Upon our completion of the Sydney Lock-Out Laws impact analysis, we decided the findings would probably be of interest to more than just our immediate page followers. So we thought what better way to share our insights with the greater Sydney population than to run a LinkedIn Advertising campaign to get exposure for the findings, to encourage engagement with the outcomes of the laws, and to ultimately test the waters with the advertising platform to see just how sticky LinkedIn advertising really is.

Objective of the campaign

Our objective for this campaign was to build a broader following for the White Box brand across Australia (with a focus on the Sydney market).

How we did it

We created an easily digestible info-mercial of our key findings of our analysis piece which complimented LinkedIn’s formatting/feed styles and gave people the choice to engage with the story in their own way.

What we did

The campaign budget was spread over a 10 day period and launched on Tuesday 3rd December, following the mayhem that was the Black Friday and Cyber Monday weekend sales.

As the period included non-business days of Saturday 7th and Sunday 8th December, we allocated the daily budget limits to an amount slightly greater than the total budget divided by the number of days it would be active, with the intention to let LinkedIn role out the ad more to groups who were engaging and to not limit this engagement (too much) as a result of the budget. We’ll discuss the outcome of this strategy below.

The LinkedIn Campaign Manager - Objectives and Audience

LinkedIn, supposedly, is the ‘stickiest’ platform for ads in terms of engagement for a number of reasons including and definitely not limited to:

• compared to Facebook and Instagram, its users are far more likely to keep their employment details up-to-date, meaning targeting specific businesses or job titles is far more accurate

• like Facebook and Google, you can add insight tags to better capture target audiences and those who have engaged with your ad and website to refine future targeting efforts

• LinkedIn is a professional network with users inherently more aligned with our business and its offerings than users interacting on purely social platforms

Prior concerns, issues and thoughts surrounding the campaign

For this campaign, our main aims were not to ‘sell’, ‘convert’, or any of the other key buzz words you’d expect to see when reading a post on advertising. This is because we first want to find an audience that values our brand and business solutions, rather than push sales messages into the open and hope something sticks.

As such, we used the ad as a way to engage the wider Sydney professional network in order to establish industries, businesses and people who are more likely in our target market and (hopefully!) more willing to engage in with our business and content.

The live campaign featured a carousel of images walking the audience through the analysis and its core findings

Analysis of Results

There was a lot to take away from this campaign, and overly, we believe it was successful. It could definitely however use further tweaking and optimisation - we’ll discuss all of these factors now.

Overly, the campaign reached a large audience and increased our follower count by over 35%. Although we didn’t have a specific number of followers we were aiming for this time around, based on the budget and the amount of exposure we received from the campaign, this acts as a great basis point for future campaigns whilst allowing us to understand the capabilities of LinkedIn ad campaigns.

A shortcoming of the LinkedIn campaign manager however does exist for analysing these metrics, and that is that it doesn’t seem to be great at determining and reporting which followers were gained through the ad, and which were organic. Hopefully this will be fixed in due time.

Secondly, as you have probably now noticed in the above graphs, the campaign didn’t receive clicks on the launch date, or for the final three days we had intended. We put this down to two reasons:

1. it takes a bit of time for the campaign to find and get pushed out by LinkedIn into the feeds of the target market

2. If LinkedIn find target markets who are engaging with the content on any specific day, they will continue to push it until the budget for that day is exhausted - hence our budget allocation strategy we discussed earlier was flawed as LinkedIn found enough people each day to exhaust the daily spend limit.

Despite the campaign being live for a shorter period than we intended, there were other wins gained!

Interestingly, aside from the days of the weekend, the average click through rate increased from approximately 2.3% on the first day of audience engagement, to 3.8% at its highest point.

This suggests that as campaigns spend more time live, LinkedIn get better at finding appropriate audiences which are more likely to engage with your content. As these audiences are saved in your campaign history for future use, this is definitely a win in our books!

Now the question remains, with more ad spend and/or a longer time period for tuning, could we achieve an even higher Click Through Rate? As we move deeper into the marketing funnel, our targeting will definitely need to be more refined, but this campaign has definitely been a great tester to see the potential that lies on the LinkedIn platform.

