Data might be getting bigger, but its definition is getting smaller


This blog post was first published on

Data is getting big right? We all know that by now. We’ve got the message. It’s getting so big that it’s bursting at the corporate seams, bubbling up from the research, insight and analytics floor in to the boardroom where more often than not it’s deposited as a vast incoherent mass on the CEO’s, CMO’s, CSO’s or C-whatever-O’s floor, in the hope that someone spots a number that they like and doesn’t ask any difficult questions.

Possibly a bit cynical, and to be fair it’s widely acknowledged that data storytelling is of paramount importance for the modern-day analyst or researcher. One often unacknowledged challenge however, is educating the rest of organization on exactly what data is.

For some reason it seems that the practice of web analytics increasingly has a monopoly on the use of the word “data.” It’s a mistake I’ve seen time and again in organizations that I’ve worked with and for: more often than not, at a senior level, the word “data” is being used synonymously with analytics data. i.e. behavioural, metered, observed and tracked data. For some reason, the vast amounts of research, analysis and insight produced through “traditional” market research methods no longer seems to count.

Suddenly self-reported, attitudinal, ethnographic and surveyed research doesn’t fall within the definition of “data”. I’ve even heard a piece of survey-based market research being referred to as “qualitative research” … heaven knows what the results of an actual focus group from the UX lab would be referred to in this context. A “quick catch-up” with our audience maybe?

I’m no Big Data luddite by any means. Helping publishers harness the power of their own first party analytics data to tell compelling stories for brands and consumers alike is something I’ve had a lot of fun doing, I just think we need to remember not to throw the baby out with the bath water as we become mesmerized by the shiny and new promise of what big data can deliver.

Behavioural web analytics data can only ever tell us so much in isolation. It struggles to tell us about people’s attitudes, or what their sentiment is towards something or indeed what it is that’s driving their behavior on a deep routed cognitive basis. Big Data has been fantastic at raising the profile of insight teams, at placing data at the heart of the strategic decision making process and at democratizing data analysis, but it comes with its own warning too. As the recent facebook video metrics debacle highlighted, if behaviours can be gamed, then so too can behavioural metrics, and often such nuances are undetectable to the uninitiated.

At the same time, the traditional research and insight world cannot rest on its laurels, safely assured that only they have the technical rigour to bring data to life. The world of Big Data (however you choose to define it) and market research become greater than the sum of their parts when combined, and indeed some of the most interesting projects I’ve ever conducted have done just that.

Nothing is going to stop data getting bigger, guiding the strategic process with ever greater degree of accuracy and accountability. If you’re going to get big with your data however, make sure you stay big with how you think about its definition too.

Measuring a deeper impact…. And why ad measurement matters more than ever before

The online advertising ecosystem has been built around an unsustainable preoccupation with clicks. Why is brand impact measurement so important and how do you do it in a scaleable cost effective way?

Presented at the Research and Analysis of Media Client Day, October 2016

Content may be king… but it’s still answerable to someone


This opinion piece was first published on and in September 2016

I read a good opinion piece in Mediatel a few weeks back by Thinkbox’s Matt Hill in which he unpicked the hierarchy of ad exposure and provided a much-needed refresher course to those who seem to have forgotten the difference between ad delivery, viewability, and attention.

The distinction is an important one: you can trade on delivery and viewability, but currently trading on attention is rare (although the Guardian have recently joined the FT and Economist in doing just that by selling time based ads). The story of how attention based metrics need to feature more prominently in programmatic algorithms in order to do justice to premium contextual environments is a story for another today however. Rather, for now I want to unpick what this all means for the still relatively nascent branded content measurement market.

Just because a metric can be traded on, it shouldn’t be confused with a measure of impact and it seems that the content industry is all too eager to repeat the same mistakes of the display market in using binary behavioural measures to assess effectiveness… and most of the ills of the digital ad space at the moment, from fraud to ad blocking, can be traced back to that same mistake.

Consider a survey from eMarketer earlier this year which claims that over three-quarters of European marketers are using website visits as their primary measure of branded content effectiveness, yet less than half are measuring ROI (whether that be brand return or actual sales return presumably). Visits are important, of course they are: Every brand manager wants people to actually see their content right? But to use a simple behavioural measure as a metric of impact on your brand is arguably doing more injustice to the rich immersive nature of branded content (when it’s done well!) than clicks do to display ads. So why do we do this then? Simple: it’s because it’s often too hard to do anything else.

