It was a packed day so let’s just get stuck into what you’re all here for – the round-up!
Navigating the Biddable Landscape Jon Myers (@JonDMyers)
Jon Myers kicked off the event with Marin’s key finding about biddable media: thinking more about audience is the key to future success. We’ve been applying cost data to revenue for a while now, but we’re just starting to applying the context (did it snow, was there a print campaign, etc) to our paid efforts to really understand what’s happening with our audience.
He then took us on a trip down memory lane, talking us through where we’ve come from with the first online ad way back in 1993 through to the current situation: Paid search is morphing with new ad types like PLAs. Facebook is already very important and the space is getting bigger. The display world is really complex.
Looking ahead to the future, Jon came back to the idea of audience with the bold statement:
Audiences are the new keywords.
We used to know nothing about our audience, but now we’re starting to connect keywords to audiences. This is a positive step, helping us not only to find our target market more easily and effectively, but also to provide ads that more effectively meet their needs.
Jon left us with some cool ideas about the possibilities for biddable media – will we see biddable TV? Sky’s AdSmart could be the start of this. And why not radio? Or Google glass? Or outdoor advertising – there are already interactive billboards that use data from the phones of people standing nearby to display ads – it’s logical to think that this could move to a biddable model.
In short, the future’s bright; the future’s paid 🙂
Tentative Steps into Twitter Advertising by Heather Robinson (@_skittish)
Heather Robinson was up next to talk us through the world of Twitter ads and add some Yorkshire charm to the day.
She started out with the basics: there are three ad types:
- Promoted tweets, which are good for increasing reach and engagement. These appear in the newsfeed so have the most visibility.
- Promoted accounts, which are good for increasing followers. At the moment this is desktop only, limiting their appeal, but they are soon to launch on mobile. Happy days.
- Promoted trends, which are good for Maximum Exposure. Heather quoted mashable, which last year reported that these could cost $200k per day – in the US at least. Although the point stands that this is not an affordable option (to say the least!)
Response (Several engagement types… Engagement rate) is higher on twitter than Facebook.
Talking about the practicalities of setting up campaigns, Heather made it very clear that you can’t just put Google campaigns on twitter! The entire advertising ethos is different so your approach should be different too. While it’s more similar to Facebook, with the focus on audience not keywords, there’s still a very different way of reporting – Twitter talks about Engagement Rate (retweets, favourites, replies, clicks) and Facebook reports on CTR.
Twitter also currently has a significantly smaller mobile ad share than Facebook – although this is likely to increase due to promoted accounts shifting to mobile soon.
Heather suggests that instead of thinking about what you’ve done on other ad platforms, you should think about your goals and match your choice of ad type accordingly.
Now it was time to get into some of the mechanics of Twitter ads – namely, matching options. You can choose to match ads based on:
- Keywords in tweets.
- Geography (Which can be narrowed down to region or city level.)
- Targeting by gender (Although Twitter doesn’t actually know this about you, they make an educated guess that they claim is 90% accurate.)
- Similar to existing followers (Either your own followers or for someone else e.g. a celebrity – if you knew who your market would like.)
And Heather finished off with some general tips that she’s found work well:
- Questions typically get a better response.
- Use images – especially now they can be seen in stream.
- Early adopters will benefit – make the most of the relatively low competition levels at the moment for some big wins.
- They saw the most favourites + retweets for “top tips” tweets or links to top tips blog posts.
- Lead generation cards (tweets) – are great for encouraging downloads and promoting offers and competitions.
- Experiment to find the best audience and the tweets that drive the most engagement.
- There’s conversion tracking that allows you to get more data on your ad performance – use it!
- Use the Opportunities Report! This can show organic tweets that are already generating engagement that can be added to a campaign to increase reach.
- Use the Followers Report – to find out interests, location, gender and who they follow so you can refine your targeting.
- It’s really important to target the right people! People don’t necessarily like seeing ads on Twitter – and they will tell you – so don’t piss them off by targeting a generic audience. Be specific and make sure that you have great content so your ad is worth a tweeter’s time.
