What is the Underlying Causes Dimension?

Underlying Causes is the Dimension that asks every piece of customer feedback "Why does this occur?" It explains the deeper reasons or conditions that lead to the situation described in the feedback: a recent update behind a crash, too few staff behind a long queue, a new supplier behind a drop in quality. Kimola groups the same cause across all of your reviews, survey responses and support conversations and counts the records behind each one, so you see why things happen to your customers and how many of them point to the same cause. Underlying Causes belongs to the Drivers group and is available in every package.

A Dimension is a research question Kimola asks of every piece of feedback. Themes tell you what customers talk about, and sentiment tells you how they feel about it. Underlying Causes goes one level deeper and collects the reasons behind what customers describe, which makes it the Dimension for root cause analysis of customer complaints. Every answer stays tied to the feedback that gives it, so you can open the records behind a cause and read them.

The Question It Asks

Underlying Causes asks one question of each record: "Why does this occur?" Feedback that explains why something happened gets an answer. Some feedback names the cause directly: "Since the last update…", "because there is only one person at the counter…", "after they changed the recipe…". Other feedback makes it clear from what it describes: a review saying that orders placed after 6 pm always arrive two days later points to a cut-off time, even if the word never appears. A cause can be a change (a new version, a new policy, a new supplier), a condition (staffing, stock, opening hours, capacity) or a way something is designed or run (a rule, a process, a default setting).

Feedback that describes a situation without any reason behind it gets no answer for Underlying Causes. That is expected: many complaints say what went wrong and stop there. "The queue was long today." gives a pain point and no cause. The share of records that answer Underlying Causes tells you how often your customers explain why something happens. One piece of feedback can give several causes at once, and each one is counted.

Underlying Causes explains a situation, whatever its tone. Most of its answers sit behind problems, but it also finds the reasons behind good experiences: "My order came a day early since they opened a warehouse in our city." gives the new warehouse as the cause of fast delivery. Underlying Causes is different from Pain Points, which names what goes wrong. In "The app crashes since the last update", the crash is the pain point and the last update is the underlying cause. It is also different from Motivations, the other "why" question of the Drivers group: Motivations asks "Why is this chosen or preferred?" and explains why customers chose the product, while Underlying Causes explains why the situations they describe happen.

What You Get

Kimola builds the answers of Underlying Causes from your data. It reads the dataset, finds the causes customers give and names each answer in their terms, so the causes in feedback about a coffee chain and those in feedback about a project management tool come out different. Across thousands of records, Kimola groups the same cause under one answer and counts the records that give it. The counts and shares come from the classification itself; an AI model writes the summaries afterwards, from numbers that are already computed.

The answers are the causes as customers see and describe them. Many customers name a cause your own data can confirm, such as a release date or a staffing change, and the record count shows how widely each cause is felt. The quotations under each answer give your team the exact words to start an investigation from.

In a Report

On the Report's Dimensions page, Underlying Causes has its own tab. A ring shows the share of the Report's records that answer the question. The summary card gives a headline and a few paragraphs on why the situations in your feedback happen, with key terms in bold. Below it, each cause is a cluster with its number of records, its share, a short summary. A cause with very few records gets no summary, and the page says "Summaries are not available due to small cluster sizes."

Cluster Scale shows each cause's share of the records that answer Underlying Causes, and Dataset Scale shows its share of all the records in the Report. Because one record can give several causes, the shares on Cluster Scale can add up to more than 100%. The Columns and Dimensions buttons on a cluster card show what lies behind a cause, for example which Pain Points it leads to or in which stores or countries it comes up most. See Dimensions in Reports.

In a Space

Spaces are available in the Enterprise package. On a Space's Dimensions page, Underlying Causes shows its answers under What they talk about and how they change Over time. A cause whose share changed significantly with the last upload is marked ▲ or ▼ "since the last upload", so a new cause, such as a release or a policy change, shows up with the next file. Kimola writes Insights after every new file: What the answers say for Underlying Causes as a whole, and Why they answer so for each cause that at least 10 of the latest records give. In a Space, the answers come from the labels of the Funnel's Underlying Causes model. See Dimensions in Spaces.

