Track logos on social media with image analysis

Watching eye

Geoffrey Hinton is the godfather of ‘deep learning’, especially in the context of Search Engine Optimisation (SEO). Deep learning is a branch of machine learning that describes the ability for computers to show aspects of artificial intelligence through complex algorithms. It’s set to revolutionise the digital marketing industry in the next five years; giving software the ability to understand what images and videos are actually showing.

TechCrunch yesterday, quoting from a Reddit AMA session, revealed Hinton’s opinion of the most exciting area in the next five years:

“I think that the most exciting areas over the next five years will be really understanding videos and text. I will be disappointed if in five years time if we do not have something that can watch a YouTube video and tell a story about what happened.”

However, a development in the social media insights industry may have just called this development five years early. Social insights tool, Crimson Hexagon, has just introduced image analysis to measure ‘Share of Eye’ within its social analytics platform. At this stage focused on logo detection, that has already proved effective for O2 and other brands.

Crimson Hexagon logo

The new offering enables brands and agencies to identify, track and analyse logos contained in billions of social media posts across social networks. As consumers increasingly rely on photos to share meaningful moments on social media, the image analysis capability gives brands the ability to better understand and measure their value in the context of customers’ everyday lives.

For years, brands and agencies have relied on social analytics technology to track and analyse social mentions in order to derive insights that help shape marketing campaigns, loyalty and retention programs, advertising strategy, corporate messaging and more. But as consumers increasingly share experiences via photo and video, brands have struggled to find a way to track visual mentions. This is especially true when a brand’s logo appears in a photo, but its name is not mentioned in the social post.

Crimson Hexagon’s image detection capabilities solve this problem. By incorporating this visual data, ForSight™ now gives brands a complete picture of how consumers are talking about them on social media.

Crimson Hexagon plans to launch additional features later this year that will continue to help customers thrive in the visual social media era. Image analysis is currently in beta. To schedule a demo, you’ll need to contact them.

Fintech revolution in UK: 30,000+ social media conversations

London City

It’s been reported that the Danish Chamber of Commerce is recommending that shops and services be given the option of going cash-free. A move, that if it goes ahead, could kick-start a cashless revolution across Europe; from a consumer option to a necessity. The proposal implies that all of us are comfortable as consumers with new methods of payment, made possible by the growth of fintech (financial technology).

Innovation is addictive. You’ll feel what I mean when you step off at Old Street station; the gateway of London’s Silicon Roundabout. It’s the home of Google Campus and a number of other technology companies who all aim to disrupt ‘business as usual’. The most successful of these companies are commonly referred to as fintech. They are making quite a splash in the UK. Our financial services sector currently accounts for approximately 9.4 percent of GDP, with fintech generating £20bn in revenue in 2014.

This year there have been over 30,000 conversations on Twitter about #fintech in the UK, with the future of payments and banking innovation appearing as popular topics, along with a strong relationship with financial innovation and start-ups. These conversations are not just limited to industry chatter, but include consumers discussing and exploring how fintech could provide them with more options in their personal and professional lives.


The fintech revolution isn’t just happening in London; Edinburgh, Belfast, Leeds, Manchester, Birmingham and Cardiff are all emerging leaders. The scale of fintech is astonishing, encompassing mobile payments, P2P lending and crowdfunding.


The popularity and rise in fintech has generally been associated with two developments: the fact UK consumers are sophisticated and open to new financial models, and high internet and mobile penetration rates. We can see both of these developments present in conversations on Twitter, through mentions of #bigdata and #iot (Internet of Things).

Fintech – aside from providing a positive B2B and B2C revolution, has serious consequences. Significant growth of cryptocurrencies (such as Bitcoin) has been noted by the Bank of England to potentially pose a material risk to monetary and financial stability of the UK. Highlighting that fintech not only provides innovations in payment technology but can also leave central banks with little control of fiscal matters (i.e. digital currencies decentralising monetary control).

Fintech will continue to grow in the UK and is a significant part of the future of financial services. For similar insights and deeper discussion, consider attending Lansons’ Future of Financial Services Conference on the 2nd June 2015.

We must protect our access to social media data (#CIPRmanifesto)

I’m not sure whether to read the first line of the CIPR’s manifesto with cautious optimism or pure elation, “The UK General Election of 2015 promises to be most unusual”. Although there is no denying this statement’s truth – from May 2015 we’re going on a roller-coaster ride of unexplored political territory.

The CIPR outlines a number of areas for the next government, including the importance of PR and public affairs, gender equality and Internet governance. Having met the CIPR editorial team, drafting this document may have been stressful – over 11,000 members are about to pass judgement!

However, I feel apprehensive with the CIPR’s summary of data protection in the document. It seems dated as it fails to understand the huge developments that have taken place in the EU. It may also be in danger of going against how CIPR members are currently using social media data to deliver services.

The PR industry relies on the trading of social media users’ data so that it can deliver services and evidence its effectiveness to clients.