How and Why to Become a Power Editor Power User by Ben Harper (@benharper87)
- Time saving – it allows you to get more done in a short space of time, thanks to its bulk editor roots.
- New changes are rolled out into power editor first – the gap before it launches to the main interface may not be long, but any edge helps! For example, Facebook now allows us to target users by what they studied, this appeared in power editor first.
- You can manually edit Optimised CPM (OCPM) bids to make them more specific to your goals.
- This makes it easier to optimise and set everything up for proper testing.
- The big one – this is where you find advanced audiences.
- Custom audiences upload your CRM data to build a custom audience of Facebook users.
- Alternatively try a similar audience, which uses the demographics of your custom audience to build a list of users who share your customers’ characteristics. Ben wasn’t keen on this because it gives you less control over targeting, but it can be good for fan acquisition.
- Partner categories pull loyalty card data into Facebook, so you can target people by buying behaviour. Not available here yet, but due in 2014 apparently.
- To truly understand the best placement for conversions you need to split things out, which is easy in editor.
- You can use it for data extraction. When you input targeting, it tells you the potential audience so you can overlay targeting for insight e.g. discover the age groups of Nike fans.
The Power Editor is only available in Chrome. The best thing to do is get started with it – test and learn!
Black Hat PPC by Nick Christian (@nickchr1stian)
Strictly speaking, Nick wasn’t here to talk black hat PPC – not in the strict sense of cloaking and all that scary gubbins. Instead he talked about down prioritising best practices and focusing on marketing.
We think that PPC and marketing are different – because we talk in more sanitised terms (creative instead of ads, conversions not sales) and because we can (allegedly) track clicks to sales.
It’s undeniable that no one genuinely knows how much a TV advert genuinely influenced sales. We think search is better, but all we know is that a sale happened after a click.
Where we are genuinely different is that we can be flexible, responsive and proactive in a way no other advertising can. But often we don’t, because we make assumptions, and Nick was out to challenge those assumptions:
- There’s no such thing as irrelevant traffic.
- Everything had a conversion rate.
- There’s nothing you can’t buy if the price is low enough.
— Rory@Rocket (@RoryAtRocket) January 22, 2014
We should care way more about CPC than we do. Low conversion rates aren’t an issue – as long as you haven’t spent a lot on those clicks.
Nick works for lastminute.com and based on his work for them he said that:
- We need to do massive keyword expansion.
- We have to expand geotargeting and language and device targeting.
- We should spend less time on adding negatives.
- We shouldn’t spend as long on copy.
- We should focus less on the website – we should care, but search is demand driven so we don’t necessarily have control.
- We have to embrace Enhanced Campaigns. They’re better than when dappy got kicked in the face by a horse.
- We need to start with the assumption that we can bid on everything – ask yourself “why not?”, not “why?”
- We should aim for as much traffic as possible.
When they used this approach they found that they had to start doing more analysis. They began using dimensions to draw out more insights than they had done previously.
Not everything went right when they used this approach – three keywords blew up. If you drastically change your approach, things will go wrong, but you have to learn from your mistakes. And ultimately, things weren’t all bad – there was a drop in CPC, so they were getting more traffic for same spend.
Nick wrapped up with a simple idea: There’s no such thing as bad traffic just more or less interested traffic. Your site doesn’t care where traffic came from and neither should you.
Was eBay Right about Paid Search? by Richard Fergie (@RichardFergie)
In March 2013, eBay released a study that had paid search marketers across the world weeping – they had concluded that paid search was ineffective. For eBay, brand queries did not make any money. Non-brand made a tiny amount of money, but only for new users. Richard Fergie was here to remind us that it doesn’t matter what’s right for eBay, it matters what’s right for you.
In his view, you have a PPC profitability curve, and you’ve got two types of task as a result:
— Richard Fergie (@RichardFergie) January 22, 2014
- You can try to change your position on curve: i.e. do you care more about spending to get as much traffic as possible and just not losing money or do you want to make as much money as possible for a limited spend?
