DataFocus · Journal
Importing CSV, Excel and JSON Files into OT Flash
A step-by-step walk from a spreadsheet on your desktop to a searchable OT Flash table bound to an ontology, using only the Tables, Ontology and Data Search screens.
To import a CSV or Excel file into OT Flash, open Tables, click Import Table, choose "Import table from local", pick the file format (CSV, EXCEL or JSON), upload the file, set the parsing options, confirm the column names and types, and click Submit. The new table appears first in the table list. To make it part of an ontology, open that ontology's Data Tables module and use Import data table; after that you can type a keyword search such as "Sales by Region" in Data Search. The Tables menu must be visible to your role, which the administrator assigns.
The steps below follow the OT Flash r57 user manual, whose own example imports a file called Manual_Sample_Store_Monthly_Sales.csv. The sample in this post is a small hypothetical file in the same shape, not data from a real company.
Which of the four import paths should you use?
Click Import Table and OT Flash offers four methods. The right one depends on where the data lives today.
| Method | Use it when | Typical owner |
|---|---|---|
| Import table from local | The data is a CSV, Excel or JSON file on your computer | Analyst |
| Import table from external data source | The data sits in a connected database | Data maintainer or administrator |
| Import from API | The data is available through an API | Data maintainer |
| HTTP source | Data arrives over HTTP | Data maintainer |
A local file is the only path that needs nothing set up beforehand. The other three rely on a connection someone has already configured; for external databases the manual recommends a dedicated read-only account with access only to the tables to be analyzed. This post covers the local file path and returns to the others at the end.
How do I prepare the file so columns import cleanly?
Most import problems are decided before you click anything. A short checklist:
- Put column names in the first row. The parsing screen asks "Are the column names already defined in the file header?" You tick it when the first row holds names. A header row also gives you readable column names in search later.
- Keep one table per sheet or file. A file imports as one table. If your workbook has several tabs, export or save each one separately.
- Keep each column to one kind of value. A column that mixes numbers and text such as "n/a" will be hard to type as a number. Clean it in the file or decide how to handle it in the type step.
- Remove summary rows and blank lines. Totals at the bottom would be summed again by search. The parser can ignore empty rows and rows with errors, but a total row is valid data and will not be skipped.
- Use column names people would type. Search matches on column names, so "Sales" beats "Net_Sales_Amt_USD_v2". You can rename later, but fewer renames means fewer surprises.
- Stay under the size limit. The manual states that the file must be smaller than 50 MB.
- Avoid names that collide with search keywords. If a column name equals a search keyword, the system renames it on import by appending a sequence number such as
_1. You can set a display name in the table detail afterwards.
Here is an illustrative file (hypothetical data, not run through the product) with the same five columns the manual's sample uses:
Store,City,Month,Orders,Sales
Harbor,Rotterdam,2026-01,420,18350.50
Harbor,Rotterdam,2026-02,398,17020.00
Ridge,Denver,2026-01,512,24110.75
Ridge,Denver,2026-02,487,22980.25
How do I import the file and check the table?
- In the left menu, open Tables, then click Import Table.
- Choose Import table from local. The Import locale table dialog opens.
- Pick the file format: CSV, EXCEL or JSON. Click Please Select File, choose your file, and click Upload.
- Set the parsing options. Choose the delimiter (usually COMMA for CSV), confirm the header-row checkbox, and decide whether to ignore empty rows, ignore error data, and discard any data found after the last column. Show/Hide Advanced Parsing Options reveals more. Click Next.
- On the column screen, check the table name, add a description if you want one, select the columns to import, and adjust column names and data types such as string, int or double. Sample values appear on the right, and Preview shows how the imported data will look.
- Click Submit.
The left menu you see depends on the roles and permissions your administrator assigned, so if Tables is missing, that is the first thing to ask about.
For JSON the flow is the same; you pick JSON as the file format in step 3. The manual does not describe nested-JSON flattening rules, so a flat array of records with consistent keys is the safe shape to start with.
What should I check after the import?
Open the table card's Detail button. On the Columns tab, check that each column has the type you expect. Two settings matter for search:
- Column Type: ATTRIBUTE columns are used for grouping and filtering, MEASURE columns are aggregated.
- Aggregation: the default aggregation of a measure, such as SUM.
In the sample, Store, City and Month should be attributes, while Orders and Sales should be measures. Use Preview data to compare a few rows with the source file. If the business calls a column something else, set its Display Name here; this is the term search will match.
How do I bind the table to an ontology?
An ontology in OT Flash is a knowledge space organized around one business topic, bringing together related tables, business documents, indicators and pinboards. Each ontology has 11 modules, including Data Tables, Data Explore and Document Center.
If you have not created one yet, creating an ontology takes a Name and an icon. Afterwards you import the related tables in Data Tables and upload business documents in Document Center.
For your new table:
- Open Ontology from the left menu and enter the ontology for your topic, for example one for retail store performance.
- Open the Data Tables module. It lists the tables already bound to the ontology.
- Click Import data table and choose your newly imported table from the system.
- If the ontology holds more than one table, use the Relationships tab to set up joins between them, in the same way as in the Tables chapter.
Importing the table and binding it to an ontology are two separate steps: the first puts the table in the system, the second adds it to the ontology.
How do I run a first search?
Open Data Search under BI Workspace, choose your table, and type column names and keywords. Typing "Sales by Region" in a table with a Sales measure and a Region attribute computes the result and picks a chart; measure columns are summed by default. With the sample file above, "Sales by City" would give one total per city.
If you type a full question instead of keywords, the system asks FocusGPT to parse it into a search statement, shows the parsed statement, and gives you a Revert option. The right rail shows the SQL behind the result and FocusGPT's analysis summary, which is a quick way to verify that Sales is being summed over the rows you expect. Comparing one city's total with a manual sum in the original spreadsheet is a good first check.
When should a one-off upload become a scheduled import?
A local upload is a snapshot. If you replace the file every Monday, you are doing the job a schedule should do. The manual's Table Task tab manages scheduled import tasks for tables imported from data sources, showing each task's Data Source, Import Type, scheduled import time, status and last import time. When you import tables from an external data source, the third step of the wizard lets you set the import type and an optional schedule.
A practical decision rule:
- Keep the local file when the data is a one-time extract, a survey, or a file that changes rarely.
- Move to an external data source when the same file is exported repeatedly from a database. Ask the data maintainer to connect the source and schedule the import.
- Consider API or HTTP import when the data is produced by another system rather than a database.
Where is this not the right fit?
A file upload will not keep two systems in sync, and a spreadsheet with merged cells, multiple header rows or several tables on one sheet needs cleaning before it can become a single table. Files of 50 MB or more are outside what the local import accepts, so those belong in a database and the external-source path.
Next step
OT Flash is distributed free. Download it from the editions page or the GitHub repository, where the installers, release notes, system requirements and supported platforms are published, and follow the r57 user manual for the Tables and Ontology chapters.