HeroConf London 2019 – Key takeaways
…Why PPC marketers are pretty clueless about the future of digital marketing.
Data-driven, AI, automation, store-visits, privacy and user experience were the very topics on everyone’s mind and the same questions remain after these 2 very interesting days. Marketers have no clear idea on what to do or how to position themselves about the future of digital marketing. Let’s navigate the main takeaways:
Is AI/Automation going to push marketers out of their jobs?
Artificial intelligence and machine learning were obviously amongst the most followed topics during these 2 days and conferences about these subjects easily filled up the conference room. Let’s temper the expectations right away, there is no groundbreaking AI that’ll replace marketing teams nor awesome hacks to crack the code and master algorithms.
Like Frederick Valleys from Optimyzr had explained very well, AI is a tool and we need experts to add the final touch. This final touch is « context » that any algorithm will never be able to fully understand.
These automation solutions allow us the work « smarter » and more efficiently and we should totally look into it to streamline our processes, but we need experts to master them. Machine to machine marketing is not likely to be the answer. As mentioned by several speakers, algorithms have increased in quality and capabilities exponentially for the past 10 years, but this curve is plateauing and computers won’t get any much better.
The conclusion would be to find a soft spot between man and machine, keeping the final user in mind.
You can have the smartest bidding algorithm and the better attribution algorithm, if you provide bad user experience, you’ll fail.
Fun fact, a speaker asked, « Do you think chatbots are interesting and would you add it to your marketing mix ». 98% of the audience hands up. This was followed by a second question « As a customer, do you like and use chatbots? ».
2 hands up. It summarizes quite well the situation, technical tools are available but we need marketers to create a great experience.
Data and audiences in a cookie-free world and why Data-driven models suck.
« You have to leverage your 1st and 3rd party data to the moon. »
« You have to make the best out of audiences and uber complex customer journey. »
You could hear that in a lot of talks and see nice examples of how the data was mastered showing awesome results.
But very few dare to speak openly about the core issue and dodge the main question in the Q&A’s.
How to tackle GDPR, ITP, e-privacy, walled gardens challenges and overcome hurdles in an ecosystem where cookies are dying?
We know it’s happening, we know we should think about it, we know it’s not going any better anytime soon but no one is really taking the bull by the horns.
Blindly trusting data and data-driven model’s in a world where that very data is not going to get any better is not « smart ». Thinking that we can map complex customer journeys and provide a tailor-made solution at every touch-point based on biased data to skyrocket performances is a fantasy.
These questions remains big question marks and marketeers are clueless.
Few solutions were discussed, but are either very complicated/costly or not applicable in most cases. Also, everyone agreed these workarounds won’t probably work anymore in a few times. It’s a cat and mouse game.
UX Designers should be the best-paid guys in the building.
The opening talk featured Rory Sutherland, and we can say it, he rocked the party!
He gave a very smart (and fun) talk on why we should focus on user experience « in a world where we have to pretend that everything makes sense ».
He showed that video where they put Ed Sheran in a 2$ peep show with a creepy sales-guy on the driveway trying to get people in. Even though these people would very likely pay 100$ tickets to live the experience of an Ed Sheran’s show, they won’t enter for 2$ dollars private show since the user experience was terrible.
This analogy was very easy to understand in our digital marketing jobs. We have to make « people marketing » not « spreadsheet marketing ». We usually spend more time analyzing data and tweaking algorithm/automation rather than write new ad copies or create new banners right?
Very often, the most effective way to increase conversions is to increase conversion rates by providing a better user experience, not just throwing more budget in the machine.
Forget shiny objects. A « store-visit » story.
Agencies used to jump on last marketing shiny objects but tend to forget the incremental value. Don’t get us wrong, research and testing is a very important part of the process. As Rory Sutherland mentioned in his presentation, luck and irrationality matter too.
Store-visit is one of these stories. After a speech about bridging the gap between online marketing and offline behavior, the Q&A has been very interesting but the conclusions stayed the same :
• Revenue could be bad on a PPC or shopping campaign be could have a positive impact on store visits
• A very good ROI campaign won’t necessarily impact in-store traffic.
• In-store results are extrapolation.
• We can’t split organic and paid “go-to store”.
• Incrementality is very hard to measure.
« Trust the tech and let it be » was the final word.
We can’t really agree with that….
Conclusion: Where are PPC managers going in 2020?
As Fred Valleys mentioned, as digital marketers today, we are PPC doctors. Taking context into consideration to provide the best treatment. We are also PPC pilots, managing auto-pilot for 95% of the flight and putting hands on the bar for the most important moments or if something is going wrong.