- You can try to change the curve: and generate loads of profit for loads of traffic.
This ability to position ourselves on a curve in this way is a unique position that doesn’t apply to other channels. But, if we don’t know the value of PPC we don’t know where we are on the curve. At the moment we use attribution modeling to try to tell us, but how do we know which model is right? All the different choices try to answer the question of where we are on the curve, but Richard isn’t satisfied by the answer.
So what other techniques can we use to get an answer? Try the eBay approach and test out switching things off to discover the impact. You can approach this in a few ways:
- The nuclear option: switch off all the things! Pros: A quick way to establish impact. Cons: You lose your job (if it proves had no impact you render yourself irrelevant, if it proves PPC has a huge impact you just killed their sales). And this doesn’t allow for any understanding of seasonality, which has an undeniable impact on every account. In short, it’s not a practical answer.
- Nuke parts of the account. If you choose intelligently (this is very tricky) you can gain great insights by carefully selecting small sections to pause.
How can you intelligently choose the sections to pause?
Just pausing a campaign or keywords doesn’t work – your ads will start to match other keywords elsewhere in the account and it just makes you less efficient overall. This method only works with super negatives in place to avoid this scenario, but it would be extremely time consuming to do this effectively. Having said it’s a poor approach, Richard suggested that this could be a good approach to discover the brand/non-brand split in your account.
Use paid/organic reports if you’re unsure about the split between the two, although there are concerns about the accuracy of organic data provided by Google Webmaster Tools and why Google doesn’t show ads for all queries.
You can use a method called time slicing and pause PPC for short periods of time. You can learn the value by not running ads, and comparing performance to when you are.
For example one week you might pause ads between 2pm and 3pm, then reactivate for the following hour. If you want to double check whatever results this gave you flip it the following week and run ads from 2pm to 3pm and pause them from 3pm to 4pm. It’s important to choose homogeneous times for this – don’t use a high traffic times or ones that typically have very different results from each other. This approach is unlikely to be effective for accounts with a long lead cycle, you’d need to pause your ads for longer to see data, which takes us dangerously close to the nuclear option that we want to avoid.
EBay used this time slicing approach based on geographic areas. They divided America into Nielsen regions, and split those regions into three. Activity was paused in just one of the three areas, and then they observed what was happening in the paused test groups compared to the 2 partner groups in the region. Effectively they were running an AAB test.
- Ignore London. It’s too weird a town – he was actually referring to population difference and IP issues, although I think the statement stands generally.
- Normalise for population size and group by size – if you skip the normalisation step, you’ll end up with strange groupings of big cities that perhaps have no other similarities.
- Alternatively group by percentage change, e.g. these towns follow the same seasonal trends.
- Whatever method of grouping you choose, groups things together in a dendrogram and check line graphs of performance to ensure they belong together.
- Keep groups bigger for lower risk. E.g. if you have groups of 10 only 1 will be paused compared to groups of three (the eBay approach), which could be a significant volume of traffic gone.
Once you’ve got your grouping, it has to be all or nothing to avoid cannibalisation. If you can’t commit, it’s not sensible to run the tests, but there’s real value in finding out the value. You can use this to test the validity of your attribution modelling and it might be easier than arguing.
Ultimately, you need an impact on revenue to get results but it doesn’t have to be extensive, you can limit the negative impact while discovering real insight.
AdWords Quality Score Types and How to Boss Them by Tara Dee West (@Koozai_Tara)
Optimising for Quality Score (QS) is more valuable than ever. Revenue is paramount, but QS can help to bring CPCs down, which does have a positive impact on revenue.
There are several different types of Quality Score:
Some facts about keyword level QS:
- It’s based on the historical CTRs of your account, if there’s no history in your account then Google will use the industry average.
- There is an impression threshold that has to be reached before keywords earn their own QS.
- It’s calculated on exact match search queries – i.e. when the search query is exactly the same as the keyword.
