At most one email a week. Unfollow anytime with one click.
Transcript
The full narration of this video. The current sentence is highlighted while the video plays; click any sentence to jump there.
Show transcript
Introduction
Hi, and welcome to the Intermediate Artifacts session. In this session we build mini apps for our own work: dashboards, quizzes, calculators, forms, and diagrams, all without writing code. In the first episode of this session we build an interactive sales dashboard. One month of data turns into number cards, charts, and filters you can click.
Step 1: Prepare the sales data
Step one is preparing the data. Laras has Sunset Coffee's sales records for October. Each row holds one transaction: the date, product, quantity, total, and sales channel. Row two is a cold brew sold through Instagram. Notice there's no column for customer names or phone numbers. Those columns actually exist in Laras's original records, but she deleted them before uploading. A sales dashboard doesn't need to know who the buyers are. The next rows hold a gift hamper from the order form, and an iced coffee cup from Instagram. In the last row, Laras writes a summary: two hundred sixty transactions, totaling twelve and a half million rupiah. She calculated this with a SUM formula, and later it becomes the benchmark for checking the dashboard. Before uploading, check five things. First, each row holds one transaction, not a whole day lumped together. Second, column names and date formats are consistent from top to bottom. Third, the period is clear, for example one full month from the first to the thirty-first. Fourth, you've calculated the total yourself as a benchmark. Without a benchmark, we have no way of knowing whether the dashboard calculates correctly. And what to avoid: customer names or phone numbers still in the data. Delete those columns first, or replace them with customer codes if you really need them.
Step 2: Ask for the first dashboard version
Step two is asking for the first dashboard version. The key is to start from business questions, not chart types. Laras wants to know the best-selling product, the busiest channel, and the best week. The data is the October sales file she attaches. The format is a one-page dashboard: three key number cards, a bar chart of revenue per product, and a summary per channel. The constraints: use numbers from the file only, and show rupiah with commas for thousands so it's easy to read. Laras sends that prompt along with her file. Claude reads the October sales CSV file. CSV is a simple table format you can export from Excel or Google Sheets. The dashboard appears in the right panel. At the top are three cards: revenue of twelve and a half million, two hundred sixty transactions, and an average of about forty-eight thousand per transaction. Below them is a bar chart of revenue per product. The iced coffee cup is highest, at almost six million, followed by the one-liter iced coffee, cold brew, and hampers. The last line sums up the channels: WhatsApp fifty-five percent, Instagram thirty, and the order form fifteen. All of this came from one prompt, without a single line of code from Laras. At the top of the panel there are two tabs: Preview and Code. Laras is curious, so she asks what's in the Code tab and whether she needs to change it. The Code tab holds the recipe that makes the dashboard work. Its lines roughly mean: read the sales file, add up the total column, show three cards in rupiah, then draw a bar chart per product. You don't need to understand this code, let alone change it. Every change can be requested in everyday language, and Claude adjusts the code.
Step 3: Add interactive filters
Step three is adding filters. Laras asks for two filters: channel and week. The cards and charts have to update with them, and she also asks for a chart of revenue per week. Version two appears. At the top there are two choices, channel and week. The example in the panel shows WhatsApp for all weeks. The cards change: WhatsApp revenue is about six point nine million from one hundred forty-three transactions, or fifty-five percent of the total. The weekly chart shows week four is the highest. Claude guesses the reason is end-of-month paydays. That guess makes sense, but it's still a guess, not a fact from the data. Before sharing, Laras tries each filter choice one by one. If a filter makes the numbers blank or strange, she simply reports it to Claude to be fixed. Which filters are useful? For most businesses, start with the sales channel, such as WhatsApp, Instagram, or a marketplace. Second, the period. Besides weeks, compare weekdays with weekends. Third, product or category. For Maju Bersama Co., the categories could be paper, folders, and writing supplies. Fourth, region, such as in town and out of town. Pick just two or three filters. Too many filters only make a dashboard confusing.
Step 4: Read the insights and check the numbers
Step four is reading the insights and checking the numbers. Laras compares the dashboard total with the SUM formula in her spreadsheet. The numbers match: twelve and a half million. She also asks how the average was calculated. Claude explains the formula: revenue divided by the number of transactions. Then it gives insights split into facts, a guess, and a suggestion. The fact: the iced coffee cup brings in almost half the revenue. The guess: week four rises because of payday. The suggestion: a hamper promo in week four. This split shows Laras what needs testing first. To test the payday guess, Laras can compare the fourth week of October with the fourth week of November. If the pattern repeats, the guess gets stronger and becomes a sound basis for a promo. Make these four checks a habit. First, total revenue must match the SUM in the original spreadsheet. Second, the number of transactions must match the number of data rows. Third, try one filter, then calculate one of its numbers by hand. For example, total Instagram revenue in the first week. Fourth, if a number looks odd, ask how it was calculated. Claude can misread a column or round incorrectly. A beautiful but wrong dashboard is more dangerous than having no dashboard at all.
Step 5: Use it every month
The last step is using the dashboard regularly. At the start of December, Laras uploads the November data and asks for the same layout and filters, plus a comparison with October. Claude reads the November file just like before. The November dashboard appears with new cards: revenue of thirteen and a half million, up eight percent. The table below compares each product month to month. It turns out the biggest rise came from cold brew. That's the power of a regular dashboard: small changes show up right away, so Laras can ask why and decide on her next move. Let's sum up. First, prepare tidy data that's free of personal information. Second, ask for a dashboard based on clear business questions. Third, add two or three filters you'll actually use. Fourth, check the numbers against the original spreadsheet before trusting them. Fifth, update it at the start of every month. You can also share this dashboard with Dimas through a link, as we learned in the Artifacts Basics session. Make sure there's no personal data before you share it.
Done!
We've built an interactive sales dashboard: preparing data without personal information, creating number cards and charts, adding filters, checking the numbers, and updating it every month. In the next episode we build a different kind of mini app: interactive quizzes and exercises for students, new employees, or customers. See you there!
Practice exercises
Try these on your own after watching to lock in what you learned.
Upload one month of sales data and ask for a dashboard with three number cards and one chart
Add a filter by sales channel or by week
Check the total revenue on the dashboard against a SUM formula in your spreadsheet
Ask the class material
Type a question or keyword; we find the part of this episode that explains it and jump the video there.
My notes
Notes are saved with the current video time and are only visible to you.
Questions & answers
Stuck on a step? Ask here; the instructor and other learners can answer.
No questions yet.
Your instructor
Cerdas Inovasi Digital Team
Software developers, BIM & 3D engineers, and AI practitioners
53 video classes1,286 episodes
We build software, BIM models, games and AI automations for clients every day, and turn that hands-on work into step-by-step classes: every step is shown on screen and comes with a PDF guide you can keep.
