In-house marketing analysis: How to distinguish between analyses that are easy and difficult to do in-house
Whether or not bringing marketing analytics in-house is successful largely depends on which types of analysis are handled internally. While it's relatively easy to handle analyses that aggregate and classify sales and customer data internally, analyses that isolate the effectiveness of strategies from other factors, analyses that determine budget allocation, and analyses that use external data become more difficult to complete solely within the company.
What is in-house marketing analytics?
In-house marketing analytics means establishing an internal infrastructure, including the necessary personnel, data, and analytical environment, so that in-house staff and teams can perform marketing data analysis without relying on external experts. In practice, this may involve handling only a portion of the analysis or specific analytical tasks internally.
Marketing analysis can be divided into four types depending on what it answers: descriptive analysis (what happened), diagnostic analysis (why it happened), predictive analysis (what will happen if things continue as they are), and prescriptive analysis (what should be done). The differences between them are...What is data-driven decision-making? Steps to turn "decisive analysis" into data-driven decision-making in marketing.As explained in [the relevant section], the ease of bringing an analysis in-house varies greatly depending on the specific analysis being performed.
Bringing analysis in-house has both advantages and disadvantages, and which is more significant depends on the type of analysis being performed.
Benefits of bringing it in-house
The benefits of bringing marketing analytics in-house are primarily the following three. All of these benefits can only be obtained when the right conditions are met.
It allows for analysis tailored to the specific circumstances of the business.
Internal teams are familiar with the company's products, sales mechanisms, and the history of past initiatives, allowing them to tailor analyses to the specific business situation. For example, an internal team member can determine from the context whether a month's sales increase is due to a special sale or increased distribution (expansion of stores carrying the product).
→ condition:The key is that the analysts can interact not only with the data but also with sales and brand representatives on a daily basis.
The time required to make a decision will be shortened.
Compared to outsourcing, the process of data transfer and ordering is reduced, allowing for faster decision-making based on the analysis results. In Japan, the eyewear brand JINS has reported that by internalizing the analysis of advertising effectiveness, they have been able to produce reports more quickly and frequently (Source: JINS Tech Blog)Practical implementation and problem-solving of in-house MMM development)).
→ condition:This means that the data is regularly updated and ready for analysis.
External payments can be reduced.
This can reduce the costs paid to external analysis and consulting firms.
→ condition:The advantage is that you already have personnel capable of performing the analysis within the company. If you were to hire new personnel, you would need to compare the total cost, including the personnel and time required for recruitment, training, and data preparation.
Thus, each of the three benefits has its own conditions, and the ease with which these conditions are met varies depending on the type of analysis.
Challenges of bringing things in-house
The following five challenges are common obstacles that companies often face when bringing marketing analytics in-house:
It is difficult to secure personnel capable of performing analysis.
According to the Information-technology Promotion Agency (IPA), 85.1% of Japanese companies lack the personnel to promote DX (Digital Transformation), a significantly higher level than in the United States and Germany (Source: IPA).DX Trends 2025 – Developing Digital Talent in the AI Era). In "DX Trends 2024," data scientists are listed as the second most in-demand type of personnel after business architects (Source: IPA "DX Trends 2024 – The worsening shortage of personnel and challenges in promoting DX)).
→ countermeasure:Instead of trying to handle all analyses in-house at once, we first narrow down the analyses that will be handled internally, and then secure and train the necessary personnel for those analyses.
Preparing the data is time-consuming.
The data needed for marketing analysis is often scattered across departments and companies; sales data is stored in internal systems, advertising data is from agencies, and customer data is stored in membership or e-commerce systems. Even if you have the right people to perform the analysis, the task of collecting and formatting the data can become overwhelming, causing the analysis to stall.
→ countermeasure:Before starting the analysis, decide what data will be prepared, who will prepare it, and how often.
It is difficult to maintain the neutrality and reliability of the results.
Because internal teams are very familiar with the business, their preconceptions can sometimes influence how they interpret the results (types of bias and points to note are "What is bias? A clear explanation of its meaning, types, and points to note in data analysis.(This is explained in "[...]"). Also, when a team within the marketing department evaluates the effectiveness of its own budget, management planning and finance may ask whether the results are neutral and objective.