- Normalised on position, ads in position 1 will have a better CTR than ads in position 8, this limits the impact of the position on QS.
So how should you optimise for QS:
- You should start with great structure – ideally with just one keyword per as group, although that’s not always practical.
- Boost impression share with Modified Broad Match to reach the impression threshold faster.
- Use exact match keywords.
- Normalisation isn’t accurate – so if you can afford to bid for a top position, that does help.
- Google will apply historical QS to a keyword that has previously performed poorly in an account – even if you deleted it before. You may be better to leave it paused.
Optimising for ad QS:
- Ad extensions should be considered as part of the ad, make sure that you always use them, and ideally set them at the ad group level.
- Split test regularly to ensure that you’re improving your CTR.
- Using keyword variations in the ad is important to help with exact search query matching.
- Display URL is important factor in deciding ad level QS – so make the most of it!
- Dynamic Keyword Insertion can be really helpful.
The important thing to remember at the account level is that bad shit will bring everything down – so working on low QS keywords does matter. Sadly though, it takes a long time to make account level improvements! Here are some things that you can do to help:
- Optimise for location – Tara tells us that even where the account has been set up matters when it comes to location.
- For new accounts, set up brand first so you get the benefit of the good positions, relevancy and CTR.
- Regularly create new campaigns and keywords.
Landing page level
This is currently a very small part of the puzzle but Tara reckons it’ll become more important.
Landing page QS is mostly about relevance, which actually relates to whether you can show at all for a given keyword. E.g. your make up reviews page is simply not eligible to show up for London Zoo keywords because it isn’t relevant. As long as you target keywords that relate to your page, you don’t have to worry about this. If you’re unsure if your page passes the test, get an outsider to tell you what it’s all about. If they can’t tell you, you need to think again about the page!
There are still things that you need to think about:
- Bounce rate and time on site matters (time has been introduced to counteract the impact of high bounce rates for PPC-specific pages).
- Speed is really important. Benchmarked by industry.
- Unique content (don’t rely on blocking with robots.text)
Device QS is affected by many of the same factors as the others – CTR, landing page, etc, but mobile in particular has a few new ones:
- Location matters more – so make sure that you use extensions and bid adjustments where appropriate.
- You need a good, speedy site and a great user journey. If your site isn’t up to scratch, don’t be afraid of using the -100% bid modifier while you sort it out.
- Create mobile ad copy and use mobile extensions to create the best experience for users.
- Segment when you are optimising your site.
If you are working on CPC bidding, the same rules apply as above. If you use CPM bidding your QS is based totally on landing page quality. While this may sound scary, remember that landing page QS is mostly about relevance. So essentially this means you need to work on targeting the right audience and placements to ensure that your ads are eligible in the auction.
We only ever have access to keyword level QS – and we don’t get to see historical data. So if you want to keep track of what’s going on with your account, make friends with AdWords Scripts! Tara recommended using FreeAdWordsScripts.com to find scripts that can track QS for you.
While scripts can help you keep track of QS they can’t tell you where you should focus your attention. A keyword with a low quality score but only a handful of impressions won’t have as much impact on your campaign as a whole as a higher traffic keyword with only a slightly higher QS. In this case, try using weighted QS (e.g. QS x impressions) to prioritise the keywords you need to work on to improve your account, so you know that you are optimising the keywords that have the most impact.
Finally, 6 is new 7. Since the change in reporting Quality Scores in 2013, there has been a shift in the figures, so where you previously had a 7, now it’s a 6. It’s harder to move from 6 to 7 than 3 to 4 but the cost savings are now worthwhile.
Online Shopping by Sarah Conway (@dothesethings)
Sarah started off by showing the context of the PLA landscape. In 2013, Google Shopping users saw slow decline in free clicks from February to August, when free listings ended. The use of PLAs has increased massively over the year – and importantly the revenue that they generated increased by significantly more than spend.
Looking ahead to the future, Sarah shared the following thoughts:
- There will be an expandable module of 16 ad spots available.
