ATX AnalyticsTable Training | Video Tutorials and Team Training

Learn ATX AnalyticsTable in minutesShort full-HD video tutorials recorded on the real ATX AnalyticsTable application, every click marked and the whole menu path shown: how to open a project from disk, how to open a table from the local DuckDB database, how to write SQL on a DuckDB table and pin the result to a sheet as static data, how to export data to Excel and schedule the export with a date parameter in the file name, how to filter SQL to yesterday’s rows with a =TODAY()-1 parameter (WHERE order_date = '${yesterday}'), pin the result and export it to a dated Excel file such as reports_2026-10-22.xlsx, how to use SQL functions (SUM_SQL, COUNT_SQL) and the AI_PROMPT function in a grid, how to create a pipeline, how to pin a query, how to build a query visually, how to mask PII data, how to find anomalies and outliers, how to build a pipeline that replaces anomalies (hidden spaces, zero-width characters, HTML entities, leading and trailing whitespace) in a named range and schedule it for every weekday morning, how to build a dashboard, how to build a sales forecast with prediction intervals, a backtest and two scenarios, how to build a histogram of order values with a range of interest, percentile markers and a split by channel, how to connect a data source and how to send a PDF report. Teams can request online, on-site or self-paced training.

Video tutorials

Open a project from disk — Browse to a project folder on this machine and open it — sheets, queries and database together. Menu path: File → Open Project → Browse → D:\ → Tutorial-Demo → Use this folder. Steps: 1. Open the File menu 2. Click Open Project 3. Choose Browse for project — a folder or an .atx file 4. Pick the drive — D:\ 5. Open ATX Agent Hub 6. Open AnalyticsTableTests 7. Open the project folder — Tutorial-Demo 8. The .atx file and its folders: queries, pipelines, schedules 9. Click Use this folder 10. Loaded — every sheet, query and the project database
Open a database table — Browse the local DuckDB and open a table in its own tab. Menu path: Get Data → Tables → sales. Steps: 1. Open the Get Data menu 2. Click Tables — the project database 3. Refresh the list of tables 4. Every table and view in the local DuckDB 5. Click sales 6. It opens in its own tab — all 108 rows
Query DuckDB and pin the result — Write SQL against a DuckDB table, run it, and pin the answer to a sheet as static data. Menu path: Get Data → Empty Grid → SQL → Editor → Run → Sheet → Get Data → Static → Insert Table. Steps: 1. Open the Get Data menu 2. Add an Empty Grid — a sheet for the result 3. Open the SQL menu 4. Click Editor 5. Connection: Local DuckDB 6. Write the SQL — revenue and margin by product 7. Click Run 8. Close the editor 9. Go back to the new Sheet 10. Click the cell where the result should start 11. Open Get Data again 12. Tick Static — the values are written into the sheet 13. Click Insert Table
Export to Excel on a schedule — Export a table to an Excel file whose name carries today's date, then let the scheduler write it every weekday. Menu path: Settings → Parameters → asOf → Export → Excel → Flat file → Schedule → Add → Run now. Steps: 1. Open Settings 2. Click Parameters 3. Parameter asOf: the expression =TODAY() 4. Format YYYYMMDD — it resolves to today's date 5. Close — ${asOf} can now go into any file name 6. Open the Export menu 7. Choose Excel 8. Source: a range 9. The sales table, header row included 10. Destination: a file on a drive 11. Format: Excel (.xlsx) 12. The folder to write to 13. File name with the date: sales_${asOf} 14. Sheet name: Sales 15. Click Schedule this export… 16. The export is already picked: sales_${asOf}.xlsx 17. Give it a name 18. Choose At specific time & days 19. Weekdays at 07:20 20. Click Add 21. Click Run now to try it 22. ok — sales_20261013.xlsx written; tomorrow a new file, by itself
