JTBD Analysis, What It Is and How to Do It
This is the third story in a series of introductions to JTBD.
JTBD Analysis, What It Is and How to Do It
An example of the JTBD Opportunity (Importance/Satisfaction) Chart as well as an example of the business value and JTBD report table is at the end of this chapter.
What is a JTBD Analysis?
“Jobs Theory provides a powerful way of understanding the causal mechanism of customer behavior”
— Clayton Christensen et al.
A Jobs-to-be-Done Analysis or simply a JTBD Analysis (originally in 1960s defined as a Segmentation Analysis) is an interdisciplinary methodology rooted in behavioral economics that is used for observational causal inference to understand what drives economic decisions, such as purchasing or voting behavior, in given situation.
Unlike traditional market, customer or policy research, JTBD Analysis uses a comprehensive approach to identify why customers make specific choices.
Causal inference in JTBD Analysis relies on observations and quasi-experiments rather than controlled experiments. The process involves identifying distinct customer segments retrospectively, based on a common need, and then profiling these segments to understand their unique characteristics. Profiling data—including circumstances, opportunity, ability, place, time, demographics, psychometrics, and firmographics—serve as independent variables, while needs, outcomes, and the progress people seek represent dependent variables.
JTBD Analysis integrates qualitative and quantitative approaches, employing diverse data collection methods like interviews, surveys, observations, ethnographic studies, and natural experiments. This is complemented by sophisticated data analysis techniques, including thematic, statistical, cluster, and factor analysis, to derive comprehensive insights into customer behavior.
This approach enables teams to transition from qualitative insights to quantitative market segments, identifying business or public opportunities and actionable strategies with a success rate of 90%. Currently, JTBD Analysis is recognized as one of the most effective methodologies for predicting outcomes in business, public policies, and innovation, offering the highest ROI with the lowest risk.
It is important to remember that JTBD Analysis, like any research methodology in social sciences, is only the first step in the research process—the discovery phase. However, the process doesn't stop here and should continue into real-world controlled experiments to determine causality with the maximum level of confidence and business or social impact. For example, the results of the JTBD Report can be implemented as an A/B test to verify if the identified profiling data and circumstances will trigger the same needs from the users, and will yield expected business or public outcomes, such as more conversions or votes in elections.
The objective of the JTBD Analysis is causal inference
JTBD Analysis is not a Task Analysis nor an observation of any existing process or journey alone.
A Job-to-be-done is a progress towards an outcome that a person has in given circumstances. It is solution and product agnostic. It exists already, demand is there, even if the current state of the world doesn’t allow a person to make such a progress.
Circumstances is what many organizations and professionals don’t pay attention to.
Until there is any causal inference, an aim to determine what causes the behavior, and not what it is, what people are doing and their needs and features alone, the work performed by a team or a professional is a Task Analysis and not a JTBD Analysis. Understanding the need is only one bucket, a vertical axis (Y). Science always requires two buckets and then the causal inference. What are circumstances, a horizontal axis (X)? And finally which X causes which Y? Answering the last question is the most complicated part and the ultimate objective of the JTBD Analysis.
JTBD always is a desire, not a mere feature. However, people can satisfy that desire, unrealized aspiration or avoid the struggle only by interacting with the physical world, e.g. using a specific product or a service, or by outsourcing it, e.g. believing that voting for a particular political party will achieve the same goal. JTBD analysis is more practical-philosophical in the sense that the main goal is to learn about all the “Why’s” in a way that benefits businesses or societies significantly.
Think of a JTBD analysis as a best friend and a doctor who tries to listen actively, understand the root cause of your problems, and help you, rather than just someone who records what you say, and helps you summarize your thoughts. Someone who gives you simply more of what you are doing right now versus giving you a better way to satisfy your true desire, and that could be through a totally different product.
Think about your own life and experience. You probably can quickly think of something that you don’t like and you wish you could finally satisfy it, however, maybe there is nothing at all at the moment to help you with that, or you probably are using something in an unusual way to help you with that, but still in a bad way.
Imagine you could read the minds of the people and learn these stories, understand their experiences, and know deeper secrets from the entire population. Imagine a magic wand that would tell you something like - millions of people really suffer with X and they need Y so much that they would buy it instantly. This is what JTBD analysis is about. It is this magic wand.
JTBD analysis is the most efficient methodology for putting exploratory continuous discovery into practice or identifying strong hypotheses for controlled experiments in the most successful way.
JTBD analysis can be performed in a week by a solo founder alone at the beginning of a birth of a new startup, as well as by a more experienced team at larger corporations or governments within an existing process. It can be done externally and internally. At any stage, and in any industry.