- Bing PLAs show promise – they’re meant to open up to general users in Q1 or early Q2 this year.
- Almost a no brainer: they’re becoming an increasingly huge part of search.
The Holy Trinity: Feed – Data – Structure – Sarah tells us that all matter for great PLA performance!
The basics will often be covered automatically, but you probably need to tweak categorisation and additional attributes. Go granular with your Google Product Categories to get off to a good start. Mimic the categorisation of your website for your product types.
You can try to get everything perfect before you put it in, but it may be better just to get it live! The Google Merchant Centre information will tell you if there are any problems – even if the interface is a little clunky. And ultimately, just adding more products will increase your clicks.
Start by targeting all products. Once you’ve got some data through you can break out the hero products and stagger above all product targets with bids. (Think of “all products as a rogue broad match – it’s expensive and bad for QS so don’t let all products win the auction with high bids.)
PLAs are like a bucket full of holes – as soon as you cut off one unprofitable keyword, another will come up behind to take its place. Remember to mine search query reports and add to your negatives to keep your ROI high.
Dig deeper into your data by segmenting performance – look at the performance across devices, check day and time to understand your user behaviour – do you need to split test or segment further? Should there be bid adjustments? These are all ways that you could win with your PLAs.
A few top tips from Sarah:
- Throw some budget at it – trying raise bids initially to get data, then bring spend back down for good ROI.
- Use adwords_labels – they really are useful.
- Break out best sellers and target them with higher bids.
- If you’re not using PLAs or only use them in the UK when you sell globally – get on it – it will never be a better time.
- Try using the accelerate to show more frequently.
- Check quality of your feed and try to break out as many targets as you can or as many as make sense for your brand.
Every Picture Tells a Story by Jim Banks (@jimbanks)
Jim Banks started off on a strange – if amazing – note. He told us about his youth in Hong Kong, modelling himself on this guy:
… and going to black tie dos with his mates dressed as The Village People. Yes you did read that correctly. And just to make the trip down memory lane complete, he even shared his party trick for standing out (apparently you blend right in dressed as the hairy one from the village people) and he popped a party popper into his mouth.
Jim wasn’t going senile (probably) – he had a great point with all of this. Just because you can do something that’ll make you stand out, doesn’t mean you should. While his questionable fashion choices did get him attention they weren’t necessarily wise or appropriate. After this eye-opening start, he then went into the meat of his talk:
Why Graphical Ads Are Great:
- They don’t look like ads.
- They tell a story.
- They resonate.
- They’re shocking.
- They’re different.
Why Graphical Ads Are Crap:
- They focus more on advertiser than customer.
- They bore people by caring about themselves – your ad needs to help people with what they’re trying to do.
- They get blocked. Use white space around ad to make it a non standard size to avoid this.
In Jim’s opinion, there’s no such thing as ad blindness, just bad banners.
He talked us through some ways that you can improve your ads – see the slides for full details – but these two jumped out at me:
- Make your ads any shape you like with white space – such a simple trick but definitely effective.
- Quite often you can use an angled picture to tilt a photo to what you want people to look at – you can even edit eyes to look at your message.
A few key takeaways from Jim’s talk:
- There’s no right or wrong you just need to try.
- Think about how to stand out from your competition.
- There’s a long way to go on attribution but it’s important to try.
- You can’t necessarily do everything on every network (e.g. Google has rules about white space around ads).
- Use fiverr.com or 4mads.com (template to create great ads cheaply these do include a watermark) to get ads made.
- WhatRunsWhere.com is great for competitive analysis (but pricey!) – who is hosting the ads for your competitors.
View Through Conversions – Not Worth the Paper They’re Printed on by Nate Wood (@natewood)
Nate Woods was up next talking about a potentially controversial topic: View Through Conversions (VTCs) don’t get enough air time – whether people love them or hate them. That may be because it is a complicated topic!