We’d love to hear your thoughts about LinkedIn advertising as well as the Sydney lock-out laws. To join the discussion, find us on LinkedIn.

As debates about climate change quite literally heat-up, many people are interested in the factors, behaviours and events driving these changes. As data scientists the use of machine learning models are becoming more and more relevant for the completion of new types of analytical tasks, and as technologies like Neural Network Models lead these advancements, it’s important to understand the positive and negative effects of their employment to derive whether they should be incorporated into best practice for the future. 

The article, Deep Learning and Climate Change by Lukas Biewald, discusses the representation of Deep Learning models in the media, and how they are often misrepresented in terms of their negative impact on the environment.

The criticism arises from a paper called Energy and Policy Considerations for Deep Learning in NLP, which uses tables like these to compare the energy and costs of the use of different types of learning models. From the table, it is quite clear that NAS, or Neural Architecture Search model, emits extreme amounts of CO2 and has enormous cloud computing costs. 

The negative portrayal in the media and through papers like these comes about as a result of a focus on the most complex types of Deep Learning including Neural Networks which subsequently have the greatest carbon emissions. As such, with climate change currently a top-of mind issue for many people, it is easy to focus on the raw numbers without contextualising the entire scenario. In this case, Biewald argues that yes, NAS models can be cause for concern if they are employed by a great number more businesses, but at the moment, the technologies behind these types of machine learning and far too complex and unnecessary for the conduction of the great majority of business tasks, and hence do not warrant restrictions on their use.

Figure 1: a NAS, or Neural Architecture Search model, can take thousands of hours to train but most businesses don’t need models as complex as these.

In the bigger picture, this article is interesting because it takes the perspective of an industry professional on how non-professionals or the media can influence messages to suit a narrative. That is, it is always important to understand the context of what a message claims, and the implications that context might have on its broader application. At White Box, some of our most important work is identifying what clients’ data means in the context of their own business and industry, and how the information gathered from their data can be used in a way that is impactful yet doesn’t stray away from the true message of the brand or business vision. As such, like in this article, it is sometimes extremely useful to have an ‘out of the box’ perspective on the real implications of trends, data and other factors influencing business, or the use of Deep Learning, in this case, in order to create clarity and drive development in a way that is beneficial, rather than crippling, to longer term goals.

Figure 2: it is worthy of mentioning, when contextualised, training neural architecture search models can omit 4x more carbon that an average car over its entire lifetime

---

For the original article, click here.

For more data analysis and visualisations, click here.

Or, get in touch for a discussion about your data strategy.

There is no denying the evidence for climate change. But just how dire is the current environmental situation and are there mechanisms being put in place to curb these impacts?

The image above displays the consistent rise in deviations above the global mean land and sea temperature average. The World Economic Forum discuss how 11,000 scientists recently came together to co-sign a letter in the journal BioScience as a result of these types of evidence, warning that further neglect can and will cause irreversible chain reaction events. Despite this evidence, graphs like these need to be taken with a grain of salt. As data scientists we know that correlation doesn’t always imply causation, but the seemingly exponential rise in global temperatures and global populations increased consumption of carbon based products and land clearing are very strong signs that we are in fact inducing a potential future catastrophe.

---

For the original article, click here.

For more data analysis and visualisations, click here.

Or, get in touch for a discussion about your data strategy.

In February 2014, Sydney introduced new laws to curb alcohol-fuelled violence within Sydney CBD (1.30am lockouts and 3am last drinks at bars, pubs and clubs).

With the latest news that the lock-out laws will be lifted from 14th January 2020, we looked back at the impact the laws had on related crime within and just outside the designated zones.

Methodology

We looked at four key crime offences (domestic and non domestic violence related assault & offensive language and conduct), from 59 months pre lock-out (Mar-09 to Jan-14) compared to 59 months post lock-out (Feb-14 to Dec-18).

Data from Bureau of Crime Statistics and Research.

