Scatter charts from Google Sheets to Slack, when correlation is the story

Scatter charts from Google Sheets to Slack, when correlation is the story

Scatter plots are the most under-used chart type in business reporting. They're the one shape made specifically to answer "is X related to Y?", and that's a question that comes up constantly: does spend correlate with revenue, does usage correlate with retention, does session length correlate with conversion.

If you're sending one to Slack regularly, you probably want to know about the relationship more than the individual numbers. Chartcastr renders scatter plots straight through.

What scatters are for

Each dot is one observation with two coordinates. The pattern of dots tells you what relationship, if any, exists between the two variables.

Cases where scatters are the right call:

• Account spend vs account retention. One dot per customer.
• Campaign budget vs campaign conversions. One dot per campaign.
• Ad creative impressions vs CTR. One dot per creative.
• Days-since-signup vs activation rate. One dot per cohort.
• Time on site vs purchase amount. One dot per session, sampled.

The defining feature: each dot is an entity, and you're asking whether the two axes are related across entities.

Reading them

Scatters reveal four things, in roughly this order:

1. Direction. Up-and-to-the-right is positive correlation. Down-and-to-the-right is negative. Cloud is no correlation.
2. Strength. Tight cluster around an imaginary line is strong. Diffuse spread is weak.
3. Outliers. Single dots far from the cloud are often the most useful information in the chart. They're customers, campaigns, or sessions that broke the pattern.
4. Clusters. Multiple distinct clouds suggest segments hiding in the data.

A trend line over a scatter is fine if Sheets is configured to show one. Chartcastr passes through what you've set.

The Slack pattern

Scatters work less well as daily reports than as periodic deep cuts. The cadence we see most:

• Weekly or bi-weekly. A Monday-morning scatter that shows the relationship for the past N days.
• Paired with a summary. "ROAS clustered around 1.8x for most campaigns. The AU Lookalike was an outlier at 4.2x. Worth digging into the creative."
• Sized small in Slack, with a link to the source. Thumbnail in the channel, click through if you want to inspect the dots.

A shape that works well

For agencies and ops teams: customer health scatter. X-axis is product usage (logins per week, or feature events per week). Y-axis is contract value or NPS. Each dot is one customer.

Send weekly. The chart instantly shows two segments to focus on:

• High-value, low-usage dots, at risk
• High-usage, low-value dots, expansion targets

That's a scatter plot earning its keep.

Setting it up

1. Build the scatter chart in Google Sheets, two numeric columns, one dot per row.
2. In Chartcastr connect the Google Sheets source and pick the chart.
3. Pick a destination, Slack channel works. Email also fine for less time-sensitive scatters.
4. Schedule it.

Range, labels, and trend line preferences come through. If you change which columns are the x and y in Sheets, the next pulse follows. Full reference at the Premade Charts docs.

Where scatters fall down

• Too few dots. Below ~10 observations, a scatter is just a few floating points. Use a table.
• Too many dots. Above ~500 the chart becomes a smudge unless you're using opacity or jitter (Sheets doesn't really do either). Sample down or aggregate.
• Categorical axes. If one of your axes is a category, region, channel, a scatter is the wrong shape. Use bars or a small-multiples panel.

For comparing time series rather than relationships across entities, see line charts. For composition, see stacked bars.