You can’t escape from advertising. It’s as old as civilisation, with ads seen in ancient Egyptian scrolls. And yet, nobody likes ads, everyone says they skip through them on TV. But you don’t hate them – how could you hate something that you are constantly surrounded by? Advertising is increasingly innovative and acceptance of it is growing.
We gain budget by talking about accountability – as an industry we fixated on the last click because we didn’t have technology to go beyond that. It’s dated.
Everybody believes in attribution and VTCs are just an extension of this.
What is a VTC?
Seeing an ad, but not clicking so the performance can’t be seen in analytics programmes, and yet that doesn’t escape the fact that VTCs are just part of the path to sale.
And here’s where Nate started to talk psychology:
— Rachel Misra (@RachelMisra) January 21, 2014
We’ve evolved to absorb ads subconsciously. There’s a cognitive bias known as the familiarity heuristic, which shows that repeated exposure to something and therefore increased familiarity leads to more favourable reactions. A study tested this behaviour on readers by placing banner ads out of the focal area of a five page article and discovered that even without consciously seeing ads people were better able to identify a brand.
Nate highly recommended that we all read The Silent Click: Building Brands Online. This discovered that the longer that a user has been exposed to a brand for, the more likely they were to search for it, and they evenly engaged more positively with the websites when they landed on them by staying on longer and spending more.
This clearly suggests that merely displaying ads can be helpful and that VTCs is a good way to test this effect. But forecasting is impossible – a viral video is often indistinguishable from a video that didn’t go viral and it’s the same with predicting the power of VTCs.
If you have some understanding of the nature of VTCs you can make better predictions of how they’ll work, but even then it’s hard. It’s hard to measure uplift exactly although you can often see correlation. And while Nate agrees that the doesn’t indicate causation, it’s hard to see curves going up and not make a connection.
His advice on finding the value of VTCs – dig deep. He looked at everything – from time on site to sales to discover their value, and the only correlation he could see was between impressions and CTR.
But as with everything – test VTCs in your platform, don’t just work on someone else’s findings. If you run a non-brand public service ad and it has n0 VTCs you know any other VTCs are likely to be accurate. If you can see that they are likely to be accurate, you then still need to interrogate them to discover if they really have value for you. And this very much depends on company and campaigns as to whether it’s worthwhile. E.g. did running the display campaign increase brand clicks and CTR enough to justify the spend?
A few final tips from John on VTCs:
- Don’t just assume it’s not working, things take time.
- Don’t launch other campaigns (e.g. email) at the same time that could affect your findings.
- Don’t look so hard that you convince yourself something exists that doesn’t.
- Always separate out VTCs and last click.
- Get buy in from all departments – whenever you’re assuming value of VTCs you’re taking credit from someone else and they need to be ok with that!
Performance Display Best Practices by John Were (@John_Were)
John Were started his presentation by talking about the background of Xelsion – and how ad buying and indeed the entire display landscape has changed dramatically since they started out. He went on to share the best practices that they have learnt over the years.
In John’s eyes, planning is really important. It’s worth thinking about the journey up front – just as you would on an actual trip, you need to think about where you’re going before you start so you know how to get there. Anything on digital is complicated and will take a long time, but the clearer you can be on your goals, the more likely you are to reach them. There’s a whole ton of information that needs to go into this process – ideally you should document it all with an initial brief (ideally).
Key takeaway: Think about goals and what success looks like – before day one.
Now you can think about how to execute the campaign. In particular you need to get your tagging set up right! This can be painful – although tag management solution had made this easier – ultimately, it’s worth the pain because it’s vital to show results! You need to think about any and everything that will show success and how to measure this.
In the planning phase think about the structure of your campaigns and avoid a complex top line structure (e.g. lots and lots of campaigns with little information in each). This will be difficult to implement and changes to settings and tagging will be time consuming to implement. Instead aim for a traditional pyramid with a small number of campaigns.
Ultimately there will be two things that you’ll use display for : prospecting for new users or retargeting.