Examples of use-cases for a JTBD Analysis
JTBD analysis can be used to identify endless possibilities and insights, such as, but not limited to:
Align on what a need is and identify customer or citizen needs and measurable outcomes (p-metrics);
Discover and prioritize business or public opportunities;
Segment markets, knowing precisely what makes them different and why;
Marketing, improve SEO, CRO and organic growth through copy that resonates with users’ circumstances and intent;
Venture design and successfully launching new startups with evidence-based strategies;
Optimize operations and enhance internal processes, productivity, and employee experiences. Align sales, marketing, research and product;
Get more people supporting your policy or win elections;
Thrive in competitive markets;
Dramatically improve UX and customer satisfaction (CSAT);
Foster customer loyalty, turning users into brand advocates;
Disruptive innovation or uncover hidden segments to identify and address previously unrecognized market segments;
What skills and attributes are required to perform a JTBD analysis?
Hence the skills required to perform JTBD analysis are way more than just UX research or marketing or interviews. JTBD analysis is a multi-functional E2E activity that requires either an experienced consultant or an entire team. The person performing or leading the JTBD analysis at a bare minimum must have those skills and attributes:
Solid knowledge of analytic philosophy - Because JTBD analysis is all about logic, understanding why, and looking at something that can’t be seen. It includes argumentation, and its components such as conclusion, premises, and assumptions; then fallacies, lacunas, cause and effect, and type I and II errors. It includes meta-analysis and synthesis.
Strong empathy, curiosity, open-mindedness, and communication - Like the previous point, this is also non-negotiable. JTBD analysis requires a mindset of an explorer, a connector who can deeper understand customers and their true desires.
Basic knowledge of psychology, environment, and behavioral science, including biases, disabilities, addictions, personality type theories, brain structure, behaviors, values, attitudes, beliefs, religions, and even knowledge of the culture and history of the target population. For example, in order to perform a JTBD analysis in China, you have to be very familiar with the local culture, and the way of thinking and behaving there which is totally different from the next example. Or if you are performing a JTBD analysis with American corporate executive leaders, you have to have a totally different mindset.
Basic experience with qualitative research and JTBD interviews and job stories
Basic knowledge of statistics and science, including the scientific method and hypothesis testing
Basic experience with quantitative research, JTBD survey design, data analysis, cluster analysis, factor analysis, profiling, and some knowledge of SQL and tools like Python/R/SPSS.
Basic knowledge of economics, management consulting, sales, and marketing, including bottom-up market sizing and research and pre-sales, monetization, and business modeling - Without a commercial mindset and ability to connect business and customers, there is no JTBD analysis but simply either UX research or market research. The report has to be made with the recommended next steps and to be presented to stakeholders in a management consulting fashion.
Optionally, depending on the industry and type of products the organization is offering and plans to offer - being familiar with how the product might look like, function, and help the target population get the job done, especially if it’s a very scientific or unusual product. The best way to empathize with customers is to put yourself into their shoes, literally. And the best way to do that is by being like them and understanding how they work already.
Different levels of the JTBD Analysis
Basic JTBD Analysis - quickly summarizing a small number of interviews and survey findings with averages. Everyone can quickly do it and easily understand the results.
Advanced JTBD Analysis - calculating percentage and the difference and forming basic tiers of customers manually. Still, everyone can do it, for example, in Google Spreadsheets with a bit more time and advanced queries. This particular story at the maximum explains this level and how you could do it yourself. The final level requires at least basic scientific, data, and statistics knowledge and experience with the relevant libraries, tools, or languages like Python. However, now it is possible to perform scientific analysis with modern end-to-end discovery platforms like ATHENNO, easily, without being a scientist.
Scientific JTBD Analysis - using all methods, including quasi-experimental and as a part of controlled experiments to achieve the highest level of confidence and causal inference. This is where you can discover all sorts of unknown patterns, and spot anomalies in order to test hypotheses or check assumptions.
How to perform JTBD Analysis and when it ends?
First, conduct JTBD interviews and define personal outcome metrics (Read more about that in the first story in the series)
Then create and send a JTBD survey, and collect as many responses, as you can (More about that in the second story in the series)
Optionally look into other data sources, such as product analytics, historical data to collect as much profiling data and needs data as possible.
Start analyzing results:
Calculate Opportunity Score
Prioritize needs based on Opportunity Score
If you have more data (at least 70-90+ responses):
Perform Cluster (and other) analysis to segment respondents into groups (clusters) based on similar rates of Importance and Satisfaction
Profile segments. The most important part of the JTBD analysis is to determine what is causing respondents in one segment to have different needs or to struggle more/differently than others. For example, it could be due to disabilities, health, lifestyle, financial situation, environment, interests, and many other unknown factors.
Finally, put anything together and define clear market segments as a “group of people” (with the same characteristics) + “job they try to get done” where each segment has clear JTBDs and personal metrics they need the most; What they have in common; Combine it with basic market research and do bottom-up market sizing. Now you have the most powerful insight in front of your eyes from operational all the way to strategic view, and you clearly know what to do next.