The measurement of branded content shouldn’t even stop at measuring brand impact. Quality of content is the very measure by which our readers judge us, and media owners need to acknowledge that what they produce has an impact on both their own and their commercial partners’ brands… and this impact needs to be measured too.

So what might a simple hierarchy of branded content metrics look like if you want to measure impact in its entirety then?

1) Content Delivery: There are all sorts of great platforms out there that allow you to interrogate your branded content and native analytics to within an inch of its life. Polar, Sharethrough, Simplereach, Parsely, Chartbeat and of course Google Analytics to name but a few. They can tell you how much, what and where your content has been viewed, shared and engaged with, with many now offering trading desk integration to allow you to optimise traffic driving assets as efficiently as possible. BUT, these are all just behavioural metrics.

They’re really important yes, but they only answer the question of how people are engaging with you content, not how the content is impacting on your brand. The metrics they track exist above the traditional branding funnel and we need to go deeper to understand longer term impact.

2) Ad Impact: Whether it’s brand or sales impact, measurement of branded content effectiveness can be hard. It’s hard to scale, it’s hard to find sample and it’s hard to assess long-term impact. But it’s not impossible, and if you’re clever about how you doing things – building reader panels, keeping a consistent methodology to build a  normative database, and using sales proxies such retail store footfall using mobile location data – you can really start to build rich datasets which exponentially unearth new insight in to how branded content and native works in general.

In their analysis of the IPA databank “The Long and Short of it,” Peter Field and Les Binet are quick to remind us that the optimum split between long term (brand oriented) ad spend and short term (response oriented) ad spend is 60:40. An immediately served branded content ad survey is by its very nature only going to measure short-term impact, but the brand metrics measured will be a better proxy for long-term impact than clicks or behavioural metrics will ever be.

3) Media Impact: This is the piece that so often gets overlooked, yet at the same time is so necessary for completing the circle in terms of content impact. Branded content doesn’t just have an impact on a sponsor’s brand, it impacts perceptions of a publisher’s own brand too. It has to be seen to be of sufficient quality otherwise the knock on effect will be a more mistrustful readership who themselves become a less attractive prospect to advertisers… and vice versa too of course!

Call it what you will – A Media Impact Framework, A Content Feedback Loop or a Theory of Change – either way tracking editorial inputs and outputs, metrics and attitudes, behaviours and impact with a bird’s eye view is vital if branded content and native advertising are to help the industry alleviate the pressures that the forces of commoditisation have brought to the bear on the market, while providing consumers with ad content that has real utility and which, fingers crossed they’ll be less inclined to block.

Content maybe be king, but just because it is, it needs to be held as accountable as everything else in the ad ecosystem if it’s to thrive.

Follow me on twitter @IGDataStories

R-Net Interview


This interview first appeared on in August 2016

Ian Gibbs is Head of Commercial Insight at Guardian News & Media, where he has taken numerous award-winning research and data initiatives to market. He joined the organisation as a Commercial Planner in 2007 after graduating from the University of Exeter with a BSc in Business Economics and several years as a Research Analyst/Manager at IDC and Dynamic Logic, Millward Brown’s Digital Practice. Ian will be joining the debate panel at the MRS Technology and Data Summit on 13 October.

I wish someone had told me at the beginning of my career that if you’re not enjoying your job you should just get out of there. You have the freedom to chop and change things a lot more at the start of your career.

I most admire anyone positive and enthusiastic who doesn’t just default to the negative mindset, which you can see so often in the workplace. Those sorts of people get places.

The best research project I have worked on during my career is “Audiences Not Platforms”. It has been the big one for me over the past few years. It’s a cross platform media planning and research tool that we developed and which had transformed how we (and many of our competitors!) sell ads.

The worst research project I have worked on during my career was probably the one at the start of my career where, with little experience, I had to produce a press release on the European tech market. Someone took a quote from my release that was in direct contradiction to something Bill Gates has said and tried to make out I was some sort of tech contrarian. As if a twenty-two year old me would ever have disagreed with Bill!

The most amazing or memorable experience when I was doing research is, coincidentally, also Bill Gates related – though at the other end of my career and in a completely different role. We’ve done some work assessing media impact for the Bill and Melinda Gates Foundation. It was fun and rewarding going out to Seattle to meet lots of other organisation from around the world doing the same thing.

