At the end of October, we had the pleasure of attending Hero Conf London, the world’s largest all-PPC conference. There were great speakers, insightful talks and a lot of great takeaways. A few trends for the direction of PPC became clear as the conference wore on.
- AI, Machine Learning and Automation
- RLSA and Audiences
- Google Shopping
- Back to Basics
Pretty much all the keynotes mentioned Machine Learning. Specifically, machine learning in the context of automating AdWords scripts. RLSA and audiences also drove a lot of discussion with several speakers stating that someone having interacted with your site or brand before is the strongest sign that they will re-engage and eventually convert and how that should drive your strategy.
Google Shopping is a rapidly growing channel where Google only gives us limited direct control over where ads display. Hacks and workarounds were the main topics of discussion around shopping.
We can’t cover all the interesting talks and content but we did want to share some details of a few of the keynotes and sessions we attended.
The Future of Google is Automated
“Within 5 years, Google AdWords will be automated.” Says Sophie Newton, COO of Brainlabs. Here are some interesting developments expected on the search marketer’s horizon:
One of the most manual elements of a PPC marketers job – Ad Creative testing, is now one of the areas that are on the path towards an automated optimisation solution. Mathematical techniques; word vectors and genetic algorithms, will provide a method akin to natural selection without the long wait for testing. In effect, results are delivered quickly.
Image recognition is being improved by machine learning, allowing AI to understand the context within images to a higher degree: Understanding if the image is ‘a woman carrying a handbag’ vs ‘a woman carrying a Gucci handbag’ will begin to evolve how advertisers provide their messaging through some of the largest repositories of images, such as Facebook or Snapchat.
Alongside machine learning and automated processes, the prediction is that voice search will grow – reaching 50% of all searches by 2020. Personal assistants like Siri, Alexa or Cortana will begin to bridge the gap and ‘cut out the middleman’ of search engines, providing access to answers, content and services more seamlessly than ever.
Brand Advertising in a Direct-Response Marketing World
Larry Kim kicked the breakout sessions off with a bang. In his provocative talk, he made a case for why you should stop all your paid search other than audience retargeting activity.
His main argument draws on the fact that people who have interacted with your brand are so much more likely to convert that they make any other optimisation signal negligible by comparison. In many of his clients’ audiences who have already interacted with his clients’ brand have CTRs and CVRs which are double or more the non-audience activity. The vast majority of the clients’ revenues comes from this activity at a fraction of the overall cost. This approach can make even previously unviable generic keywords strong revenue drivers.
Hence instead on focusing on doing the prospecting and initial brand engagement step in search Larry Kim advocates sticking to targeting already engaged users on search and redeploying the budget saved into channels like Facebook where new audience acquisition and engagement is much cheaper. Demographic and interest targeting on these social and display platforms can be used to hone in on potential customers and people who match your current customer base.
This approach would certainly be of interest for companies with tight margins working in a very competitive environment. The examples used certainly fit that profile with one being a greeting card business for example.
Google Shopping Feed Hacks
Rob Watson, Head of Digital at Supplyant shared some of his most interesting Google Shopping hacks:
- Maximise 150 Character Titles, even though only 70 characters are shown, use the additional space to increase coverage over your most relevant terms.
- Alternative Landing Page attribute can be populated with a page different to the original link, allowing for more customised landing pages for Google Shopping visitors.
- Use Labels to promote Best Sellers, group your top products and bid differently for them compared to the rest.
- Broader terms, such as ‘living room furniture’ can use the AdWords redirect to land on the category page. Recently, Google has rolled out Showcase Shopping Ads for this.
- Use the dimension reporting to slice and dice the shopping data to understand performance at a deeper level.
- Break product targets through the quality score barrier and generates more traffic for a cheaper CPC by giving high bids for a short period of time, then pulling back.
- Replace the product assigned to a high traffic but low performing product ID with another product. Impressions are mostly retained, but the caveat is that the product must be like-for-like.
Statistical Concepts Every PPC Specialist Should Know
With all this discussion of the current PPC cutting edge and new developments, it is also very useful to be reminded of some of the very important basics. In this session, Frederik Hyldig of s360 covered some statistical and mathematical pitfalls which are easy to overlook.
His talk covered the law of large numbers, the complications around discussing and working with relative and absolute numbers and percentages, the Pareto principle, Simpson’s paradox, misleading graphs and so forth all illustrated with examples taken from everyday PPC practice.
We could only pick out a few of the many excellent talks we saw during this conference and there is so much more we could cover. This is not even going into the out of session chats we had with many experienced paid search experts. One of the striking things was the audience survey at the start of the conference which revealed that well over half of attendees had more than 5 years of PPC experience. This drove a generally high standard of discussion and hence great learning opportunities.