In scientific research, you very often need to perform, what’s called a dimensionality reduction, which means, representing multiple scales with just one number in order to make it easier for everyone. Researchers can even plot insights on the traditional 2D chart humans can read. Executives can easily interpret results.
Sometimes, if you just have 2 scales, you can generalize both of them with just one number without doing any reduction.
In JTBD analysis and surveys, this one number is Opportunity Score.
JTBD analysis ends with:
Calculating an Opportunity Score and plotting Opportunity Map Scatterplot
Forming specific markets and segments - advanced analysis (In this case Opportunity Map can be plotted for different segments individually)
Though, for the scientific JTBD analysis you have to continue the process by implementing the insights as experiments in the real world.
How many responses do I need for a JTBD survey?
Not as many, as you might think. JTBD analysis can be done quickly by one person with a small group of people as successfully as by a team at a large corporation with thousands of respondents. No matter the stage of the business.
Start by collecting at least 20-30 responses. Opportunity Score here already will help you understand which needs are more or less important and to what degree they are satisfied.
You can’t do any cluster or advanced analysis with that amount, but you might not need to. If you are starting a new startup, you can start by understanding where to focus more and that’s it. Detailed clustering and segmentation are usually more valuable for companies at a later stage to help them improve existing products and services.
If you want to perform cluster and/or factor analysis, aim for at least 70-90 bare minimum responses. New startups with no traction and no pre-sales or community can do that by using platforms where you can pay to recruit participants quickly.
That amount of respondents can help you to start seeing different segments.
Of course, in a most ideal situation, and for the best statistical significance results get as closer to 250-350 responses as you can.
So, in conclusion:
20-30 (Low fidelity) - start here. Till this point use the average score to represent your findings and now you can start applying %.
70-90 (Mid fidelity) - if you can, get here and this is where you can start doing cluster and/or factor analysis. At least divide your data set (respondents or customers) into two clusters (groups): a) those who responded positively to the majority of questions and are more likely to use/buy from you and b) those who responded negatively or did not respond to the most questions at all and won’t be interested in using or buying your offerings. Definitely from now on use only % and not average score here (the difference is explained below under the Opportunity Score section)
250-350 (High fidelity) - if you are a larger company, ideally you need to get here for maximum statistical significance. However, there is no need to survey beyond that and thousands of people since results won’t change much (until you are surveying all kinds of consumers in a very large market or doing public/gov research).
Don’t be afraid to take it gradually. Better to start with small research and low fidelity with 20 respondents in order to refine your next steps and focus (or even find this focus if you are a new startup), and only then update research and JTBD parameters and continue to the next level of fidelity.
Remember from the first story in the series, that only you define this journey-to-be-done and its structure. In a similar way here, you can go level up or down whenever you need with surveys and JTBD analysis.
Calculating Opportunity Score or how to analyze JTBD survey results?
Let’s recall that in the previous story in the series, we created a JTBD Survey. Each JTBD and personal metric is evaluated on 2 scales - importance and satisfaction.
Now, we can represent both importance and satisfaction with just one number - opportunity score.
Opportunity score = Importance + MAX ( Importance - Satisfaction ; 0 )
In a simpler way, it means Importance x 2 times and minus Satisfaction.
However, if Importance minus satisfaction is negative, then it is just Importance.
In other words, the opportunity score is a weighted difference with a higher weight in importance.
Simple (Average) Opportunity Score & Individual Opportunity Score
That brings us to a total scale of between 1 and 9 (Since we surveyed Importance and Satisfaction between 1-5) for the Simple Opportunity Score or Average Opportunity Score.
For basic JTBD analysis that’s it. You can quickly calculate averages and simple average opportunity scores.
However, you can treat it as an Individual Opportunity Score if you merge both answers together for each respondent. I will stress here again “for each”.
Later, you can segment respondents based on their original answers to both Importance and Satisfaction, however, you can also segment them based on the merged Individual Opportunity Score instead of two numbers. That will be important for a scientific JTBD analysis.
Percentage Opportunity Score and Top Leads/Contacts
For advanced (and for scientific) JTBD analysis, it works in a bit different way.
Usually, whenever we hear and talk about Opportunity Score, we mean Percentage one and the difference between those percentages.
This is calculated from the % of people who said that a particular need is very important to them AND at the same time % of people who are satisfied where we look at a smaller level of satisfaction for important needs. So we look into the HIGHER % who said Importance=4,5 and LOWER % who said Satisfaction=4,5.
In the illustration above think of Opportunity Score as a flag area. You are looking for a bigger flag.
We can flip it around and make it % of people who said Importance=4,5 AND Satisfaction=1,2,3. Those people become top leads or top contacts.