If you’re prospecting consider using the following targeting tactics:
- Audience (demographics)
- Contextual (using page content as a proxy for audience)
- Reach (high level groupings to get as much exposure as possible)
If you’re retargeting don’t treat your audience as all the same! There are lots of different ways that you can segment users, e.g. recency of visits to website, as the value of segments differs.
- Separate out bids depending on who works well. In the case of an online pharmacy he’s worked with, really recent traffic is expensive, gets gradually less until it hits the 30 day mark and then it picks up again.
- These segments potentially need different ads to suit their needs.
Tends to recommend CPM bidding when working on display, although CPC and CPA are available too. His reasoning is that all bids are translated to CPM, so they can be compared against each other so this offers the most transparency. He also showed also how it can be cheaper and talked about the potential for fraud with these systems as they can be fooled into thinking that there have been more clicks and sales than actually took place.
A few tips on ad creative:
- Simple is a lot better than some of the rubbish created.
- Engaging is better – how can you encourage interaction with your ad?
- Think hard about the final frame of your ad – simple and branded is best.
The plan should be communicated clearly and signed off by the client – once this is done, you’re ready to build your campaigns! Testing is crucial – you’ve got to make sure everything works, even when you’re sick and tired of testing. Only after extensive testing should you go live, but even then go live slowly launching only a few campaigns or lower bids so any problems that testing missed can be sorted out. When you’re absolutely positive no mistakes exists, turn everything up to 11.
Every time you buy an ad – if you’ve followed John’s advice and tagged everything up – you create a huge volume of data. Make sure that your reporting is efficient you don’t want to spend too long on reports, you want to take action on what the data shows! This doesn’t mean you sure skimp on reporting – the commentary is important to help clients understand what’s happening. Analysis will help you to understanding and using techniques such as data visualisation is likely to help you get more from display – not only because you can make improvements, but also because it’ll be easier to persuade clients to give you money to run new campaigns.
As with (almost) every speaker, John touched on the tricky subject of attribution. His advice was to run controlled experiments, showing a limited group of people a non-brand “public service” ad to determine whether conversions can really be attributed to display. In one particular case they found that 31% of conversions from users that visited a site less than a day ago is directly attributable to display. This percentage rose to 41% of conversions later on, although this dropped off again after 30 days, when old users start to behave like new users again.
And just a final thought from John: document everything so everything is clear – to you, colleagues and clients – and so there’s a record. This both helps if things work really well and you want to replicate the results or for seeing where things went wrong – either way it will help you to create even better campaigns in future.
Top 10 Marketing Predictions for 2014 by Martyn Bentley (@martynbentley)
Martyn Bentley from programmatic advertising platform (aka ad sellers) Chango, introduced their marketing predictions for 2014:
- Audience and device fragmentation will increase. This isn’t a new phenomenon, Martyn is merely predicting we’ll see more of the same, here are some stats to back this up:
- Currently 21x an hour people are swapping devices.
- On average 43% of people watching TV are using a second screen.
- 50% of internet users will also be tablet users by 2015.
- Flow or sequential advertising that can follow users across devices will launch. There have been tests using microchips to show mobile users ads based on the TV show they’ve just watched and this in Martyn’s view is the future.
- Marketing silos will continue to disappear. Traditionally different aspects of marketing have been treated as entirely separate, but with attribution becoming an ever hotter topic the issue of who claims a conversion is going to continue in importance. We’re already starting to connect the dots and consider the role of everything and this will increase.
- Native advertising (e.g. branded content on big publisher’s front pages) won’t scale well. Some newspapers are beginning to offer this on a CPA basis, but it can’t become programmatic, because there will always be a need for manual control for quality and to fit editorial quidelines.
- Mobile will get better. Again, this isn’t new, it’s been “the year of the mobile” for how long now? And yet, it’s undeniable that mobile use is growing and if you haven’t already got the message, you need to take mobile seriously. While we aren’t there yet, there’s the potential to use cookieless IDs to start tracking users across devices, and we need to be ready to make the most of that. Again, Martyn hit us with some stats (all from the US):
- 43% year on year increase in Thanksgiving traffic from mobiles in 2013.
