10,000 bouquets. 10,000 colours.

A Data Methodology

Does the time of year affect the colours of the bouquets we choose to buy?

Does our gender affect the colours of the bouquets we choose to buy?

How about where we live? Or the occasion the bouquets were purchased for? Or even the amount we spend on those bouquets?

We sought to answer these questions using data, our data, specifically. For each of the tens-of-thousands of bouquet orders in our database, we calculated the most prominent (or 'dominant') colours apparent in the photos of those bouquets.

Here's an example of this process:

So what was the mathematical formula we used? It works in a similar way to how you would find the average of a set of numbers; but instead of numbers, we used the colour-value of the pixels that made up each of our bouquet photos (after all, a digital image is merely a set of numbers representing red, green and blue!). And we repeated this process for every order in our database, in essence, turning a database of numbers into a database of colours: these our 10,000 colours.

From here we were able to arrange these colours in different ways. For example, in the image below we've taken all of our bouquet orders in January and stacked up their associated dominant colours like bricks in a wall, or bars in a barcode.

What this allows us to see is a birds-eye-view of the colours our customers prefer in the first month of the year. Looks cool, but on its own, it doesn't tell us much. We need to go a step further.

When you plot the dominant colour of each of our bouquet orders, split by each month over an entire year, you begin to see patterns emerge. Red bouquets spike in popularity during February (for Valentine's Day), but also during December… perhaps warming red flowers, consciously or unconsciously, are purchased to counterbalance the cold beginnings of winter.

Orange bouquets make a very slim proportion of orders throughout the spring and summer months, but come autumn they become far more popular… perhaps matching the colours of the leaves on the trees as they turn from green to orange to brown. (You might be wondering what all the dark green bars represent: the leaves and stalks - some bouquets feature more greenery than petals, and so this is reflected in the study.)

But what about flower bouquets on different occasions? What about how much people spend on those different occasions? Take Valentine's Day, for example. Do people who spend more on Valentine's bouquets go for different colours than those who spend less? Turns out, they do:

In the graphic above, the dominant colour of each Valentine's order was converted into a ring instead of a bar, creating a kind of fingerprint of bouquet colour choice: low spenders go for blues, pinks, and violets. High spenders go for deep reds and light pinks. And when we compare the colour fingerprint of Valentine's Day to other occasions...

Each occasion has a unique colour fingerprint:

• Overall, Valentine's Day features intense and passionate tones.
• Christmas features warming fireside red and orange tones.
• Mother's Day features pink rose tones.
• Anniversaries are a mixed bag, featuring roughly equal amounts of red, violet, yellow and orange.
• Bouquets ordered to celebrate the birth of babies, naturally, feature baby pink tones.
• And the energetic atmosphere of a birthday is matched by the bright, electric yellows and oranges.

We're busy building out more stories and finding more insights into this data. So stay tuned! In the meantime, we think these graphics make for some pretty sweet posters.