That way it is also easy to present findings to the team, stakeholders, executives, or client. By saying something like “85% of our users consider X very important, however, only 15% are satisfied with it, which yields the highest opportunity score for us”.
So, in order to calculate the Percentage Opportunity Score or just Opportunity Score (or for short OppScore):
For each JTBD/metric, create 5 columns for Importance to count # of times people answered with 1, 2, 3, 4, or 5.
Do the same for Satisfaction with another 5 columns.
Now calculate % of people who said Importance=4 or 5. Which is (COUNT(importance4)+COUNT(importance5))/total_responses * 100
Do the same for % of people who said Satisfaction=4 or 5. Which is (COUNT(satisfaction4)+COUNT(satisfaction5))/total_responses * 100
Finally, apply the Opportunity Score formula to those percentages.
And normalize results. Tony Ulwick (author of the opportunity score formula) normalizes percentages to 10 instead of 100, and Opportunity score on a scale between 0-20, hence anything above 12 is very good. If you were looking only at Average Opportunity Score between 1-9 or a percentage opportunity score normalized between 0-10 then 6 or above is a good score. However, I would recommend normalizing everything always to a traditional scale of 0-10, or to 0-100%.
Percentage Opportunity Score = (
(% of people for whom it’s important (=4 or 5))
+
MAX (
(% of people for whom it’s important (=4 or 5)) — (% of people who are satisfied (=4 or 5))
; 0 )
)
* 5 (to normalize it between 0-10)
In the example above with 85% importance and 15% satisfaction, we get 85% + 85% - 15% = 155%. Normalized to a scale of 20, it becomes 1.55 * 10 = 15.5 (If you are using Tony Ulwick’s scale or the Outcome-Driven Innovation process).
However, it is recommended to keep everything consistent and simple, hence the final score on a scale of 0-10 becomes 1.55 * 5 = 7.75 out of 10.
Keep in mind, that this is for percentages as percentages, meaning that 85% = 0.85. If you are doing math with numbers without percentages, then instead you might need to divide by 10 or 20 depending on the scale you want.
Last but not least, Opportunity Score alone for business (and not just people or products) is useless. It only helps you to answer the question - how risky or how likely is this opportunity to succeed if we will address it? So in that sense sometimes it is worth representing Opportunity Score as a certain probability % of success (which can be used in decision trees and EMV calculations)
Hence, in the end, we get a 77.5% probability of success (or if flipped around - the decision will have only 22.5% risk)
Beyond advanced JTBD analysis - with statistical JTBD analysis, clustering, and factor analysis, merged together with market and competitive research, and business modeling, assessed at the same time with the team’s capabilities to address this opportunity x10 better than anyone else - this risk % can be reduced even lower.
In future stories, the Innovation Triangle, Innovation Sprint, and Market Score will be introduced for that purpose. Which will contain the main Impact/ROI metric.
All-in-one modern E2E and free platforms like ATHENNO bring all that together with just one ATHENNO Score or Main Score.
But let’s go back to a JTBD analysis…
Different tiers of customers, identifying champions, and evidence of future payments
Optionally, you can also sort respondents even on the degree of urgency, interest, or loyalty. While you want to focus on the personal outcome metrics and needs that are Important 4 or 5 and Satisfied 1, 2, or 3 at the same time for most people (top leads/contacts), you sometimes might want to know how critical a specific need is and how big might be your potential fan base?
Tier 1 respondents are only respondents for whom specific need is both important only at the level of 5 and satisfaction is at the level of only 1.
You could continue this with as many tiers as you wish, usually up to 3.
While clustering and segments separate respondents into groups. Different tiers help you to organize each segment and identify potential champions.
Sometimes it is a better decision to focus only on a segment that needs something more than anyone else and suffers the most. This can also be called a tier who is “pissed off”. If you will address the needs of exactly those people, they for sure will pay you, will love you, and will talk to you for free so you could finally help them with their biggest pains which they can’t satisfy anywhere else at all.
In JTBD analysis the best way to have evidence of future payments is to look exactly at Tier 1 of respondents.
You can also add questions to your JTBD survey like “How much are you paying now to get this job done?” or “How much are you willing to pay to get this job done perfectly?”. However, usually, these questions often don’t add any extra insight. Just filtering responses by Tiers might be enough.
How to visualize JTBD analysis results using Opportunity Map
An Opportunity Map (or Opportunity Landscape) is the most common visual representation in a form of a scatterplot chart that presents an important part of the JTBD analysis and research and allows anyone to see the best and worst opportunities themselves. This is also one of the main deliverables of the JTBD analysis.
Importance is, well, important. Opportunity score puts a higher weight on Importance. This is because there might be many people who responded with average importance, and let's say 4 satisfaction. This will skew results a lot if you are not using the correct formula and don’t do scientific analysis.
In Opportunity Map: X-axis always represents Importance and Y - satisfaction. Either as an average of 1-5 or as a percentage.