When you sign up to Twitter (or any social network), as mentioned by the CIPR, users need to give away their personal data so that it can be bought and sold by Internet organisations. This is the lifeblood of every social media monitoring tool; be it Brandwatch, Radian6 or Pulsar. Without it social media practitioners, such as myself, are unable to effectively manage or measure online activities.

Cynically it could be said that we rely on users blindly signing up to a social networks’ 15,000 word terms and conditions, because without access to user data our jobs are near impossible. Of course, a user only gains from giving away personal (disposable) information and therefore shouldn’t worry what happens to it. It’s just a tiny sacrifice that is made so that a social network can be used.

Keeping all of this in mind, the CIPR concluded on data protection,

“We therefore have one part of our data environment which is highly regulated and in which citizen’s rights are regarded as pre-eminently important, and another in which there is effectively no meaningful control on the use and re-use of data. This lacks coherence, is not sustainable and will ultimately lead to frustration.”

It’s extremely unclear and disconcerting that the CIPR has suddenly decided to take a stand on data protection without consulting the membership. Especially when it is such a hazy view – not good for clear manifesto points. The “… effectively no meaningful control on the use and re-use of data” directly implies the transaction of a social network selling data to a social insights company. That, for the PR and public affairs industry, must be protected at all costs.

Especially as a large public debate will continue to bubble as the current EU Data Protection Regulation considers the impact of globalisation and social media on data protection (across every EU member state).

The point of this post isn’t to say that the CIPR’s stance is wrong, only unclear. It frustrates me that my representative body hasn’t consulted its members on data protection, especially as there must be hundreds of better-qualified thinkers than I who could have contributed to this manifesto.

I would like to see the CIPR take more action on data protection and develop the ideas summarised in their manifesto. However, we won’t even impress our political allies unless we can demonstrate an up-to-date knowledge of current data protection developments.

But… despite the little bit on data protection, the rest of the manifesto is fab!



Google Analytics adds Cohort Analysis (Beta)

Google has released cohort analysis as an option on the Google Analytics dashboard. It allows you to breakdown an overall visitor number into different groups (known as cohorts), and then analyse visitor behaviour based upon common experiences visitors may have had. For example, you may want to compare how users who visited your website from an email campaign compared with those who had visited from Twitter Ads.

It’s a form of analysis I first came across in my past role as a Multinational Account Manager at Microsoft – online advertising analysis is generally way ahead of digital PR measurement.

Google Analytics’ cohort analysis is still in beta version and therefore it is extremely basic. As of writing this post you are able to select from a number of segments (existing metrics on Google Analytics such as device or traffic type), but the rest of the options are a little bit… simple.

  • Cohort type: Can only view how visitors to a website behaved after a certain date
  • Cohort size: Can only be by day, week or month (so a general overview)
  • Metric: Choose a Google Analytics metric that you want to measure
  • Date: Choose the date range

Despite some limitations, cohort analysis can still provide some interesting results. For example, the below screenshots show how the duration times vary between mobile users and desktop and tablet users.

Cohort Analysis, Google Analytics, visit duration


I’m sure this new feature, whilst only basic at the moment, will provide valuable feedback about client campaigns.

3 Simple questions before you write your next blog post

It was only a couple of years ago that every slide presentation wielded by self-proclaimed social media gurus hailed “content is king”. This later changed to “curation is king”. Today the industry seems to have settled for ‘Ahh! Quick! Post it, post it!’ and most social media sites have been left with a desecrated mountain of link-baits, posts that are just lists of resources or observations, and rehashed content from years gone by.

The secret to standing out in all this noise is to plan:

  • Who do you want to speak to?
  • What you are going to say?
  • And why?

These possibly overly simplistic questions are the main reasons behind why blogs survive, and why others fail. For example, you could be a tourist board who is attempting to lure travellers to a destination; this is your audience. Therefore you may want to write about a tourist attraction, tips for living on a budget or how to spend 3 days in ‘blah’; this is your content. The final point is often the least followed, it is the reason why you are blogging. Do you want to generate higher social media numbers, sell more services or products, and refer traffic to an external link? This is your why. Remember it, this is your measure of success.

Going about this process means really getting to grips with what your blog is about. It may just be a personal project designed to give you satisfaction. If that is the case, enjoy! However, if you are a business or a determined individual, then your reasons may be far reaching. Think about where you would like your blog to be in a year’s time, do you want to make money from your blog, receive products to review, make yourself an internet celebrity? There are no boundaries, choose your direction and run. Blogging is a flexible platform that will allow you to succeed as a small online business, a large multinational or perky individual.

You just need to plan the purpose of your blog (which is another subject altogether) and develop a content publishing process that works for you.

Please stop talking about Big Data… sheesh!

‘Big data’ is becoming a phrase synonymous with the public relations (PR) industry. Just do a Google Search for “big data public relations” and 105,000,000 results are delivered. Ironically none of these articles are really explaining big data, but instead the service used to deliver those results is making use of big data.