The one story I always wanted to tell but never had a chance? Everyone always has the chance. Self-publishing an ebook these days is easy!

A research project I wish I had done is Channel 4’s “In VOD we trust” from last year. People are still talking about that.

If I wasn’t doing this, I would be helping my other half set up her massage therapy business in Ibiza.

The biggest challenge for our field in the next 10 years is watching the ad tech world encroach on the audience insight space. There’s a lot to be said of the reputation of more established research suppliers, however. We shouldn’t throw the baby out with the bath water when it comes to big data.

My advice for young researchers at the start of their career is be curious and be enthusiastic. Don’t be overly cynical. This is what will get you noticed.

Media Voices

This interview appeared on in August 2016


What does your role include as Head of Commercial Insight at the Guardian? How has this role changed over the past 6 years?

My role is all about harnessing the power of data of all shapes and sizes and transforming it in to compelling narratives for brands, agencies and advertiser partners. This includes building the tools and planning systems to get data-driven insight in to the hands of our commercial teams and partners as effectively as possible.

The work my team does touches all stages of the media planning cycle: from opening doors for the sales team through thought leadership research, to “traditional” media planning, to tracking and measuring campaign performance.

The market has transformed at such pace over the past six years that it’s hard to know where to start when it comes to thinking about what has changed! I think overall though, there are probably four key trends I’ve seen:

1. A real shift towards measuring true campaign effectiveness.Measuring ad impact is nothing new, but measuring it using the correct metrics – particular in the digital space – is. Our industry has built an entire advertising ecosystem upon acquiring clicks: clicks on content and clicks on ads. Such a binary measure might be fine for a DR campaign, but it does a huge disservice to the effectiveness of a brand or content campaign and that’s what we look to address with our research.

2. A focus on first party data. Everyone has access to the same off-the-shelf planning and research tools these days. They’re simply a hygiene factor now and are no way to really differentiate your audience from a competitors. First party data absolutely is though and we’ve been using it more and more to inspire creativity and campaign plans amongst advertisers.

3. Cross platform campaign planning: Print and digital planning and buying in agencies and publishers used to exist in splendid isolation from each other which was an incredibly inefficient way of doing things when you think about it. We’ve built tools like our award winning “Audiences Not Platforms” planning system to allow agencies to plan in a truly audience centric multi-platform way. The third iteration of ANP is being released later this year.

4. Internal truths vs external narratives: I think a lot of organisations are sometimes guilty of confusing their externally facing sales bluster with their own internal truths. Keeping the two separate is really important and over the past half-decade, a culture of using strategic insight in the commercial decision making process has noticeably evolved.

How do you use first party data to cooperate with your advertisers?

Through a system that we call Audience Explorer we are creating an advertiser audience-centric, rather than Guardian-audience centric view of our data. The Guardian reaches over half the UK online adult population each month so arguably has more information on the nation’s quality content consumption habits than anyone else.

We use this data to inspire branded content. For one educational establishment about to launch a forensic science degree, for example, we were able to see that their audience was really engaged with crosswords and “Scandinavian Noir” content (all very “Guardian!”). What better way to raise awareness of their new degree than through an online interactive crime scene perhaps—sating the desire to solve both puzzles and crimes simultaneously?

And of course our data can be used to make planning decisions on the delivery of display campaigns. We have data scientists using our data to build audience segments that are delivering far more efficient results than those that can be bought from third parties. This is the sort of stuff that agencies and brands are getting really excited about.

How is the data then filtered through the rest of commercial team? For example, do you have a feedback system?

We attempt to democratise data as much as possible at the Guardian. Everyone in the business can access our live Editorial analytics dashboard “Ophan” – and this is how we aspire to deliver all insight. There are also plenty of cross-functional and cross-departmental working groups or huddles that have been brought together to solve specific business issues and it’s proven to be quite an inspiring way to work.

What measurement tools do you use to evaluate return on ad spend?

It’s a mixed bag. For years we’ve partnered with the Swedish ad effectiveness research company RAM to assess the brand impact of campaigns across all platforms: web, print, mobile and app. We’ve got to the point now where we’ve built up a really robust normative dataset that we can use to unearth new insight. For example, evaluating the value of context in digital display. You can see the results here.

Over the years we’ve worked with Aimia and Kantar to prove ROI and there’s now a noticeable shift towards proving that advertising prompts action in our audience. We’ve commissioned YouGov to track the online and offline impact of one of our largest branded content deals from Guardian Labs this year and are also talking to a number of vendors about using geolocation information to measure store visitation after campaign exposure.