By looking at this chart anyone can quickly see what type of JTBD analysis was performed. For small teams and new startups at the beginning or just surveys with a very small number of respondents - this often will be just average. With more data and for larger teams this always will be a percentage.
Never use percentages for a small amount of data! If you increased revenues from $100 to $600, it’s not a dramatic 500% growth rate, it’s just a tiny +$500. If you only did basic qualitative UX research and interviewed 5 people, then the average representation will be better. Saying that 80% want X when you talked only to 5 people will leave a very bad impression on executives or investors. Instead of doing personas, just do traditional synthesis and summarize findings from all interviews in the most basic statistical way possible, then scale your JTBD research, journey, and metrics, and only then create a survey, collect more data and now you can use %.
The opposite problem is true as well. Never use the average for larger amounts of data because the average will bring more results to the center of the opportunity space. This is where you need to use % and with even more data - cluster it and have a separate diagram and summary for each cluster (segment).
Below, are both examples of a JTBD Opportunity Map:
The problem with advanced JTBD analysis and why scientific JTBD analysis is required
Teams and organizations shall always aim to perform a scientific JTBD analysis as often as they can. Only due to limited resources or time constraints, you may just do a quick advanced research that can provide enough results to make certain decisions, for example, if you are looking to build a totally new startup and just have an idea, or you have a small agency or individual consultant, working with few clients who need results in few weeks.
The main problem with the advanced JTBD analysis and Percentage Opportunity Score and the example above is that because 85% said important and only 15% are satisfied, it doesn’t mean that 85% said important AND dissatisfied at the same time, or some smaller % in the middle, or 15%.
Mathematically speaking, in this particular example, it is possible that 85% who said important and 15% who said satisfied are totally 2 different groups and in reality, 0 people said important and dissatisfied at the same time. 85 + 15 = 100, so there is a probability that there are no overlaps whatsoever.
This is why exactly it is critically important to look into data holistically.
And this is where even if you can do some of that in the spreadsheets to some extent, it will take way too much time with many custom queries and lookups again and again until some decent insights will be uncovered.
Even answering the most basic JTBD analysis question - How many % need X (where need means - BOTH important AND dissatisfied AT THE SAME time)…
Is not as easy as it seems. Many will rush into the JTBD analysis conclusion and will provide very misleading insights to the team. However, this is where even basic cluster analysis can help researchers and organizations save hours, days, and even weeks of manual work.
Finally and as previously mentioned, it is critical to remember that the ultimate goal of the JTBD Analysis is the causal inference, i.e. to determine what causes customer behavior. Any report based on interviews and surveys alone is just a first step in the scientific and business research process that can help form strong hypotheses, but it doesn’t stop here. Scientific JTBD analysis should continue into real-world experiments, such as A/B tests, fake door tests, Concierge MVP tests, etc. Businesses must make revenues, therefore the discovery process can only be concluded with at least some evidence that people themselves convert or click “Buy”, and not just tell you that during interviews or surveys.
Final JTBD Report - Intro to scientific JTBD analysis with cluster & factor analysis for business value & identifying specific market segments
Scientific JTBD analysis is outside the scope of this story. But let’s quickly look into how it looks in the end and what value businesses and product teams get out of it.
As mentioned in a previous section, statistical JTBD analysis allows for answering core JTBD questions more precisely. For example, we could confidently say something like “65% of people need X where they said important AND dissatisfied at the same time”.
Then cluster analysis is a way to automatically group respondents (rows in a spreadsheet or a table) into groups based on the similar way they responded to a survey.
Factor analysis, on the other hand, helps to group questions, JTBDs, and personal outcome metrics (columns in a spreadsheet or a table) together based on the same unknown underlying factor. This can be applied to reduce # of questions in a survey or features. It can be used to merge different needs together in a unique way and create something new, to innovate. It can be used to recommend other features to a user that this particular user is more likely to use, but is not aware of. Even multiple surveys or different research techniques, marketing campaigns, product strategies and so much more can be analyzed together in order to find a common denominator that actually is a main driver/lever for business outcomes and customer success.
One important aspect of a JTBD analysis is that JTBD always shall be the unit of analysis, and never a person or any personal questions, psychometrics. And solution also shall never be part of the JTBD analysis. Only the job at all times is the main unit of analysis. People are segmented together based on the same needs, not their demographics or psychometrics, or any other factors.
Only after a cluster analysis is performed and clear segments are identified. Only then “profiling” is performed to help teams understand what actually makes different people be in the same segment and have exactly the same need, or why they struggle the most when others - don’t.
This is where clear markets & segments (real business “personas”) are formed as a “group of people” (with similar characteristics) + “job they try to get done” (with exactly the same personal outcome metrics they need more than other segments)
The Opportunity Map is created per each cluster separately rather than one Opportunity Map with all respondents messed up together. The average customer doesn’t exist. With more data, you can uncover hidden segments, niches, and unmet needs very successfully.
