Data Analytics

Data analytics that turns your business data into forecasts and decisions

We work with the data you already have, such as sales history, customer purchase records and operating data, to forecast, analyse and simulate, then turn the results into recommendations your team can use.

Talk to us about your data See the questions we answer

Illustration: a monthly sales forecast

Actual sales
Forecast
80% likely range

December: Next peak month
12 items: Products to restock
8%: Customers likely to lapse

Illustrative data

Business data analytics uses the data an organization already has to answer four kinds of question: what happened, why it happened, what is likely to happen next, and what to do about it. The work that helps decisions most usually sits in the last two, because it lets a team act before the outcome arrives.

Read the full summary

NXT Consulting Group provides data analytics for private companies and public organizations in Thailand. The work covers sales and demand forecasting, churn analysis, purchase-data analysis for customer segmentation and lifetime value (RFM and CLV), market basket analysis, propensity models for purchase or campaign response, promotion and pricing analysis, inventory planning, and scenario simulation to show the range of likely outcomes. Every engagement starts with a check of the data already in place, and the method follows the business question. Results are delivered as recommendations, dashboards or models the client's team can keep using, built in the tools the organization already has. Before NXT, our team analysed data for e-commerce and retail platforms and for government agencies.

4 levels: From what happened to what to do next
Your data: We start from the data you have, with no new system to buy
4–8 weeks: Typical length of a focused analytics project
NDA + PDPA: Confidentiality agreement and personal data protection practice

Levels of analytics

The questions data can answer

Most organizations already have first-level reporting. Data helps decisions more once it starts to say what is likely to happen next and what to do about it.

Descriptive
What happened?
Sales reports, KPI dashboards
Diagnostic
Why did it happen?
Which customers, branches or products changed
Predictive
What is likely next?
Sales forecasts, customers likely to lapse
Prescriptive
What should we do?
Stock levels, who to target, simulated options

What we analyse

Data questions we help answer

We choose the method from the business question, and say plainly when a simpler analysis will answer it.

What will sales be next month?

Methods

Time series
Seasonality
Promotion effects

You getForecasts by product or branch with a likely range, for stock planning and targets

Which customers are likely to stop buying?

Methods

Churn model
Survival analysis

You getA risk score for each customer, the main reasons, and who to look after first

Which customers create the most value?

Methods

RFM
Customer lifetime value
Cohort

You getCustomer groups built from real purchase data, with the lifetime value of each group

Which products are bought together?

Methods

Market basket
Association rules

You getProduct bundles and cross-sell suggestions based on purchase data

Who is likely to buy or respond to a campaign?

Methods

Propensity model
Classification

You getA target list ranked by likelihood, so marketing budget goes where response is higher

Which promotions actually add sales?

Methods

Promotion analysis
Price elasticity
A/B test

You getThe extra sales from each promotion and price level, compared with no promotion

If the plan changes, what range of results should we expect?

Methods

Monte Carlo
Scenario model
What-if

You getThe likely range of results for each option, such as price, capacity or budget

How much stock should we hold of each item?

Methods

Demand planning
Safety stock

You getStock levels and reorder points by item, balancing stock-outs against overstock

How we work

How a data analytics project runs

01 · Audit
Check the data
Collect data from existing systems such as POS, ERP, CRM or Excel files, check completeness and accuracy, and say at the start how far the data can answer the question.
02 · Model
Analyse and model
Choose the method for the question, test accuracy against past data, and explain which factors drive the result.
03 · Decide
Turn it into decisions
Summarize the recommendations and numbers leaders can use straight away, such as sales targets, stock levels or the customers to look after first.
04 · Handover
Hand it to the team
Deliver the dashboard, model and data-update routine, and train the client's team to use and adjust them.

Data we use

Data most organizations already have

The data does not need to be perfect at the start. We begin with a data quality check and say plainly how far the data can answer the question.

Sales history (POS, ERP)
Line-item purchase records
Membership and CRM data
Stock and purchasing data
Campaign and promotion data
Website and app data
Customer surveys
External data such as holidays and seasons

We work under a confidentiality agreement (NDA), use only the data the question needs, and recommend removing personally identifiable information before data is shared, in line with Thailand's Personal Data Protection Act (PDPA).

What you receive

What we deliver

The format depends on the question. The previews below are mock-ups of the formats, not client work.

Mock-up

Tracking dashboard

Indicators tied to decisions, refreshed from the team's own data

Mock-up

Forecast model

Forecasts with a likely range and an explanation of the method

Mock-up

Customer scores

A risk or propensity score for each customer, with the main reasons

Mock-up

Scenario simulator

Change the assumptions and see the range of results for each option

Mock-up

Recommendations

What the data says, the options and next steps, in a few pages

Our team's experience

Data work our team has done

Examples of data analysis our people did before joining NXT.

E-commerce and retail

Big-data analysis for e-commerce and retail platforms, and development of a geo-data platform business

Retail

Customer data analysis to find the root cause of rising marketing costs for CP ALL

Government

Research and survey projects for government agencies with more than 6,000 respondents

Economic policy

Economic impact analysis to support policy decisions

These are our team's prior-role experience at other firms and organizations, not work performed under NXT. Details are on the team profiles.

See NXT's own work

FAQ

Questions before starting a data analytics project

How much history do we need to forecast sales? For seasonal sales, at least two years of monthly or weekly data, so the seasonal pattern repeats at least twice. Shorter histories can still be forecast, but the uncertainty range is wider, and we say so at the start.

Our data sits in several systems and is not clean. Can we still start? Yes. Most projects start from Excel files and POS, ERP or CRM systems that are not yet connected. The first step is to combine and check the data, which also shows where data collection should improve.

Do we need to buy new software or systems? No. We work in the tools the organization already uses, such as Excel, Google Sheets, Power BI or Looker Studio, with statistical tools for the modelling. If you want a permanent system, we help define the requirements without tying you to any vendor.

How is data analytics different from building a dashboard? A dashboard shows what happened. Analytics goes on to why, what is likely next and what to do. We often deliver a dashboard together with the recommendations, so the team can keep tracking results after the project.

Will our customer data be safe? We work under a confidentiality agreement, ask only for the data the question needs, and recommend removing names, phone numbers and other identifying details before sharing, in line with Thailand's Personal Data Protection Act.

How long does it take, and what does it cost? A focused analysis, such as a sales forecast or customer analysis, usually runs as a 4–8 week Advisory Sprint, depending on how ready the data is. Scope and price are agreed before work begins.

Insights