Yesterday's report to Excel — Parameter =TODAY()-1 in a SQL WHERE, pinned, then saved as reports_<date>.xlsx. Menu path: Settings → Parameters → Add parameter → yesterday → Get Data → Empty Grid → SQL → Editor → WHERE ${yesterday} → Run → Sheet → Get Data → Static → Insert Table → Right-click → Query Binding → Query Properties → Export → Excel → Flat file → reports_${yesterday} → Export now. Steps: 1. Open Settings 2. Click Parameters 3. Click Add parameter — a new row appears 4. Name it yesterday 5. Expression =TODAY()-1 — today minus one day 6. Format YYYY-MM-DD — the form SQL dates use 7. Resolved: yesterday's date — it moves on by itself every day 8. Close — ${yesterday} now works in SQL and file names 9. Open the Get Data menu 10. Add an Empty Grid — a sheet for the result 11. Open the SQL menu 12. Click Editor 13. Connection: Local DuckDB 14. WHERE order_date = '${yesterday}' — only yesterday's orders 15. Click Run — the parameter is filled in before DuckDB sees the SQL 16. Close the editor 17. Go back to the new Sheet 18. Click the first cell — the result starts here 19. Open Get Data again 20. Tick Static — the rows are written into the sheet 21. Click Insert Table — the result is pinned 22. Right-click the pinned result 23. Open Query Binding 24. Choose Query Properties 25. Headers: Include Headers — the column names become row 1 26. Click Update & refresh 27. Column names on top — every row is dated yesterday 28. Open the Export menu 29. Choose Excel 30. Source: Active grid table — the pinned result 31. Use first row as column headers — the names from row 1 32. Destination: a file on a drive 33. Format: Excel (.xlsx) 34. The folder to write to 35. File name with the date suffix: reports_${yesterday} 36. Sheet name: Reports 37. Click Export now
SQL and AI functions in the grid — Formulas that ask the database (SUM_SQL, COUNT_SQL) and a language model (AI_PROMPT) — right in a cell. Menu path: New sheet → Functions → Cell → SUM_SQL → COUNT_SQL → AI_PROMPT. Steps: 1. Click + to add a sheet 2. Open the Functions menu 3. The formula bar is on — formulas go there 4. Click the cell for the result 5. SUM_SQL(connection, table, column, filter) — DuckDB sums it 6. The database answered: 772,200 — no rows pulled into the sheet 7. Next cell 8. COUNT_SQL — counts five million rows in place 9. Every _SQL function runs on the database, not in the grid 10. Now a cell for the model 11. AI_PROMPT(model, prompt, range) — the range is the only data sent 12. The model's answer lands in the cell 13. Refresh AI asks again; Hold AI stops calls while you edit
Create a pipeline — Turn a manual routine into a repeatable, scheduled recipe. Menu path: Transform → Pipelines → New pipeline → Name → Bind → Top products → Add step → Sort rows → Save step → Add step → Format header → Step through → Run → Apply → History → Schedule → Add. Steps: 1. Open the Transform menu 2. Choose Pipelines 3. Click New pipeline 4. Give the recipe a name 5. Click Bind… to choose the data it works on 6. Pick the pinned Top products block 7. Click Add step 8. Choose Sort rows 9. Sort by units… 10. …Descending — biggest first 11. Click Save step 12. Add a second step 13. Choose Format header row 14. Centre the header text 15. Click Save 16. Click Step through to rehearse it 17. Run to end — a preview; nothing is written yet 18. Happy with it? Apply to data 19. Every run is logged in History 20. Click Schedule… to run it unattended 21. Choose At specific time & days 22. Every weekday at 06:00 23. Click Add
Pin a query — Place a saved query result into a sheet and keep it fresh. Menu path: Get Data → Empty Grid → SQL → Queries → revenue-by-month → Edit → Run → Sheet → Get Data → Static → Insert Table → SQL → Refresh Scheduler → Add. Steps: 1. Open the Get Data menu 2. Add an Empty Grid — a sheet for the result 3. Open the SQL menu 4. Click Queries 5. Choose Queries — every saved query 6. Pick revenue-by-month 7. Click Edit to open it in the SQL editor 8. Connection: Local DuckDB 9. Click Run — six months of revenue 10. Close the editor 11. Go back to the new Sheet 12. Click the cell where the result should start 13. Open Get Data again 14. Tick Static — load the full result 15. Click Insert Table — the query is pinned 16. Open SQL 17. Open the Refresh Scheduler 18. What to refresh: all grid bindings — the new pin included 19. Click Add — refreshed every hour
Build a query visually — Click columns, filters and sorting — the SQL writes itself. Menu path: Get Data → Tables → sales → Visual Query → SELECT → WHERE → ORDER BY → Code preview → Run query. Steps: 1. Open the Get Data menu 2. Click Tables to browse the database 3. Refresh the list of tables 4. Right-click the sales table 5. Choose Create Visual Query 6. Name the query 7. Click product to select it 8. Click revenue 9. GROUP BY product 10. SUM the revenue 11. Open WHERE to filter 12. Click Add condition 13. region = EMEA 14. Open ORDER BY 15. Click Add sort rule 16. Sort by product 17. Code preview shows the SQL it wrote 18. Ready-to-run SQL — no typing 19. Click Run query