Most teams need to address the needs of the majority in order to increase profits. However, at a much later stage large organizations might not be able to uncover many needs which are important and dissatisfied for the majority. Scientific JTBD analysis becomes a crucial technique later where unknown segments even within the existing customer base can be discovered, and then the need becomes a need of the majority again, but in this case for this particular segment.
JTBD analysis provides a way of identifying strong scientific hypothesis and sampling data that can be used to attract the right audience and create impactful experiments that yield real world business or social impact.
The ability to move quickly, diversify, address the needs of different segments, and personalize in the modern fast-paced world are critical components for business success and growth.
Below, is an example of what final JTBD analysis results (called a JTBD Report) usually look like. It is based on the “Outcome-Based Segmentation” originally developed by Tony Ulwick in the early 1990s.
By just looking at it, you might see why exactly JTBD analysis is the most powerful technique today that drives discovery and innovation globally better than anything else.
Marketing knows who exactly to target, why, and how to speak to different audiences and channels.
The product knows exactly who is the customer, and a user, what features to design and to build, and how they will be used.
Sales know who are main champions and how to talk to prospects better.
Executives, startup founders, and investors get a higher ROI with the lowest risks, know the market size, and growth rate, and have proof of payments, usage, demand, and what action to take with confidence before a solution even exists.
Innovation is not random, chaos or luck. Building products quickly and seeing what happens yields lower results and costs too much.
Innovation is a predictable, measurable, scientific and business process. Always remember the example of you holding an object in your hand above the ground and releasing, dropping it. If you know the laws and science, you know 100% what will happen next and what key drivers or levers that connect cause and effect are.
The future is predictable. The choice is yours.
Generating JTBD analysis results automatically
As you have seen JTBD analysis is the end-to-end process and creating and adjusting so many tools, calculating different scores, and using different algorithms even at the advanced level in a spreadsheet is very time-consuming work. Plus you might need to test it to make sure everything is correct. Adjusting it to different use-cases and surveys is another challenge.
In addition, performing scientific JTBD analysis usually requires hiring an expensive consultant and a data scientist.
However, what if you could just generate all those charts and segments automatically in a second without doing anything?
In ATHENNO, the world’s most advanced all-in-one continuous discovery, JTBD, Outcome-Driven Innovation (ODI), startup management platform and tool, you can do exactly that!
Just create your journey, and break it down into personal outcome metrics.
Then generate a survey, and publish it. Send the link. Collect responses.
And that’s it. You don’t need to do anything!
In ATHENNO, as soon as you get the first response, you will unlock all those powerful survey insights, and the more responses you get, the more insights you will be unlocking.
Save not just days, but weeks, or even months of manual work for your entire team today!
Common JTBD analysis mistakes and myths
Asking just one question or too few questions.
JTBD analysis is not a CSAT survey or any other technique where you ask only one question. It doesn’t cost you anything to have more questions upfront so you could collect the data that you need in order to explore it and uncover findings. With fewer questions, it might be hard (and sometimes not even possible) to test many assumptions, form clusters, and discover underlying factors.
Not having something real, measurable, like a hypothetical or existing solution and goal upfront
Not having any solution or something real, physical, or practical in the real world that people can use in order to satisfy their desires in the surveys. Even if you are at an early stage and don’t have anything yet - those are assumptions or any other existing solutions and competitors. People can only satisfy their desires by interacting with the world around them through some senses, experiences, products, or services. Asking people, for example, “How satisfied are you with your mental state” is totally worthless.
Have a clear goal in mind before doing JTBD analysis. In needs and questions, you are about to ask and analyze, is there enough information there to be of any help? Randomly shooting around in different directions without knowing upfront if you want to improve a particular product or discover something totally new might be too expensive.
Instead of the example above, if you are working on a mental health app, you could ask something like “When using a <solution/your app> how satisfied are you with how quickly you fall asleep”.
Break down your features and possibilities into many personal metrics in order to uncover where exactly to improve your product and gradually help people get the job done better until they can do it perfectly. Or at least organize an immersion session and talk to a few domain experts who will help you to break the journey-to-be-done into little blocks.
Not breaking JTBD down into personal metrics (p-metrics)
In relation to the previous point. Remember that we are evaluating JTBD on two scales - Importance and Satisfaction. Here specifically I am talking about satisfaction. It can only be experienced in relation to something real in the world. Even if it is the absence of it.
If you are just asking people how satisfied they are with the core JTBD itself, rather than breaking it down into specific personal outcome metrics that they use to interact with the physical world and how they measure success, you will get way too abstract and low-quality results and even anomalies.
Even if you really are analyzing one simple job-to-be-done. From previous stories in the series, you have seen how it is important to break it down and uncover personal metrics. And when you create a survey, you survey people around these little p-metrics, not the job-to-be-done or something large itself.
