What these tools are for
Most online tools for working with data fall into two groups. Spreadsheets store rows and columns and let you calculate on them. Charting tools take a table and draw it as bars, lines or points. Many spreadsheets include charting, and many charting tools accept pasted or uploaded tables, so the line between them is blurry. What matters is the sequence of work: get the data into a tidy table, summarize it, then draw it.
Skipping the middle step is the most common beginner habit. A chart made straight from raw rows often shows noise, not a pattern. A short summary first tells you what the chart should actually show.
Start with a tidy table
A table that works well in almost any tool has one row per observation, one column per kind of value, and a single header row with clear names. Avoid merged cells, blank spacer rows and notes typed into the middle of the data. If your numbers came from several sources, some cleanup is usually needed first, and data cleaning is worth doing before any calculation, because duplicates, stray spaces and mixed date formats quietly distort totals.
If you are still getting comfortable with the tools themselves, building up your spreadsheet skills first makes everything below easier.
A small worked example
The numbers below are hypothetical, made up only to show the arithmetic. Imagine a small newsletter with these new sign-ups per month:
| Month | New sign-ups |
|---|---|
| Jan | 120 |
| Feb | 135 |
| Mar | 150 |
| Apr | 130 |
| May | 160 |
| Jun | 185 |
Three quick summaries are usually enough to start.
- Total: 120 + 135 + 150 + 130 + 160 + 185 = 880 sign-ups over six months.
- Average: 880 divided by 6 is about 146.7 per month.
- Median: put the values in order (120, 130, 135, 150, 160, 185). With six values, the median is the mean of the middle two: (135 + 150) / 2 = 142.5.
The average and median are close here, which suggests no extreme month is pulling the average around. When they differ a lot, the data is skewed, and a note on median versus average explains why one number can hide so much.
Change over time is the other useful summary. From January to June, sign-ups rose by 185 - 120 = 65, and 65 / 120 is about 54 percent. From March to April they fell by 20, and 20 / 150 is about 13 percent. Both are true, and a chart shows them together better than a sentence can.
In an online spreadsheet, each of these is one short formula, covered in basic formulas. If the table grows to hundreds of rows with categories, pivot tables do the grouping for you, and filtering data lets you look at one slice at a time.
Turning the summary into a chart
Pick the chart from the question you are asking, not from what looks impressive.
- Change over time: a line chart, or bars if there are only a few periods.
- Comparing categories: a bar chart, ideally sorted from largest to smallest.
- Parts of a whole: a pie only works with a few slices that clearly add up to a whole; otherwise a bar chart is easier to read.
- Relationship between two numbers: a scatter plot, with the caution that a visible pattern is not proof that one thing causes the other.
For the sign-up example, a line chart with months along the bottom and sign-ups up the side answers the question “is this going up?” immediately. The details of matching chart types to questions deserve their own page.
Common chart mistakes
- A truncated vertical axis. If a bar chart starts at 100 instead of 0, the January bar in our example would look tiny next to June, exaggerating a 54 percent rise into something that looks like several times larger. Bar charts should generally start at zero. Line charts have more room, but say so if the axis does not start at zero.
- Too many series or colors. If readers must keep looking back at a legend, simplify. Show fewer lines or split into several small charts.
- Missing labels. Every chart needs a title that states what it shows, labeled axes with units, and a note on the time period.
- Decoration over information. Three-dimensional effects, heavy shadows and background images make values harder to read.
- Cherry-picked ranges. Showing only the months that make a trend look good misleads, even if each number is correct.
- Colors that fail some viewers. Do not rely on red versus green alone, and keep enough contrast between colors for people with color vision differences.
Check before you share
Most embarrassing chart errors begin as spreadsheet errors: a formula that skips a row, a range that stops one cell short, a number stored as text. Compare your totals against a hand calculation on a small sample, and read through the usual spreadsheet errors before sending anything to others. If you share the file or chart online, also think about who can view or edit it, since sharing permissions apply to data as much as to documents.
When one chart becomes several
Once you track the same few numbers regularly, you may want them on a single page. A simple dashboards approach, with a handful of numbers and two or three charts, is usually more useful than a crowded wall of visuals. Keep each chart tied to a question someone actually asks, and drop the ones nobody looks at.
Keeping conclusions modest
Charts make patterns look authoritative, but they only describe the data you gave them. Six months of sign-ups says little about the seventh. A small data set, a single unusual month or a change in how the data was collected can all produce a pattern that does not last. Describe what the chart shows, note what it cannot show, and treat forecasts as possibilities rather than predictions.