In both Reports and Spaces, you can filter the Content and Performance pages by a cause, or cross Underlying Causes with a column or a classifier in Tables, for example Underlying Causes by store or by sentiment.

An Example

Take this invented review of a coffee chain:

Underlying Causes finds three answers in it. "Mobile orders served first" is a rule of how the branch is run, and it explains why the queue at the counter grows. "Too few staff in the morning" is a condition. "App sends orders to the wrong branch" is a fault of the new app that explains the wrong order. The launch of the new app ties them together: it is the change after which the customer started to notice the problems. The long queue and the wrong order are not causes; they are answers for Pain Points. The morning rush is picked up by Experience Stages, and the branch by Experience Context, when those Dimensions are added.

A second review, "The queue was very long again this morning.", describes a problem and gives no reason for it, so it gets a Pain Points answer and no Underlying Causes answer.

Across a dataset of 5,000 reviews, the same causes gather into clusters. "Too few staff at peak hours" might be the largest, with 540 records, followed by "Mobile orders served first" and "App sends orders to the wrong branch". Clicking Dimensions on the largest cluster would show that it sits behind several pain points at once: long queues, cold drinks and wrong orders. Fixing that one cause would then address all three.

When to Use It

Use Underlying Causes whenever knowing what goes wrong is not enough to act and you need to know why it happens. It reads best in feedback where customers tell the story of what happened to them: support tickets and chat conversations, product and app store reviews, and open-ended survey answers to questions such as "What could we have done better?". For a full walkthrough with Pain Points, see How to Find Customer Pain Points.

Root cause analysis. Add Underlying Causes next to Pain Points and open the Dimensions button on a cause's cluster card to see which problems it leads to. A cause behind several pain points is usually the one to fix first, because one change solves many complaints.

Handing work to the right team. A cause often tells you who owns a problem: a release points to the product team, staffing and opening hours to operations, a return rule to customer service. Filter the Content page by a cause to send that team the feedback that names it.

After a change. In a Space, check the ▲ and ▼ marks on Underlying Causes after each upload to see whether a release, a new supplier or a new policy has started to cause problems, and whether a fix has made a cause fall.

Segments. Cross Underlying Causes with a column in Tables to see whether customers in different stores, regions or plans point to different causes. A cause that comes up in one region only is often a local condition.

What works. Read the causes behind positive feedback too. They show which decisions customers notice and value, so you keep them when other things change.

Underlying Causes works best together with the other Dimensions of the Drivers group. Motivations ("Why is this chosen or preferred?") shows why customers choose you, next to why things happen to them, and Unmet Needs ("What is missing or desired?") shows what customers want added or changed. Outside the group, Pain Points gives the problems each cause leads to, and Experience Stages shows at which moment a cause hits.

Use It in Kimola

How you add Underlying Causes depends on whether you work in a Report or a Space.

In a Report

Select Underlying Causes in the Dimensions step while you create a Report. It is listed under All and under Drivers. Click it to add it to My List, then check the Required Query in the Review step. To add it to a Report that is already analyzed, open the Report's Dimensions page, click the Underlying Causes tab without a check mark and click Generate Now. Kimola classifies the Report's existing records with Underlying Causes and writes its analysis. The steps are in Add Dimensions to a Report.

In a Space

A Space gets Underlying Causes from a Dimension model in its Funnel. In Models, click Create AI Model, choose By Prompt, click Dimension in the Purpose step and choose Underlying Causes under Which dimension?. Its Question and Description are filled in, and you can fine-tune them for your business. In Labels, define the causes your Spaces can find, such as "Too few staff at peak hours" or "Changes in the latest app version", each with a short description. Then pick the model on the DIMENSIONS tab of the Funnel wizard's Models step. Every file uploaded to the Funnel's Spaces after that is classified with Underlying Causes. The steps are in Add Dimensions to a Funnel.