The PPC manager role is evolving. We used to crunch numbers in spreadsheets and make manual changes in slow tools. The role is evolving towards settings goals/KPI’s, picking automation and monitor them, teaching the machines and troubleshooting as well as leveraging insights across platforms.
• We have to challenge tech and ad tech. Not blindly follow it to avoid self-fulfilling prophecies.
• We have to challenge ourselves as marketers. Cut the bullshit, place user experience in a central position, look at the big picture and think “incrementality”.
• We have to challenge our organizations. Stop working in silo’s, think about integrated marketing and open communication and reporting.
But most importantly, we have to use our common sense more than ever. As things are evolving so fast, there is no time for the status quo.
Thanks again to all the speakers and special shout out the staff for making this event possible. The app to interact with the speaker was great. Following these events allow us at Clicktrust to be on top of what’s going on and try to find answers to help our clients and partners to overcome the hurdles about their digital marketing journey in a world in constant evolution.
3 months ago
Julien Delbauve
Data-driven + lots of brand keywords? Start thinking like a marketeer…
[activecampaign form=3]PPC & attribution experts only. This requires a deep understanding of search and an obsessive passion for paid digital media. I expect some negative feedback on this. Allow me to indulge 😉
Context:
- Client is ROI and Cost/sale sensitive (duh) and is currently facing very high cost per sale numbers on generic search terms in a market with CPC’s averaging 9 EUR .
- They desperately want to expand generic volume, and have gone through several agencies who each have failed to deliver this.
- The account contains both generic and branded keywords on both exact and broad match.
- The average time between first click and conversion is more than 3 days.
Studying the path length reports in the search attribution tool showed us that generic queries almost always convert combined with branded queries. We can hence assume that the data driven algorithm that is looking at numbers would consider branded keywords pretty important.
Comparing data driven to first click attribution
Upon comparing data driven attribution to first click attribution through the attribution tool, the following happened:
- very strong decrease in sales from generic keywords and huge increase in cost per sale both in percentage and absolute numbers.
- massive increase in sales from branded keywords, and limited decrease in cost per sale in absolute numbers.
When showing this to the client, they were schocked. Nobody cared about the cost/sale of the branded keywords decreasing from 4 eur to 2 eur. The cost per sale of generic keywords that increased from 100 eur to 300 eur in the data driven model was however enough to leave them gobsmacked.
When data driven becomes a self-fulfilling prophecy
Statistically speaking, I’m sure the data driven model is correct. If it is providing weight to the number of times a specific query appears in the customer journey, than obviously branded queries will get a lot of weight. As data driven attribution will then decrease conversions on generic keywords, ROI sensitive clients will decrease bids. Branded queries will then become even more prominent, and the whole thing becomes a self-fulfilling prophecy.
Think revenue across channels – not within silo’s
From an outcome-driven marketing & “lets grow revenue” point of view, this makes no sense. Cost per sale on branded search moved from cheap to too cheap to care, while generic cost per sale increased by 3 compared to a first click attribution model. Meanwhile, the site is ranking very well organically on all branded search queries and has limited competitors bidding on their brand.
As a result, the actual value of generic queries is massively undervalued due to the presence of the branded queries in the account. Is is highly likely that the client would have gotten the sales that happened on the branded query through organic search anyway.
So if datadriven is not taking the less important role of branded search queries into account on sites that are ok SEO wise, and uses frequency of search queries in the customer journey as a factor in the algorithm, we can expect ROI sensitive PPC managers to decrease overal media spend in the long run, as they decrease investment on generic terms. That can’t be the goal.
I’ve seen this happen several times now on pure ecommerce players. They fail to scale the accounts, especially in smaller markets with products/services that have decision cycles beyond several days.
Possible solutions:
- Split branded queries and generic queries into seperate accounts so that data driven attribution can be used on generic queries only
- Create a test to remove branded queries alltogether. Incremental impact on sales volume from branded paid search queries together with organic queries should of course be assessed.
In both scenarios’s, by taking branded queries out of the account, conversion attribution will shift back to generic keywords and data driven will do its work on generic queries alone.
Sales and media spend on generic search queries will go up while overall sales from organic and paid search combined should increase.
In any case: before using data driven, check the importance of branded search queries in the account. If your sales are primarily coming from branded queries, and you wish to keep them in the account – you’re probably better off with first click attribution…(the horror!)
Of course, savvy digital marketeers should be going beyond that and look for attribution tools that take the incremental value of branded search queries and all forms of remarketing into account across the entire media mix…
If you’re interested by this post, this is the stuff the CLICKTRUST team thinks about daily while optimizing your accounts. We’d be happy to exchange ideas.
1 year ago
Fabian Van De Wiele