- 22% of Black Friday sales were on mobiles in 2013.
- Content marketing will go mainstream. it’s easy to think that it’s already mainstream, because the digital industry – and B2Bs more generally – are so good at this already, but more people will do more. This isn’t because more companies are caring about SEO – just that consumers are more interested in brand and ethos and content is an excellent way to communicate that. Martyn pointed to the Screwfix page as a good example of content on Facebook – go check them out.
- Corporate marketers will embraced networked innovation. E.g. funding start ups to learn from them. In Martyn’s words “you can’t be the disco vicar anymore”, you have to have something to say to people and working with a start up is a great way to stay relevant.
- The cookie won’t die – at least not yet. Having mentioned that cookieless IDs are being considered, they aren’t going to die out anytime soon. The growth of the Facebook Exchange is based on cookies. Yes there is work on alternatives, but they’re too valuable to go.
- Marketers will get cosy with their CTOs and CIOs. In other words, people are going to make better use of data to improve performance.
- Video will explode. 20 million UK users watch at least one online video per week. And while TV viewing is increasing, users are often double screening – refer back to those earlier stats – or simply not watching at a set time anymore – appointment viewing is dying out. As a result, marketers are preparing to move money from TV to online video, where it’s like.
Martyn’s final note is quite simply that 2014 is going to be a year of growth all round! In particular, mobile display is going to grow and as yet we aren’t creating separate strategies to handle this – we really need to start doing this now to keep ahead of the curve.
International Campaigns Why Biddable Media is the Answer by Nandita Patkar
Throughout the day many speakers were blighted by the Curse of the Unreliable Clicker. No one handled it better than speaker Nandita Patkar, who, having realised that she’d accidentally skipped through half her presentation in minutes, announced “thank God I’m not paying per click”. I salute you on your super pun lady!
But of course the presentation was more than just quality puns, Nandita also had some interesting stuff to share with us all. We’re seeing huge growth in Internet usage – but that’s not just in western markets – Asia and Latin Americas are growing enormously too. Their economies are growing, which is changing the structure of their societies completely. A rising affluent class is creating demand for more products and services. These people are tech savvy – so there’s the technology in place to support internet growth, which is where the opportunity lies for clever marketers.
In these cultures, there’s a lack of integrated brand development and because there’s so much to do, people don’t know where to start. Biddable media is the way forward as it’s ideally placed to connect with these emerging markets. We’re already great at balancing a mix of demands and optimising for success; to win in a global market we need to introduce culture into that mix.
It may sound obvious, but Nandita makes it clear that “culture” is very broad concept. We need to look at it practically – Google for information about how people operate in other countries to create hypothesis of what you need to do. Make use of the technology available to seek out audience insights.
We need to find the right context for our ads – the language barrier undoubtedly makes paid targeting more complex. While language is important, local is more important, in other words you need to use local speakers because simply translating text doesn’t give the right keywords.
And as with all biddable media, it’s important that we target the right people who are likely to become customers. Take a look at user behaviour on your site, use 3rd party data to get more understanding of the audience and don’t forget to do remarketing!
Nandita tells us it’s all about “glocalising” (klaxon! Wanky word alert!). In other words, we’ve got to localise ads for cultural resonance. Think about how can you apply your global brand values in a way that speaks to everyone. Nielsen tells us that humour is the theme that resonates most with consumers globally, but jokes don’t work in different cultures, which Nandita demonstrated with a Japanese joke about Hawaii?
Moving swiftly on, Nandita tells us that we generally don’t talk about websites in optimisation but post-click is super important for global users. We have to think about conversion data, because it’s really important. We have better understanding of attribution, but we must use this to prioritise data. Don’t have preconceptions of how you can use channels, it’s different for different markets. We can create highly targeted strategies based on local data. Build your strategy around a person – your target customer.
How Profitable is Google AdWords for You? by Ali White (@AlistairWhite)