→ countermeasure:The analysis procedure and data will be disclosed, and the relevant departments will decide in advance which metrics will be used and what the conclusions will be. For analyses that have a significant impact on the budget, it is also possible to incorporate third-party verification. An example in advertising effectiveness analysis is:Comparing in-house vs. outsourcing of MMM (Multi-Marketing Memory): Which tasks should be handled in-house in the open-source era?This is explained in the "Agreement" section of "[the document]".
It's difficult to find a benchmark to compare it to other companies.
Internal analysis alone can make it difficult to determine whether a company's figures are high or low compared to the industry average. Nielsen cites market benchmarks to verify results as one of the criteria when selecting an external partner for marketing mix modeling (MMM) (Source: Nielsen)The path to success in internal analytics)).
→ countermeasure:We use publicly available industry data and insights from external partners who support multiple companies as references to verify our results.
We need to keep up with the changes in methodology.
Marketing analytics methods and software are constantly evolving, with open-source libraries for MMM (Marketing Memory Management) being released and AI-powered analytics becoming increasingly widespread. Internal teams need to continuously learn new techniques alongside their daily analysis.
→ countermeasure:We will either allocate time for learning as part of our work schedule, or incorporate new methods through external training or collaboration with external partners.
The difficulty of these challenges also varies depending on the type of analysis. You can determine which type of analysis is easiest for your company to handle using the following three criteria.
Three criteria for distinguishing analyses that are easy to perform in-house from those that are difficult to perform in-house
Whether an analysis is easy to perform in-house can be determined by three criteria: how easily the results can be verified, whether the necessary data is available within the company, and how much budget the results will affect and who needs to be convinced.
Is it easy to verify the results?
While the aggregated results can be verified by comparing them with the original data, it's not possible to directly verify whether the estimation of "what increased sales" is correct.
- If it is easy to do in-house:This analysis allows you to verify results such as sales figures, customer visit counts, and customer classifications using the original data.
- If it is difficult to manufacture in-house:This is an analysis that uses statistical models to isolate causes, such as estimating the effectiveness of advertising and promotions. For example, if sales increase in a month when a campaign is run during peak season, the answer to how much of that increase is due to the campaign and how much is due to the season will vary depending on the assumptions of the model. Google's Meridian library of MMMs also explains that it is difficult to directly verify whether the causal estimation of whether sales growth is truly due to a measure is correct, and that indirect indicators are usually relied upon (Source: Google Meridian documentation)Assess the model fit and the results)).
Is the necessary data available within the company?
The initial hurdle to bringing the process in-house depends on whether the necessary data is already available within the company.
- If it is easy to do in-house:This type of analysis can be answered using data from internal systems, such as your company's sales, customer data, and website traffic.
- If it is difficult to manufacture in-house:This type of analysis requires sales data from wholesalers and retailers, external data such as social media and word-of-mouth, or newly conducted surveys. In such analyses, the challenge lies not in determining the analytical method, but in deciding how to obtain the data and how to define the questions and target audience.
How much budget will the results move, and who needs to be convinced?
When the results of an analysis influence significant budgets or decisions in other departments, it becomes difficult to complete the process solely within the company.
If it is easy to do in-house:This is an analysis used within the team to improve their strategies.
If it is difficult to manufacture in-house:This type of analysis involves significant budget changes, such as budget allocation between media outlets or price revisions, and requires acceptance of the results by management planning, finance, and agencies. Because there is a risk of errors being used in major decisions without realizing it, external expertise and third-party verification are helpful.
Applying these three criteria to analyses commonly used in marketing, we get the following:
Ease of in-house analysis by type
When applying three criteria to analyses commonly used in marketing, those closer to descriptive analysis are easier to handle internally, while those closer to root cause analysis or budget allocation decisions require more external expertise.
The following table shows general trends, which can vary depending on the data and internal company structure.