As defined by Wikipedia (3rd August 2014),

“Big data usually includes data sets with sizes beyond the ability of commonly used software tools to capture, curate, manage, and process the data within a tolerable elapsed time. Big data sizes are a constantly moving target, as of 2012 ranging from a few dozen terabytes to many petabytes of data in a single data set.”

The fact is most PR agencies struggle to keep their media lists up-to-date, so industry discussions concerning big data seem rather farcical. Big data should only be used to reference certain situations, such as describing the total computational capacity of Google which apparently equates to 40 petaflops, which means 40 quadrillion operations per second. Such power doesn’t rely on ‘commonly used software’, whereas the service of PR typically does.

Unsurprisingly there are not reports on the actual amounts of data that PR agencies operate, but in my experience online programmes can range between 10 kilobytes – 50 gigabytes. Inevitably these numbers will increase as tools progress, the data we can collect advances and visualisations improve.

In my view PR operates in the arena of Small Data; defined by Wikipedia (3rd August 2014),

“Small data is data that is small enough size for human comprehension. About one quarter of the human brain is involved in visual processing, and the only way to comprehend Big data is to reduce the data into small, visually-appealing objects representing various aspects of large data sets or data “features” (such as histogram describing data projections and relationships, charts, scatter plots) that can be understood by humans. The term “big data” is about machines and “small data” is about people.”

The aim for PR agencies should be to develop solutions delivering humanly comprehensible information, by using commonly used tools (understandable for manual human interaction); computer scientists are not necessary.

Strikes me that the majority of those Google search results about big data were primarily placed to scare fees out of susceptible new business, rather than provide any real insight or analysis. Small data is beautiful, formidable, and offers outstanding value to PR. Don’t underestimate the value of these micro insights.



Evidence! There is life on Google+

Google+ is notorious for being one of the social networks which is the most difficult to measure. Google just doesn’t provide a measurement dashboard in the same way that Facebook does. Cynics would say this because Google don’t want us to know how small their network is in comparison to other social networks. Personally I think Google is simply developing their platform gradually and haven’t yet worked out a measurement dashboard yet. Although there is a small feature which will allow you to get basic Google Analytic stats for your page if it’s connected to a website.

Yesterday evening Google introduced a “total views” stat to Google+ pages and accounts. From what I can see I do not think these are unique views and it’s only a top-line number. It isn’t broken down individually for each post yet (which is frustrating) and counts from October 2012. What is important to note: it looks like Google+ have ditched their +1 number stat in favour of the total views metric.

This would make a lot of sense. Early last year many digital agencies reported that the Google+ pages they managed on behalf of clients had seen a dramatic reduction in +1s. Only last week did we at Keene detect another drop of +1s across the network. To remove this stat altogether is a step in the right direction for Google+ and I’m pleasantly thrilled to find out that I have 32,000+ views of my profile!

Michael White Google+ profile

Let’s hope that Google continues to enhance the measurement capabilities of Google+ and this new feature shows the social media world how big the network has become.

What does your Twitter network say about you?

Currently my profile description on Twitter reads as the following:

Digital Consultant for @keenecomms. Writer for Thought Symposium. Sell ideas, words and attention for a living. Fuelled by coffee and ale.

It’s written as an attempt to introduce myself to potential new Twitter followers, in the hope of converting them to avid fans. It sort of works! If you manage social media programmes for clients then you will know that it tends to be the descriptions on Twitter that helps us identify who Twitter users are. My description attempts to convey, in a sort of creative way, that I work in the PR industry and enjoy blogging.

For social media programmes Twitter descriptions are crucial because once you’ve identified the influencers in a client’s campaign, you want to know everything about them. Yet, most of the time agencies will only look at Twitter descriptions and ignore the depth of data behind each Twitter account. Assuming you’ve identified an influencer in the first place then you probably already know the following aspects about the account:

  • The user’s lifecycle of content, the type of content they share and how popular it is
  • The number of followers the user has and what that ratio is to following
  • Number of reactions the user receives online (RTs, shares and favourites)

The next step is to perform exploratory data analysis (EDA) to discover the main characteristics of an influencer’s Twitter account. In the case of the graphic below, I have visualised the latest 500 followers to my Twitter account (data captured in Sept 2013 but it’s still relevant).

As with my last blog post, the key to the image below is as follows:

  • Colours represent organic communities (AKA. value groups)
  • Lines indicate follow/followed relationships
  • Node size refers to influence (bigger the better), as calculate by the number of connections

michaeltwitternetworkfull copy

In an instant I can see from my personal network that the groups who follow me and engage with me the most are:

  • Travel blogging community (DARK PINK)
  • Media industry (GREEN)
  • Social media & SEO (BLUE)
  • Cheltenham community (LIGHT GREEN)
  • PR Students (DARK PURPLE)
  • Pirate Party UK (RED)

With a larger dataset and graphic, I’m sure other sub-communities of my Twitter account would also be discoverable. Even with capturing the last 500 followers of my account, those who know me will understand the communities I’ve identified in this post summarise my online activities well.