What is the purpose of Facebook advertising for publishers if they are already publishing content on Facebook?

How a publisher’s content appears in user’s news feeds is always rather at the mercy of the Facebook algorithm, so I suppose with advertising you have more certainty! That’s not really the point though. Facebook is a vital source of traffic for many publishers and initiatives like Instant Articles just make that content even more accessible in a mobile first world.

What are your thoughts on Facebook paying media companies/celebrities to create Facebook Live videos?

Facebook needs content just like everyone else. Who better to produce it than the experts? With fewer people sharing personal updates on Facebook these days, it makes sense to constantly be looking for new ways to keep audiences engaged. Facebook obviously benefit enormously from the engagement that content brings with it but I suspect in the long run a more sophisticated model of reimbursement will need to be found.

Online branding and offline ROI: A data driven boost for premium?


This post first appeared on the

It’s been quite the couple of weeks for premium content… well, from an insight point of view, that is. With Comscore and Newsworks both releasing studies that prove whether it be in digital or print, there are still deep rewards for brands who’s ads appear in and around quality content environments, the premium ad market has received a much needed shot in the arm.

And it is a shot in the arm that’s needed. We live in an age of mass distraction, where consumers are paying less attention to traditional media channels and media planners are paying more attention to methods of campaign planning that enable more effective ways of reaching audiences at scale irrespective of context (warning: this opinion piece isn’t going to turn in to some sort of anti-programmatic diatribe, but will instead question the prevalence of automated audience trading used as an exclusive means-to-an-end irrespective of campaign goals: The concept of programmatic is often unfairly tarnished with this brush.)

This week Dominic Mills argued in his Mediatel column that the effectiveness industry is having an existential crisis, and it’s hard to disagree. The argument is a nuanced one for print and online however. Where print can often be damned by not being measurable enough (or at least perceived as not being measurable enough – especially for those with shoe string budgets), online is damned by being too measurable. It’s therefore been great to see separate pieces from Comscore and Newsworks address these issues from very different angles.

First off, Comscore’s study “The Halo Effect.”

In an independent study commissioned in in the US, ComScore has proven that advertising on quality content sites (defined as members of the trade body Digital Context Next) delivers a 67 percent uplift in brand metrics vs advertising running on non-premium sires.

As Peter Field and Les Binet are quick to remind us, a focus on short term measures will only yield short term results, and focusing on short terms measures is exactly what pretty much the entire online advertising industry is doing at the moment. A perpetual preoccupation with the click whereby marketers are assessing performance simply in terms of what they can measure, rather than what they should measure has created an ecosystem that is not only having a negative knock-on effect for brands and publishers through increased levels of ad blocking and fraud, but on content too through the proliferation of low quality content which has been created with the primary purpose of generating a click rather than creating genuine utility for a reader (next time you get to the bottom of an article and you are greeted with an invitation to “Read more from the best of the web” ask yourself how closely headlines like “Get rich in just thirty days” really match this description).

It’s therefore encouraging to see a piece of work like Comscore’s that focuses on the branding effect of digital advertising in quality contextual environments get such traction. A focus on binary measures such as clicks yields only short term results and it’s only with premium that the long term effects of brand advertising can be more effectively realised.

Second up: Newswork’s study “Newsbrand Effectiveness: The Evidence.”

Meanwhile, in a watershed study, Newsworks have mined five years’ worth of econometric data to reveal that advertising in news print yields a ROI multiplier of three times its value. Again, this study points to real long term rewards for brands engaging with premium content environments. Offline media is undoubtedly harder to measure than online media (although arguably both are as hard as each other to measure correctly), but it’s great to see that Newsworks haven’t shirked the task and that by mining the IPA’s effectiveness data bank have done so with the objectivity so often lacking from research that comes directly from industry bodies.

The study also points to the fact that print advertising for brands in the finance and services categories offers a significant multiplier effect to TV and radio. And this is what I like about this study: It doesn’t look to pitch news brands in a “My ROI is bigger than yours” battle with other media, but rather seeks to place news brands in the context of other media on the plan, arguing for a return to 2013 spend levels in print, but not asking for more.