| analysis | Questions to answer (types) | Main data | Parts that are easy to handle within the company | Areas where external expertise is useful |
|---|---|---|---|---|
| KPI aggregation and reporting | What happened (description) | Sales, advertising management screen, web analytics | Data aggregation and dashboard operation | Review of the indicators used for decision-making |
| Customer data analysis | Who is buying, and who is likely to stop buying? (Description to prediction) | Purchase history, membership data | Customer classification, aggregation of purchasing trends | Predicting customer defection and LTV, and estimating the effectiveness of policies. |
| Measuring the effectiveness of advertising and campaigns, and budget allocation. | What worked, and how should it be distributed? (Diagnosis to prescription) | Advertisement data, sales, external factors | Data preparation and planning based on results | Model design, results verification |
| Price and promotional effectiveness analysis | What will happen if we change prices and promotions? (Diagnosis to prediction) | POS data, price and promotional records | Summary of sales promotion results | Estimation of the effects of discounts and special offers, analysis by customer. |
| Analysis of the reasons for selection | Why are you and your competitors chosen? (Diagnosis) | Questionnaire survey | Summary of survey results | Survey design, model for isolating factors |
| Analysis of external data such as social media and customer reviews. | What are consumers interested in? (Descriptive to diagnostic) | Social media, word of mouth, search | Confirmation of mention of our company | Data acquisition, organization by topic |
KPI aggregation and reporting
The aggregation and reporting of KPIs is the most easily handled descriptive analysis within a company.
- Parts that can be easily handled within the company:Sales and advertising data can be aggregated internally using BI (dashboards) or spreadsheets.
- Areas where external expertise is helpful:It's time to re-evaluate which metrics you use to make decisions. Because the performance metrics in the ad management screen are counted according to the standards of each media outlet, the sum of the performance metrics from multiple media outlets may exceed the actual performance.
Customer data analysis
Analyzing customer data using purchase history and membership data is relatively easy to do in-house because the data is located within the company.
- Parts that can be easily handled within the company:This type of analysis involves segmenting customers by purchase amount and frequency, and compiling purchasing trends.
- Areas where external expertise is helpful:This involves models that predict signs of customer churn and LTV (Customer Lifetime Value), as well as analyses that estimate how much a campaign has changed customer behavior. If a campaign is applied to customers who were already likely to buy, the effect of the campaign may appear greater than it actually is. Therefore, it is necessary to design an analysis that correctly compares these customers with customers who did not receive the campaign, ensuring that the conditions are properly controlled.
Measuring the effectiveness of advertising and campaigns, and budget allocation.
Analyzing the effectiveness of advertising and campaigns and determining budget allocation is one of the types of analysis that is difficult to handle solely within a company, as it involves both identifying causes and making budget decisions.
- Parts that can be easily handled within the company:This involves organizing advertising and sales data, and planning strategies based on the results.
- Areas where external expertise is helpful:This involves designing a model that incorporates factors such as competitor activity and seasonality, and verifying the results when they deviate from the reality on the ground. It also involves deciding which tasks to handle internally and which to outsource, using MMM as an example.Comparing in-house vs. outsourcing of MMM (Multi-Marketing Memory): Which tasks should be handled in-house in the open-source era?The explanation is divided into six tasks.
Price and promotional effectiveness analysis
Analyzing pricing and the effectiveness of promotions is difficult to complete entirely within the company, because even if the promotional performance data is available internally, the data needed to estimate the effectiveness may be located outside the company.
- Parts that can be easily handled within the company:This involves compiling and reviewing the details of promotional activities and the resulting sales performance.
- Areas where external expertise is helpful:This analysis involves estimating the quantity that would have been sold even without discounts or special offers (baseline) and separating the increase due to promotions, as well as analyzing the effects on each trading partner, such as wholesalers and retailers. In the case of consumer goods manufacturers, since in-store sales data is held by retailers, it is necessary to consider how to obtain the data.
Analysis of the reasons for selection
Analyzing why your own or your competitors' products are chosen depends on whether existing survey data is sufficient or if new surveys need to be designed, which makes it easier to bring this analysis in-house.
- Parts that can be easily handled within the company:This involves compiling existing survey results and comparing them by attribute.
- Areas where external expertise is helpful:This involves designing questionnaires that include the necessary questions for analysis, and building statistical models that break down how much multiple factors such as price, quality, and brand image influence purchasing decisions. Simple aggregation alone does not reveal which factors actually influence the choice.
Analysis of external data such as social media and customer reviews.
When using external data such as social media, reviews, and search data for analysis, the first challenge is how to obtain that data.
- Parts that can be easily handled within the company:This involves checking for reactions to your company's official accounts and mentions of your products.
- Areas where external expertise is helpful:This analysis involves continuously collecting a large volume of posts and reviews, organizing them by topic, and interpreting changes in consumer interests. It requires both data acquisition and techniques for categorizing text by topic.
Thus, for tasks such as root cause analysis and analysis involving external data, dividing the work between internal and external teams becomes a more practical approach.
How to proceed: Determine the scope of responsibility within the company.
