What working with data involves
Data work has several parts, and not every project needs all of them:
- collection: automatically pulling data from the systems where it lives;
- cleaning: fixing errors, duplicates, empty fields, and inconsistent formats;
- integration: linking records across systems: ad source, lead in the CRM, payment in accounting;
- reporting: dashboards that update automatically and show consistent numbers;
- analysis: finding patterns and answering specific business questions.
Sometimes the task is narrow: build one report on sales by channel. Sometimes you need a full reporting system for leadership, sales, and marketing.
Who needs this
This service makes sense when you already have enough data, but decisions are still made on gut feeling or from spreadsheets someone assembles by hand. Typical situations:
- marketing and sales see different numbers and argue over whose are right;
- the leadership report takes several hours to compile from five different systems every time;
- an online store wants to understand which customers come back and which buy only once;
- the company is planning forecasting or ML models and needs to prepare historical data first.
If the business is young and there isn't much data yet, a complex system would be overkill. A few well-built reports will do, and you can come back to a data warehouse later.
Spreadsheets, BI tools, or your own data warehouse
Excel or Google Sheets is the simplest option. Pro: everyone knows how to use them. Con: manual work, copy-paste errors, and slow performance on large volumes.
Dashboards in Looker Studio or another BI tool connected directly to your sources. Pro: reports update automatically, and setup doesn't require developers. Con: when there are many sources that need to be linked together, reports become slow and confusing.
Your own data warehouse in BigQuery, PostgreSQL, or ClickHouse with automated loading. Pro: one database for all reports, speed, and the ability to run any calculation. Con: it takes development up front and maintenance whenever source systems change.
Why data preparation takes the most time
The hardest part of analysis isn't the charts; it's everything that happens before them. Typical problems we run into:
- the same customer is recorded in the CRM three times with different phone numbers or name spellings;
- products have different names and SKUs in the accounting system and on the website;
- some orders were canceled or returned, but that shows up in one system and not in another;
- dates are stored in different formats and time zones, and amounts in different currencies;
- business rules changed, and data for older periods was calculated differently.
We don't hide these problems; we document them and agree with you on how to handle them. Otherwise, a great-looking dashboard will show the wrong numbers.
What your data can tell you
- which ad channels bring in customers who buy again, not just once;
- how sales change by season, on weekdays vs. weekends, and by region;
- which customer groups generate the most revenue and how they differ from the rest;
- which products are often bought together and which sell only at a discount;
- where leads most often drop off in the sales process: on the website, with a sales rep, or at checkout.
What to prepare to get started
- a list of the systems your company uses and the people responsible for each;
- read access or data exports for the available period;
- examples of reports you use now, even if they're spreadsheets compiled by hand;
- the questions the reports should answer first.
How to evaluate the result
A good result shows in simple ways: everyone in the meeting looks at the same numbers, nobody compiles reports by hand anymore, and a question like "why did sales drop" is answered in a few clicks instead of a day of back-and-forth emails.
For analysis projects, the result is a conclusion you can test with action. For example, if the analysis shows that a certain customer group buys again more often, the next step is a dedicated offer for that group and a comparison of the results.
Common mistakes in data projects
Starting with a pretty dashboard. A panel with dozens of charts impresses in a presentation, but nobody uses it because it doesn't answer the team's everyday questions.
Not agreeing on definitions. If marketing counts any inquiry as a lead while sales counts only confirmed orders, the report will always be a source of disputes.
Collecting everything "just in case." Loading every field from every system complicates the warehouse and slows down reports. It's better to start with what you need and add data as required.
Forgetting about maintenance. Source systems get updated, and fields and rules change. Without oversight, automated loading quietly breaks and reports show outdated numbers.
What affects the cost
- Number and type of sources: services with a ready-made API are easier to connect than legacy systems without one.
- Data quality: the more duplicates and discrepancies, the more cleanup work.
- Volume and update frequency: daily reports on millions of records need a different architecture than a weekly summary.
- Number of reports and users with different access permissions.
- Analysis: a one-off question or ongoing analysis with recommendations.
Frequently asked questions
Can you connect our accounting system? In most cases, yes: via API, a direct database connection, or scheduled file exports. We choose the method after getting familiar with the system. If, alongside reporting, you also need to set up data exchange between the systems themselves, that's a job for integrations and process automation.
Who has access to our data? Only the people you grant it to. We create the warehouse in your account and assign access by role.
Do we need a dedicated in-house analyst? Not necessarily. We build reports so that managers can use them. But if you have a lot of questions, having someone in-house who owns the data helps a lot.
How is this different from web analytics? Web analytics focuses on website visitor behavior. Here we work with all of your company's data: sales, customers, inventory, finance.
Why Promodex
Promodex has been in business since 2015 and has delivered 750+ projects, including websites, web services, and internal systems that exchange data with CRM and accounting software. Our team includes developers who can set up data collection and marketers who understand what you actually need to see in reports. We explain where every number comes from and are upfront about where there isn't enough data to support a conclusion.