Print is pretty much the only media that we ask young millennial agency planners to plan with little understanding of what it actually takes to be a consumer that media. It’s certainly less of a challenge than that faced by TV , radio or outdoor for example and yet with the one of the few silver linings to the Brexit cloud manifesting itself as a boost to national newspaper circulations in June, it is clear that news print is still a medium that offers tremendous value to consumers and brands alike.

To sum up…

Quality context isn’t dead. Far from it: in a world where programmatic has become a byword for automated audience trading that lacks the subtlety to sift quality content from the rest of the pile; and in a world where news print can still offer the impactful and trusted environments that brands crave, context matters more than ever before.

With June’s Global Marketing Index reporting that European marketing budgets are set to tumble post Brexit, every penny of marketing spend needs to count more than ever before. Never has there been greater opportunity (and more evidence) for advertisers to stand out from the crowd by going premium both on and offline.

Big data, smart data, fast data…. creative data?


This post has appeared on, and

“You might be making mistakes on the margin of error, but I believe the speed of which you get data and make decisions is more important than the accuracy of the answer you get.”

So said Jorn Socquet, AB InBev’s VP of US marketing, in an interview with Marketing Week last month in which he endorsed the use of “fast data” analytics in business decision making. It’s a comment that doubtless sent a cold shiver down the spines of research die-hards. Margins of error, statistical significance tests, sampling methodologies: they exist for a reason right? Do we really want to discard years of carefully crafted market research and analysis theory—theory that has been developed explicitly to provide us with an as accurate as possible estimation of the truth—just for the sake of some quick and dirty real-time insight?

Perhaps this comes as no surprise? Much comment has been made of late around how we now live in a post-truth world. Facts just simply don’t seem to matter in political discourse anymore, as evidenced by the result of the Brexit referendum. So why should they matter elsewhere?

Once you consider the fact that according to KPMG’s recently released Global CEO Outlook Survey 86% of global CEOs say they lack time to think strategically about forces of disruption and innovation, but 84% also say that they are concerned about the quality of data they have to base decisions on, you have to ask what it is that they really want: fast inaccurate data, or slow high quality data?

This is a glib response of course, and I do actually think Jorn Socquet makes a good point. As with everything, it’s all about context. If you’re making important strategic decisions about the long-term future of your business or brand, take your time with the analysis and stay robust. But if you’re simply trying to inspire some creative thinking, go fast. And that’s the crucial point here: data doesn’t just have to point to one single truth, it can simply perform a function in inspiring creative thinking and ideas. In other words, it could be used as one of many means to a creative end, rather than the end point itself.

This is just what we’ve been doing at the Guardian recently: harnessing the power of our first-party data to inspire creative content ideas in our advertiser clients. By placing a DMP-generated pixel tag on client digital properties, we can now create an advertiser-centric view of our data rather than a Guardian-centric view. In other words, we can tell our clients about their audience’s quality content consumption habits to find out what makes them tick. For one educational establishment about to launch a forensic science degree, for example, we were able to see that their audience was really engaged with crosswords and “Scandinavian Noir” content (very “Guardian”, I hear you say!). What better way to raise awareness of their new degree than through an online interactive crime scene perhaps—sating the desire to solve both puzzles and crimes simultaneously? We’re not suggesting that they stake the future of their business on this one piece of insight, but this kind of data can be provided fast, and it can unearth potentially exciting creative ideas for brands.

I read of a data sceptic once questioning what piece of data drove the decision to use a drumming gorilla to advertise a chocolate brand, but I don’t think that this is as farfetched as it sounds. Data can be a lens through which we make creative decisions: not the be all and end all, but an interesting reference point nonetheless. I see no reason why data can’t inspire such creative thinking. And of course as Cannes Lions are always quick to remind us, creativity matters: it drives real business results.

The final point I’d make is that perhaps fast insight and robust insight aren’t going to be as mutually exclusive as they sound in the future. I read with interest last week the news from Savitz Consulting that through the identification of inliers, research sample sizes can now be reduced without forfeiting robustness: reducing research costs by up to 33%. Music to the ears of many a research budget holder no doubt.

At the end of the day you just need to be clear on your objectives before you set out. If you’re a die-hard market researcher or data analyst, don’t get too caught up in the detail if the aim is just to inspire some creative thinking. But on the flip side, if you’re a senior business decisions maker or strategist, don’t put undue pressure on your analytics people to deliver fast data that might compromise credibility when you’re staking important strategic decisions on the outcome.

Get fast with your data, but not loose!