When bringing marketing analysis in-house, it's realistic to start with analyses that are easier to handle internally, and then outsource more complex analyses on a task-by-task basis.
Start with descriptive analysis and analysis of your own company's data.
Aggregation, reporting, and customer data classification are suitable initial targets for in-house implementation because the results are easy to verify and the data is located within the company. Only after establishing a system for regularly collecting data can a foundation be laid for smooth subsequent performance measurement and predictive analysis.
The task of identifying the cause and analyzing budget allocation will be divided among different departments.
The estimation of effects and the analysis of budget allocation are divided into tasks such as question design, data preparation, model building and updating, interpretation, agreement, and implementation, and it is decided which of these tasks will be handled in-house. Nielsen also recommends a hybrid approach, where you can choose to handle the interpretation of results and strategic support in-house, or outsource some or all of it. The way to divide the tasks is as follows:Comparing in-house vs. outsourcing of MMM (Multi-Marketing Memory): Which tasks should be handled in-house in the open-source era?" for a detailed introduction.
Start building your system small.
Rather than setting up an analytical system all at once, it's more manageable to achieve results with one analysis and then gradually expand it while gaining internal understanding. Questions to confirm the organization's current position and approaches to building the system are:How to Build a Data-Driven Marketing Organization: How to Determine Your Current Position and Take the First StepsIt is introduced in "[...]."
How XICA is involved as an external partner
At XICA, our data scientists and consultants work with company representatives to conduct analyses that involve root cause analysis and decision-making, such as estimating the effectiveness of initiatives, budget allocation, pricing and promotional effectiveness, and analyzing the reasons for customer selection. We also sometimes participate as a third party to verify results for companies that are already conducting internal analyses, assisting with result interpretation and internal consensus building.
FAQ
What kind of personnel are needed to bring marketing analytics in-house?
The required personnel depend on the type of analysis to be performed within the company. If the focus is on aggregation and reporting, you can get started with someone who can organize data and visualize it using BI software. If you plan to estimate and predict effects in-house, you will need an analyst who can explain the assumptions of the statistical model and a business person who can interpret the results in a business context.
Which is more cost-effective: bringing the work in-house or outsourcing it?
Which option is cheaper depends on the type of analysis and the existing personnel within the company. If you have analytical personnel in-house, you can reduce external costs by bringing the analysis in-house. However, if you need to hire new personnel, you need to compare the costs including recruitment, training, and the personnel and time required to prepare the data. Some companies choose a hybrid approach, outsourcing only the more complex analyses.
Where should I begin when bringing marketing analytics in-house?
It's practical to start with analyses where the results are easily verifiable and the data is readily available within the company, such as KPI aggregation and reporting, or analysis of your own customer data. In parallel, deciding who will prepare which data and how often will be helpful for subsequent effectiveness measurement and predictive analysis.
Can marketing analysis be done in-house using AI?
AI has the potential to expand the scope of tasks that can be handled in-house by speeding up data aggregation, program creation, and initial interpretation of results. However, for analyses where it is difficult to directly verify the results, such as estimating effects, there will still be a need for people to judge whether the assumptions of the model are valid.
Summary
The feasibility of bringing marketing analytics in-house varies greatly depending on the type of analysis. It's more practical to handle analyses where the results are easily verifiable and the data is available internally, while analyses such as root cause analysis and budget allocation should be outsourced on a task-by-task basis.
If you'd like to learn more about how to conduct marketing analysis, please see the following article.
- What is data-driven marketing? Its meaning, implementation, and how to use it to drive decision-making.
- Comparing in-house vs. outsourcing of MMM (Multi-Marketing Memory): Which tasks should be handled in-house in the open-source era?
- What is data-driven decision-making? Steps to turn "decisive analysis" into data-driven decision-making in marketing.
- How to Build a Data-Driven Marketing Organization: How to Determine Your Current Position and Take the First Steps
- What is marketing data analysis? Explanation of methods, procedures, and tools for different purposes.
Related Articles
-
ColumnWhat is price elasticity? Examples of high and low price elasticity, estimation methods, and practical considerations.
-
ColumnResponding to advertising budget cuts: Three data strategies for discussing "business growth" with management.
-
ColumnFour things you should understand first in order to leverage generative AI in marketing strategy decision-making.


