We have contextualised by showing the NSW overall trend.

Interactive map

The analysis below uses an interactive map that we built. You too can dig for insights. Click here (best viewed on a laptop).

Analysis summary

Over the analysis time period, NSW has seen a decrease of 11.6% in these four offences:

So with this in mind, we would expect our locked down entertainment zone to also exhibit the same trends, which it does (see the black border indicating the lock-out areas).

The green colour represents those suburbs that have shown a decrease between the time periods; so all of the Lock-out zones have seen a positive impact.

Red indicates an increase, which we can see in Pyrmont, Double Bay, Zetland and the inner west areas around Newtown.

Our tool allows you to hover over a suburb and see the underlying data.

So for Potts Point, home of Kings Cross, one of the main troublesome areas, we can see the huge impact the lock-out law had; a 56% decrease in these offences.

Whereas Pyrmont (home of the Star casino) has seen an increase of 51%!

Given that the rest of NSW has seen an overall drop of 11%, this is a huge discrepancy.

This hasn’t been the case for all proximal displacement areas, such as Surry Hills and Paddington, even though there is a large pub and bar scene.

Newtown (part of the distal displacement area) has seen an 11.3% increase, with similar increases in nearby Camperdown and Erskineville.

Another prime displacement location is the Golden Sheaf in Double Bay. This area has seen an increase of 25%.

While both of these percentages are large, the underlying volumes of incidents is still relatively small compared to Kings Cross and Sydney CBD in general.

Conclusion

The lock-out laws have had the desired impact on the zoned areas and although the displacement impact has been quite heavy in terms of percentage change, the overall volumes of incidents has decreased dramatically.

Here are the numbers:

The lock-out area has seen a decrease of over 6,000 incidents.

The largest volume increase was seen in Pyrmont of 434 incidents.

So the net gain has been very positive in terms of crime statistics but at what cost? With pedestrian traffic dropping by 80% and Sydney having to deal with a nanny state image problem, will the reversal of the lock-out laws bring back a nightlife culture with lower crime statistics? Time will tell.

Side analysis

Whilst investigating the lock-out laws, we spotted an overall increase in Domestic violence related assaults throughout NSW. Although not a direct link to the lock-out laws, this was highlighted in 2015 as a Priority from the Premier (“Reducing domestic violence reoffending”), so this is rather concerning.

If we pan out, the number of red hot spots (% increase) is quite varied across the Sydney region, although keep in mind, that some volumes are quite low (we’ve removed any suburbs with less than 100 incidents):

We took this side analysis a little further and calibrated to rising population figures in NSW but the numbers are still tracking higher.

“Unreported as a crime” context

It is possible that the publicity of the Premier’s campaign and movements like the #MeToo would have caused more people to come forward and report domestic violence, compared to the pre period.

We’re hopeful that this is the reason but will continue to dig for more data sources and reports to find out more.

For more data analysis and visualisations, click here.

Or, get in touch for a discussion about your data strategy.

As the UK election race hots up and the challenges of the NHS come back into the limelight, the NHSX have produced a report detailing the incorporation of AI into the service and what impacts this could have.

The NHS is of crucial concern to a great proportion of the UK economy, and hence the developments of technology which improve the services and increase its staff capabilities are of equally crucial importance. Figure 3 shows exactly where the use of AI will come into effect, with diagnosis of illness consisting almost 50% of the total usage, highlighting the advancements in image recognition machine learning and the standardised digital format of data inputs for a range of diagnoses.

Despite the upsides present, it must be noted that the role out of newer technologies won’t happen instantaneously. A quarter of the proposed products are ‘very likely’ to be introduced to the NHS in the next year, but it will take at least 5 years for the great majority to be incorporated. 

Some other facts from the report include: 

• 19% of AI solutions are being developed on algorithmically generated datasets (machines gaining insights from processes and using these insights to make further decisions about products) 

• 44% of all GDHP members have a national or regional policy framework for the development and/or deployment of AI

---

For the full report.

For more data analysis and visualisations, click here.

Or, get in touch for a discussion about your data strategy.