Mask PII data — Make personal data safe to share — one way and repeatable. Menu path: Transform → Mask PII → A range → Header → Methods → Mask. Steps: 1. Open the Transform menu 2. Choose Mask PII 3. Mask a range — just the customer table 4. The customer table: C1R9:C10R27 5. Tick First row is a header — names stay readable 6. Customer numbers: Hash — still joinable, no longer readable 7. A city is not personal here — untick city 8. E-mails: Hash too 9. Phones: Partial… 10. …keep the last 3 digits 11. Tick bank_account too 12. Accounts: Partial… 13. …keep the last 4 characters 14. Click Mask
Secure rows and columns — Preview the project as the EMEA sales desk — its rows only, a bypass refused. Menu path: Settings → Set / Restore → Data access → Validate → Preview as → Pinned trades → Salaries → Show what runs → Bypass → Stop preview. Steps: 1. Open Settings 2. Click Set / Restore 3. Open Data access 4. Secured — a policy on 2 tables and 1 sheet 5. You own the project: every row, every column 6. Validate — what each persona would see 7. EMEA sales: EMEA trades only, no staff data 8. Preview as EMEA sales 9. Click Start preview — the project reopens as they see it 10. Commission empty, traders pseudonymised, e-mails masked 11. Close Settings — the strip says whose view this is 12. Pinned trades — re-read through the policy: EMEA only 13. Salaries — blank below the header for sales 14. Back to Settings 15. Click Set / Restore 16. Open Data access 17. Type a query into Show what runs 18. Click Show — nothing is executed 19. The SQL the server really runs: trades replaced by what EMEA sales may see 20. Now a bypass — SQL hidden in a string 21. Click Show 22. Refused — with a sentence that says why 23. Stop preview — back to the owner's view
Find anomalies and outliers — Clean invisible junk out of text, then flag numbers that do not belong. Menu path: Transform → Anomalies → Scan → Replace all → Outliers → Scan. Steps: 1. Open the Transform menu 2. Choose Anomalies — junk hidden in text 3. Scan a cell range 4. The imported block: C2R7:C7R27 5. Click Scan 6. Padding, hidden spaces, HTML entities — Replace all 7. Close 8. Now choose Outliers 9. Check a cell range 10. The revenue column: C8R7:C8R27 11. Click Scan — several tests at once 12. 384,000 — an extra zero; −12,600 — a credit note
Clean anomalies on a schedule — A pipeline that strips hidden junk from every new file, run every morning. Menu path: RawImport → Select → Transform → Pipelines → New pipeline → Name → Bind → Named range → Define & bind → Add step → Replace anomalies → Save step → Step through → Check → Apply → Schedule → Add. Steps: 1. The supplier file — click its first header cell 2. Shift-click the last cell — header plus 20 rows 3. Open the Transform menu 4. Choose Pipelines 5. Click New pipeline 6. Give the recipe a name 7. Click Bind… to choose the data 8. Name the current selection as a range 9. Call it supplier_file 10. Click Define & bind 11. Tick First row is headers 12. Click Add step 13. Choose Replace anomalies 14. Tick HTML entities — & becomes & 15. Hidden spaces and zero-width characters are ticked already 16. Tick Leading / trailing whitespace — "emea " loses its padding 17. Replace them with a space; edges are trimmed 18. Click Save step 19. Click Step through to check it first 20. Run to end — a preview; nothing is written yet 21. The fixes look right — Apply to data 22. Click Schedule… to run it unattended 23. What to refresh: the pipeline, already chosen 24. Choose At specific time & days 25. Every weekday at 06:00 26. Click Add
Build a dashboard — KPIs and charts bound straight to a database table. Menu path: Dashboard → New → KPI → Chart → Layout. Steps: 1. Open the Dashboard menu 2. Click New — an empty board 3. Add a KPI tile 4. Open Table 5. Choose sales 6. Open Value column 7. Choose revenue 8. Open Aggregation 9. SUM — total revenue, live from the database 10. Close the properties 11. Add a Chart 12. Open Table 13. Choose sales 14. Open Category (X axis) 15. Choose region 16. Click Add series 17. Open the series Column 18. Choose revenue 19. Close the properties 20. Drag the chart into place
