Blog

BLOG

Keep it human for AI & Analytics success: eMetrics 2017

PUBLISHED ON 8 NOVEMBER, 2017

The eMetrics Summit was held last week in New York and offered an incredible opportunity for marketing analytics professionals to get up to date with latest industry developments and look ahead to potential future applications of technology and expertise.

While the Artificial Intelligence and Analytics tracks enabled us to learn about practical implementations, the overriding theme of the conference was how to ensure you’re optimizing the value of your human attributes alongside technology. Jim Sterne and Adam Greco shared some practical tips in their keynotes that we’ve summarized below.

“Computers are incredibly fast, accurate and stupid. Human beings are incredibly slow, inaccurate and brilliant. Together they are powerful beyond imagination” – Albert Einstein

Jim Sterne, eMetrics Summit Founder and industry leader, reminded us not to fear Artificial Intelligence (AI) but to work to optimize its application in marketing. After sharing a history of the 50 year path to present-day AI, he discussed its 5 principal applications and shared some examples of implementations. But more valuable than his breakdown of Natural Language Processing, Computer Vision, Conversation Bots and Machine Learning, was his outline of how to successfully onboard AI in organizations. He shared the following recommendations:

1. Clearly identify goals prior to getting started for a laser focused implementation.
2. Know your data. Ensure that you are able to access reliable and accurate data in a timely and consistent format. Equally importantly, make certain that your data is relevant, clearly defined and presented in a manner that promotes understanding of results.
3. Start with repetitive, taxing tasks machines can do better, such as: ranking, counting, measuring, sorting big data, and finding patterns, irregularities and look-alikes. You’ll likely find that Machine Learning will bring you efficiencies in testing, lead scoring, scheduling meetings, personalizing content, inbound e-mail sorting, social media monitoring, programmatic advertising, and creating social media messages & ad copy. Only once you’re confident that these tasks can be properly completed without your intervention, can you consider using AI for more complex tasks.
4. Sort out your technology partners. In the battle for build or buy, Jim recommends buy. Enjoy the efficiencies a solution provider can bring you and instead invest your time in improving outcomes.
5. Determine which data sets are useful. Utilizing clean and relevant data sets will ensure the quickest arrival at the best solution and reduce the margin for error.
6. Become proficient with the smell test. Sometimes things just don’t feel right. The commonly misconstrued ‘gut reaction’ is often the result of experience, intuition and emotion. While your ML algorithm can gain experience over time, those other factors are uniquely human and shouldn’t be taken lightly. If your results seem unlikely, they probably are; so make sure you’re always critically examining AI solutions to recalculate improbable or strange outcomes.
7. Be the change you want to see. In many cases, implementing AI across a range of tasks within an organization will mean significant change. Just as programmatic advertising changed the scope of work for many traditional advertising agencies, AI can also modify the distribution of work across many functions in your company. While some may fear the change this brings, lead by example by applying AI to some of your most repetitive tasks and then sharing the performance improvements you’ve gained by re-adjusting your focus.
8. Hone your domain knowledge. Never stop learning. Applying AI to a specific function is far from the end of the story. It’s your chance to restructure your work, leverage your skills and look for the next opportunity. Whether it leads to an incremental improvement or a massive shift, continual learning is the best way to constantly add value to your role.

Adam Greco, Senior Partner at Analytics Demystified and commonly known for his extensive Adobe Analytics expertise, ran us through his top lessons learned across a decade of #measure.

Interestingly, Adam’s recommendations centered around a few common themes:

Improve the way you work with business stakeholders

Taking it back to basics, Adam said that to increase the impact your work can have on the business, you actually need to get up and talk to your business-minded colleagues. He shared a couple of ideas to really ensure that your message resonates with them.

For example, always framing your findings in financial terms will have a great impact on these business-minded individuals, and additionally the use of visuals can be instrumental in enabling your (less number-driven) audience to retain your message. In addition, he recommended presenting conclusions from a consumer standpoint, rather than that of the analytics team. By sharing insights from “our customers” rather than starting with “we think…”, you’ll really be able to capture the attention of your user experience-driven counterparts.

He also recommended, however, ensuring that your business stakeholders don’t hijack your time for ineffectual analysis. For example, he explained that it’s essential to work with your colleagues to build analytics into processes at the beginning for optimal results. One (unfortunately all too common) example he shared is when tagging projects are requested after a new website implementation. Stand strong, he suggested – this is the only way to ensure these stakeholders consider analytics at the beginning of the process next time.

Furthermore, business requirements should help provide a framework for analytics success as they are often aligned with larger objectives and designed to help meet them. Be critical though, says Adam, ensure that the stakeholder is clear on what they will do with the data and how it will help ensure the meeting of goals, before you begin implementing.

Build a stronger team

Aside from those tips shared above on working effectively with business stakeholders, Adam shared some additional recommendations for building a strong analytics team and protecting their efficiency and effectiveness. He noted how important it is to ‘guard your team’s internal reputation’ by touting the quality of their work, but also that the most effective team members are those who feel valued. Listening to your colleagues and ensuring that the impact of their ideas is visible, is critical for helping you to retain a motivated and talented team.

Be the best analyst possible

To do so, Adam believes you should:
a) Be creative. Sometimes this is just as valuable as being smart.

1. b) Be persistent. You might not win every battle but don’t get caught up on short term losses when working towards success in the long term.
2. c) Be skeptical. Data quality matters so don’t settle for inferior data sets.
3. d) Be competent. Too often these days, individuals think they can get away with not being an expert on the tools they’re working with. Anything from mild incompetence to just sub-optimal knowledge can have an exponential impact on outcomes however, so make sure that you’re delivering the greatest value through expert product knowledge.
4. e) Be thorough, but not excessive. Integrate data across sources for a synergistic effect. In addition, providing content around your data in the form of feedback can truly strengthen the recommendations you’re making. On the other hand, don’t track something just because it’s possible to do so. Always focus on relevant and outcome-affecting data.

Plan for success

Adam reminded us that what we choose to analyze is critical to our success, and that only where we change the future can we claim to be successful. Afterall, analytics teams and tools can only provide ROI where their analysis leads to a data-based change and results in a positive impact.

Finding KPIs in which a significant rise or fall will result in someone being promoted or fired can guide us in how we invest our time and prioritize tasks, but most importantly, continuing to adapt implementations to new business requirements is critical for ongoing success.

 

If you attended eMetrics, we’d love to hear about the most valuable lessons you learnt from the incredible panel of speakers and knowledgeable attendees. Otherwise, what tips can you share for successfully implementing analytics and/or AI?