) = (4.0900 - 4)/(4 - 1) = 0.0300

Step 4.5: Finally, calculate the consistency ratio (CR) using the formula:

CR = Consistency Index / RI,

where RI = Random Index, which is nothing but the average consistency index obtained from a large set of
matrices of different sizes and with random entries. Table 6 shows one such estimation of RI values, generated for
matrixes with size of up to 10.
CR is a rescaled value of CI with respect to RI.

Table 6
Values of the random index (RI) for randomly generated pairwise matrices

N 2 3 4 5 6 7 8 9 10
RI 0 0.58 0.9 1.12 1.24 1.32 1.41 1.45 1.51

Since in our example n = 4, the corresponding RI = 0.90

Therefore, CR = 0.0300/0.90 = 0.0333

Page 6 of 9

This document is authorized for use only in Prof. Suvechcha Sengupta's MBA Business Analytics Minor Decision analytics 24.11.2023 at Somaiya Vidyavihar from Nov 2023 to May 2024.
 Step-by-Step Tutorial of Analytical Hierarchy Process with an Example

Note: If CR < 0.10, this means that the comparison matrix is reasonably consistent and AHP can be used to take
decision.

If CR > 0.10, this means that the comparison matrix is inconsistent and cannot be used in AHP to take a decision.
The DM would have to go back and check the pairwise comparison matrix to identify where the problem lies,
correct it and re-compute CR using the above method. Only when CR < 0.10 can the DM continue with the next
steps to arrive at a decision.

In our example, as CR = 0.0333 < 0.10, we can say the pairwise comparison matrix we arrived at is consistent and
we can continue the decision making using the criteria weightages.

Step 5: Evaluate alternatives according to weighting

Once we confirm that our pairwise comparison matrix is consistent, we can arrive at the relative rank of
importance for each criterion using the final criteria weights as shown in Table 7.

Table 7
Final criteria weights and relative rank of importance

Criteria weights Criteria weights (%) Rank

Price 0.6034 60.34% 1
Aesthetics 0.1364 13.64% 3
Camera 0.1957 19.57% 2
RAM 0.0646 6.46% 4

Note that once the weights of each criterion have been thus computed, the DM might still have preferences
among the sub-criteria, if they are present. For example, if there were options available under the aesthetics
dimension (such as Matte Black, Shiny Blue, Shiny Grey, Golden Pink and Glaze Black), the DM might prefer Golden
Pink over the rest. In such instances, Steps 3 and 4 can be repeated to make pairwise comparison of these sub-
criteria to calculate the sub-criteria weights. For the purpose of illustration, sample sub-criteria weights are taken
in Table 8, without showing the Step 3 and 4 calculations.

Table 8
Illustrative values for sub-criteria

Camera
Price - Aesthetics RAM -
Quality -
Mobile Sub- - Sub- Camera Sub-
Price Aesthetics Sub- RAM
Name criteria criteria Quality criteria
criteria
Weight Weight Weight
Weight
Matte
Mobile A Rs. 31999 25% 16% 12MP 10% 6GB 15%
Black 5.9”
Shiny Blue
Mobile B Rs. 32999 23% 20% 64MP 23% 12GB 25%
6.2”
Shiny
Mobile C Rs. 36999 21% 20% 48MP 21% 8GB 20%
Grey 6.0”
Golden
Mobile D Rs. 48999 12% 24% 108MP 25% 8GB 20%
Pink 6.2”
Glaze
Mobile E Rs. 37999 19% 20% 48MP 21% 8GB 20%
Black 6.0”

Page 7 of 9

This document is authorized for use only in Prof. Suvechcha Sengupta's MBA Business Analytics Minor Decision analytics 24.11.2023 at Somaiya Vidyavihar from Nov 2023 to May 2024.
 Step-by-Step Tutorial of Analytical Hierarchy Process with an Example

Next, the sub-criteria are weighted according to the main criteria, as shown in Table 9. The weighted sub-criteria
values add up to the criteria weights calculated earlier. For example, criteria weight of “Price” is 60.34% and the
sub-criteria values for the five alternatives add up to this value.

Table 9
Sub-criteria weighted according to main criteria

Camera Camera
Price - Price - Aesthetics Aesthetics RAM - RAM -
Quality - Quality -
Mobile Sub- Sub- - Sub- - Sub- Sub- Sub-
Sub- Sub-
Name criteria criteria criteria criteria criteria criteria
criteria criteria
Weight Weighted Weight Weighted Weight Weighted
Weight Weighted
Mobile A 25% 15.09% 16% 2.21% 10% 1.96% 15% 0.97%
Mobile B 23% 13.88% 20% 2.76% 23% 4.50% 25% 1.62%
Mobile C 21% 12.67% 20% 2.72% 21% 4.11% 20% 1.29%
Mobile D 12% 7.24% 24% 3.21% 25% 4.89% 20% 1.29%
Mobile E 19% 11.46% 20% 2.74% 21% 4.11% 20% 1.29%
Sum
(Main
60.34% 13.64% 19.57% 6.46%
criteria
value)

Finally, the weighted sub-criteria values can be added for each alternative to arrive at the final weightages and the
corresponding ranking as shown in Table 10. Based on the ranking in Table 10, Mobile B appears to be the best
option for you as per your chosen criteria.

Table 10
Final weightages and ranking arrived at using pairwise comparisons of different criteria

Mobile Price - Sub- Aesthetics - Camera RAM - Sub- Final Ranking

Name criteria Sub-criteria Quality - criteria weightage
Weighted Weighted Sub-criteria Weighted
Weighted
Mobile A 15.09% 2.21% 1.96% 0.97% 20.2% 3
Mobile B 13.88% 2.76% 4.50% 1.62% 22.8% 1
Mobile C 12.67% 2.72% 4.11% 1.29% 20.8% 2
Mobile D 7.24% 3.21% 4.89% 1.29% 16.6% 5
Mobile E 11.46% 2.74% 4.11% 1.29% 19.6% 4
Main criteria 60.34% 13.64% 19.57% 6.46% 100.0%
value

In this manner, AHP enables the decision-maker to evaluate a complex set of alternatives with multiple criteria at
various levels and arrive at an optimal decision. The final ranking also provides the necessary information to the
decision maker in terms of the other choices to follow, if the optimal option is found to be either not suitable or is
out of stock.

Page 8 of 9

This document is authorized for use only in Prof. Suvechcha Sengupta's MBA Business Analytics Minor Decision analytics 24.11.2023 at Somaiya Vidyavihar from Nov 2023 to May 2024.
 Step-by-Step Tutorial of Analytical Hierarchy Process with an Example

References

Thomas L. Saaty, How to make a decision: The analytic hierarchy process, European Journal of Operational
Research, 1990, 48(1), 9-26; https://doi.org/10.1016/0377-2217(90)90057-I

Thomas L. Saaty, Decision making with the analytic hierarchy process, E International Journal of Services Sciences,
2008, 1, (1); https://www.inderscienceonline.com/doi/abs/10.1504/IJSSci.2008.01759

Brunelli, Matteo. 2015. Introduction to the analytic hierarchy process. Springer Briefs in Operations Research. P.
83. 978-3-319-12502-2 (electronic). 10.1007/978-3-319-12502-2.

Page 9 of 9

This document is authorized for use only in Prof. Suvechcha Sengupta's MBA Business Analytics Minor Decision analytics 24.11.2023 at Somaiya Vidyavihar from Nov 2023 to May 2024.