People can only satisfy their needs through these personal outcome metrics and it is the most critically important part of the JTBD analysis to get it right.
Instead, try to have many questions like “How satisfied are you with how you regenerate your mental state using meditation in the morning”.
JTBD analysis is all about mathematical precision and granularity. Something called a Mathematical Culture. It’s about achieving the right result, and unfortunately, it doesn’t matter at all how hard you have tried and how much time you have spent trying to solve the problem. In data science, people also often say: “Garbage in - garbage out.”
But don’t let it discourage you. Just be as specific and granular as you can. That’s THE whole point of the JTBD analysis. Start simple, and slow with a smaller sample size. Learn. Adjust. Continue with more people. Repeat. “Lean JTBD analysis” in a sense.
Not adding high-quality profiling (personal) questions besides JTBD questions
While JTBD is the unit of analysis. After the main analysis is done, if there is no qualitative data to understand what makes different people be part of different segments and have different needs or struggle in a different way, then teams will only end up with a fancy scientific table, but businesses won’t be able to target those segments at all. Of course, product teams can still develop the solution and then figure out later who the most loyal customers for each solution are, but that won’t be much different from the current approach of how most teams perform innovation and discovery activities. Both product and product distribution are essential for business success.
Not just adding a good amount of profiling questions, but also making sure that those questions are of high quality and are more likely to be relevant to identifying different characteristics for each segment later, are both important. Asking only typical demographic questions like age and gender will be worthless. If, for example, you are working on a travel app, you might want to put as many different attributes related to travelers or drivers such as experience, background, preferred type of transportation, budget, frequency of travel, and distance.
While it is easy to add free-form text questions into a survey where participants could type whatever they want, creating good qualitative profiling questions in a pre-defined select/dropdown manner might be way easier for analysis. Otherwise, advanced natural language techniques will have to be performed, and there won’t be an easy way to find common characteristics. Nonetheless, sometimes it is better to have a free-form textarea rather than a bad set of pre-defined options to choose from.
It is worth investing more time into profiling questions as well, and not just breaking JTBD and journey down into many p-metrics. It is fine to start with what you have on the initial ~20 participants, then analyze the results, and see if you missed something important. Then adjust a survey and finally launch full-scale research after that.
Not putting a higher weight on the Importance
While many people rush with JTBD analysis and forming opportunities, calculating scores in different ways, it doesn’t cost more to get it right up front.
Opportunity Score is a weighted difference between Importance and Satisfaction with x2 higher weight on Importance.
If something is not extremely important to many people, nothing else matters.
Thinking JTBD is not for discovery and only for later stage and existing products, or that you need many responses
This one is, perhaps, the most common misconception I see. People often argue with me back when I say that there is nothing better than JTBD analysis for just ideas, early-stage startups, and discovery from 0 to 1 when you have nothing. They tell me that JTBD can only be used when you have an existing product out there to help you improve it…
This comes from an absolute misunderstanding of what JTBD even is at its core.
JTBD is not a task analysis. It is solution independent. If you have an existing product or just an idea in your mind, it doesn’t matter.
Jobs-to-be-done are about understanding why people do what they do, what they are really trying to achieve, and if they can’t, why exactly? This is the core of customer research, customer segmentation, customer-first, customer obsession, customer discovery, need-first, need discovery, unmet need research, or whatever you want to call it.
Everything starts with understanding JTBD. Then it helps you to have an already existing Product-Market-Fit and launch a product that WILL be successful. This is how science works, right? For example, if you pick an object in your hand above the ground and then release it. I know, you know, everyone else knows exactly that in less than a second this object will kiss the floor. This is how JTBD analysis helps to predict the future. It helps you to improve existing products and grow. It helps you to figure out what to do next and where to focus exactly. It helps you to launch new companies in new verticals. And like that object above the floor, I know what will happen if you will help people address their deepest desires and struggles. You will succeed.
One point is true though. Doing JTBD analysis at the 0-1 stage of a business requires a bit different mindset and extra work, especially on lead-generation and pre-sales work.
If you don’t have real people to interview and survey on a regular basis. JTBD analysis is worthless. It is never about imaginary personas and sitting inside your comfy room or office. It is not about buying all your responses. Every part of the JTBD work shall be connected to a real person. Specifically to a real person who is more likely to buy from you.
Understanding the JTBD of just one customer, but a customer nonetheless, is more important than statistically valid results from 10,000 strangers who won’t buy from you.
Remember that JTBD analysis is always commercial.
And when you are about to launch a new startup, you already know who will buy and can instantly send them a message.
Instead of “fake it till you make it”, do “understand JTBD, pre-sell till you make it with and for your potential customers”.