Build a sales forecast — Monthly revenue projected nine months ahead, with bands and two scenarios. Menu path: Dashboard → New → Forecast → Properties → Time series → Forecast → Uncertainty → Scenarios → Accuracy. Steps: 1. Open the Dashboard menu 2. Click New — an empty board 3. Add a Forecast widget 4. Open Table 5. Choose sales — 10,000 orders in DuckDB 6. Open Date column 7. Choose order_date 8. Open Value column 9. Choose revenue — summed per period 10. Period: Month — or day, week, quarter, year 11. Forecast ahead: 9 months 12. Automatic — every method backtested, the best kept 13. Prediction intervals: 80, 95 % 14. Click Add scenario — Upside +10 % 15. Once more — Downside −10 % 16. Draw the test forecast over the history 17. Method, MAPE, MASE and RMSE print under the chart 18. Close the properties
Build a histogram — How order values are distributed — range, percentiles and a split by channel. Menu path: Dashboard → New → Histogram → Properties → Distribution → Bins → Range → Markers → Split. Steps: 1. Open the Dashboard menu 2. Click New — an empty board 3. Add a Histogram widget 4. Open Table 5. Choose sales — 10,000 orders in DuckDB 6. Open Value column 7. Choose revenue — one bar per price band 8. Bin rule: Automatic — or a count, a width, your own edges 9. Range of interest from 0… 10. …to 6,000 — larger orders gather in one end bar 11. Mark the Median beside the mean 12. Percentile markers: 5, 95 13. Open Split by 14. Choose channel — one distribution each 15. Open Bar height 16. Percent — each channel against its own orders 17. Close the properties
Connect a data source — Define a connection once, then import from it in a few clicks. Menu path: Connections → Add new → Add → Test → Get Data → Excel Imp → Preview → Import. Steps: 1. Click the Connections icon 2. Name it: Daily orders file 3. Type: EXCEL — or CSV, Parquet, Postgres, BigQuery… 4. Point it at the file 5. Click Add 6. Click Test — it opens the file 7. Click Save & close 8. Open the Get Data menu 9. Choose Excel Imp 10. Saved connection: Daily orders file 11. Click Load sheets 12. Load into a new sheet: Daily orders 13. Click Preview 14. 9 rows, clean — Import to grid
Send a PDF report — Package a heading and a dashboard into a PDF — now or on a schedule. Menu path: Export → PDF report → New report → Add block → Dashboard → Destination → Export PDF → Schedule → Add. Steps: 1. Open the Export menu 2. Choose PDF report 3. Click New report 4. Click Add block 5. Choose Text 6. Write the heading 7. Heading level 1 8. Add another block 9. Choose Dashboard 10. The Executive dashboard 11. Open Destination 12. Name the report 13. File name: sales_pack 14. Click Export PDF 15. Or Schedule… it 16. Choose At specific time & days 17. Click Add — the pack prints itself
Ask Greg about BigQuery — Connect BigQuery through MCP, then ask Greg a question it answers from it. Menu path: Settings → Set / Restore → MCP servers & API → BigQuery template → Name → Environment → Test → Save → Tools → Ask Greg → Data policy → Allow MCP → Ask. Steps: 1. Open Settings 2. Click Set / Restore 3. Open MCP servers & API 4. Add from template — Google BigQuery (MCP Toolbox) 5. The Parameters table explains every line you fill in 6. Give the server a name 7. An id for the API — it cannot change later 8. Project id, the PATH of the key file, the location — never the key itself 9. Test — checks the values, starts MCP Toolbox, lists its tools 10. Connected — nine tools found 11. Save — the server is ready 12. Click Tools to see what it may do 13. execute_sql is allowed — but only one read-only SELECT 14. Close Settings 15. Open the Ask Greg tab 16. Click Data policy 17. Turn on Let the agent use MCP servers 18. Close — the policy is saved 19. Start a new session 20. Ask in plain language — name the server 21. Ask 22. Each line is a call to BigQuery — a failed one is fixed and retried 23. Shipped leads — answered from BigQuery, masked and audited

Related guides

Automated Excel & scheduled reports

Glossary terms on this page

KPI dashboard: definition

ATX Institute on LinkedIn

This site needs JavaScript for the full experience. Every page above is reachable without it.