Exploring Google data in the run up to the 2017 election

As we get closer to New Zealand’s election, people seem to be getting more and more addicted to the polls. It seems like data’s never been so sexy.

When we look at New Zealanders searching habits (thanks to Google data) it doesn’t necessarily suggest voting patterns. But we can definitely get insight into what party NZers are curious about.

In this post, I’ll be exploring the number of Google searches for each New Zealand political party.

As of writing this, August was the most recent data released, as new data comes available I’ll update this post.

Dates of searches: Jan – August 2017
Search Location: New Zealand

1. Party Search Query

A. Keywords

I examined 3 keywords for each party.

  1. (_____) party – – – – – – – – – Eg: Labour party
  2. the (_____) party – – – – – – Eg: the Labour party
  3. (______)  – – – – – – – – – – – – Eg: Labour (more on this one later)

These search terms were added together to get the total search traffic for each month from January – August 2017.

B. Proportion of Searches

Total times NZers searched for a political party (August). Eg – “Greens”

What it means:

When we add all searches made in August 2017, we can see that Labour takes 23% of the pie – eclipsing the Greens who’re on 19%.

TOP landed a massive 18% of the total search traffic with 21,700 searches for August – this is huge for a new party. Despite National’s position as one of the major parties, they only received 17% of the search queries.

C. How have searches changed?

Political party searches per month – Eg: Googles for “National” during May

What it means:

Looking back at August, the total number of searches have dramatically increased, as we get closer to the end of the election season.

You can see a massive spike in search traffic for Labour as Jacinda Ardern becomes the leader (1st August). Labour netted nearly 20,000 additional searches between July and August – an increase of 380%.

Despite the Greens poor performance in the polls, there’s still a fair bit of public interest in the party. Being the second highest searched party in August, with an increased search volume of 330%.

TOP came in third for August, but had one of the lower percentage increases from July at 200%. For a party that’s polling at around 2% with not a huge amount of traditional media airtime, there’s an incredible amount of public interest in them.

National is driving the 4th most searches with just over 20,000 searches in August. There’s been a 350% increase in searches for them since July.

D. Notes on the Method

For the third keyword I mentioned – the (_____) keyword – I did some data scrubbing to ensure the data was as accurate as possible.

To be safe, I took the number of searches in the most politically quiet time (Jan 2016) and subtracted them from all of these top level searches.

For example: “Act” had 1000 searches in Jan 2016, so 1000 was subtracted from any “Act” searches from January – August 2017.

This is an important baseline for the more general words like “National” and “Labour”, where searchers could be seeking a variety of different information.

The one keyword this didn’t apply for was “Maori”. This is because if someone searches “Maori” they’re most likely not looking for the Maori Party. If they are, they won’t find the information they’re looking for, so will likely do another search. This next search will probably be “maori party” – which has been captured by the other keywords I tracked.

Therefore, the “Maori” keyword was excluded from this analysis.

2. Meta Data

A. Where are people searching from?

What it means:

Wellington seems to outperform everywhere when it comes to the number of political searches per capita.

B. What device are people searching from?

What it means:

There’s a large number of people who are searching from mobile devices and tablets. However most searching of this type is done from desktop computers or laptops.

3. Enrolling to Vote

How have searches changed?

Political party searches per month – Eg: Googles for “National” during May

What it means:

As the election dates are very similar (20/9/14 vs 23/9/17), this is a rather fair comparison. Obviously, there are a number of factors here, not limited to the amount of orange-guy enrol to vote advertising. However, these numbers can give us some insights.

If we compare August last election (2014) with this election, we can see that the numbers of people who are Googling “Enrol to vote” are actually up by 1,800 searches.

This is likely driven by having such an interesting election coverage, with new leaders, scandals and massive changes in polls.

It’ll be interesting to see the number of people Googling this for September, and whether or not this is higher this election vs the last one.

Tim Dorrian
Managing Director / Founder

Tim kicked off Aro Digital from his flat back in 2017. Seeing digital campaigns go live and driving results for purpose-driven businesses is what makes him front flip out of bed in the morning. A parkour master, WWII bunker expert, and SEO whiz, Tim is as adventurous as he is passionate about all things digital. He’s on a mission to do every nature walk in Wellington, so if you know any hidden gems - let him know!