After all, the only work I do for the last years is working with early-stage startups only, and every time I start working with a new founder, we do JTBD analysis out of nowhere, and then they successfully launch, get users, customers and funding.
The future is clearly predictable.
Thinking JTBD is only for product or marketing
JTBD can be used internally, for non-profits, governments, and everywhere where people are.
In the same way, Newton’s laws apply to any object you pick in your hand and then release.
Not aiming for causal inference and real world experiments
And finally and most importantly, JTBD Analysis is not a Task Analysis. The aim of the Jobs-to-be-Done Analysis is to determine the causal mechanism of customer (or citizen) behavior and validating it in a real world via experiments, e.g. for businesses an A/B test or a landing page fake door test could be implemented to see an actual higher conversion rate or intent to buy.
Summary
The ultimate goal of the JTBD Analysis is causal inference, it is to determine what causes customer or social behavior. A Jobs-to-be-Done Analysis is an interdisciplinary methodology rooted in behavioral economics that is used for observational causal inference to understand what drives customer decisions, such as purchasing behavior.
JTBD analysis is a complex end-to-end interdisciplinary methodology that must be:
Behavioral,
Mixed & End-to-end where many qual and quant methods and profiling are used,
Exploratory,
Scientific with the goal to determine the cause of customer or social behavior,
And commercial, or impactful in other way via experimentation in the real world.
JTBD analysis is not a Task Analysis nor an observation of any existing process alone. Only the need, the outcome shall be the unit of analysis. Not a feature, solution, product or a service. It is about understanding why people do what they do, what makes them buy, what is their situation and environment they are in when they consider starting the journey, what triggers that, and what are their end goals and desires, what progress they are trying to make and in what circumstances. What prevents them from getting where they want to be?
People shall never ever be segmented using demographics or circumstances as a whole analysis, i.e. a traditional market or customer research.
Circumstances is what many researches and marketers don’t pay attention to.
This research can be applied in various real-life business and social, practical use cases such as:
Align the entire organization on what the need even is.
Form definitive market segments.
Launch new startups or organizations, successfully, and have proof of future payments.
Identify winning opportunities and strategies for new offerings and grow existing businesses, products, and services.
Speak to each audience in the language that resonates with them and their intent the most.
Get support for your public policy or win elections.
Skills and attributes required to perform a JTBD analysis can be put into a Venn diagram, an intersection between:
People
Logic
And Business.
There are 3 different levels or types of JTBD analysis:
Basic
Advanced
Scientific
JTBD analysis ends with the deliverables:
(Always) Calculating an Opportunity Score and plotting Opportunity Map Scatterplot
(For Advanced+) Forming specific markets and segments, creating and presenting a JTBD Report
For scientific experiments - JTBD analysis becomes an initial part that forms strong hypothesis and sampling data that will have a high probability of success during the experiment.
Amount of participants in the research needed is lower than you might think
20-30 (Low fidelity) - for quickly testing new business ideas or testing if a big team is ready for a full-scale survey. Use averages to summarize your small findings, and then start using percentages after.
70-90 (Mid fidelity) - if you can, get here and this is where you can start doing cluster and/or factor analysis.
250-350 (High fidelity) - there is no need to survey beyond that
Opportunity Score can be:
Simple or Average (on a scale of 1..9)
Percentage (on a scale of 10, 20, or 100%)
Then Opportunity score formula = Importance + MAX ( Importance - Satisfaction ; 0 )
You can think of a flag area analogy. Look for a bigger flag.
Optionally, you can filter segments by tiers where Tier 1 is only those who said Imp=5 & Sat=1 and are very likely to represent your loyal champions and act as proof of future payments.
People who responded with Imp=4 or 5 AND Sat=1,2 or 3 at the same time are top contacts or top leads. Segments are formed from those people.
Results can be visualized in an Opportunity Map (or Landscape)
A scatterplot chart with X - Importance and Y - satisfaction. Focus on the bottom right corner, don’t mess up basic expectations, and ignore the left side.
Depending on the type of JTBD analysis, for the basic one, the chart also will represent average scores.
Even an advanced JTBD analysis has flaws, however, it is a sweet middle spot where most of the JTBD research can end.
Finally, in the JTBD report (which looks like a table) everything comes together and clear market segments are sized and defined as a “group of people” (with similar characteristics) + “job they try to get done” (with the same personal metrics this segment needs the most).
As of today, the JTBD Analysis is the most powerful technique that drives discovery and innovation globally better than anything else. Innovation is not random. It is a clear predictable, measurable, scientific business process. The future is clearly predictable. The choice is yours.
In the next and last story of the series, we will put anything to a conclusion with the bottom-up JTBD market sizing and basic business modeling that will finalize our JTBD report and will help us to see a clear action forward.
Thank you for reading this story.
This was the third story in a series of introductions to JTBD.
JTBD Analysis, What It Is and How to Do It
