Two years ago, I prognosticated about what might happen with AI in CX in 2018. With the confidence of a seasoned pundit never held to account for his transgressions, I managed to get some readers lathered up about the future-state of AI tech affecting customer experience (CX), marketing, and sales efforts. Looking back, it’s surprising the authorities didn’t indict me as a co-conspirator in various AI hype-crimes.
I’m glad to report, so far anyway, no one has shown up at my door to stop me. Nonetheless, these days I do hold myself to higher standards. As such, I’ve decided to adjust my ways. Henceforth, my predictions are for what happens in 10 years not one. This way should someone email me in 2030 writing, “Vince, that sh#t you predicted in 2020 never came to fruition,” autoreply will fire back, “Thanks for your email. Please be advised responses are slow these days due to my semi-retirement.”
And here’s a bonus promise: From here on tips will come in prime numbers. Gone are the days of the 4 golden rules or the 6 pack of tips. In the ’20s, like this blog, you’ll get five things or maybe 37.
And with that, here are five AI in CX trends gaining steam and worth keeping eyes on as we boldly go into the next decade.
Those with the biggest, fastest, and best data win
Data’s value is fleeting. Like air trapped in an underwater cave, time sucks away at its utility. But also, like crap in storage units, we just keep stashing it away in the hopes someday the junk will appreciate. 99.9% of it won’t. So, stop storing it and start figuring out a way to use it while it’s useful.
Those with the widest-ranging data access rights, like the MAGA’s of the world (Microsoft, Amazon, Google, Apple) who can swiftly tap it, clean it, structure it, refine it, and exploit it will command the most value.
The number of firms focused on the data processing economy is staggering. Anytime someone stuffs this many logos on one slide that it blurs into a beautiful piece of abstract art, we can all agree the number of companies in that landscape is high:
Figure 1: Data Processing Landscape[i]
And the number of categories of database technologies is equally abstract and impressive:
Figure 2: Database technologies[ii]
Unless a large rock strikes from the Taurid meteoroid stream[iii] in 2022 or 2025 wiping us out, assume our database processing and storage fauna will continue to flourish and multiply. And for CX pros to get a leg up, they must stay on top of these developments, pivoting and flexing with the agility of a techno-acrobat whenever a speed or cost advantage arises. Those that can change the tires and refuel their moving CX race car with the latest database technologies will get an edge.
Why all this tech talk? Because great customer experience depends on great data collection, transformation, and most importantly gleaning great insights from it fast. But to surface insights such as:
- A more accurate readout on long-term consumer sentiment, loyalty, or propensity to churn
- Customers’ intent (in the moment) in terms of current shopping behavior and journey stage
- The propensity to open, click, respond, compare, test-drive, register, apply, buy, or refer
…you first need data, and yes with data time is of the essence.
Scikit happens… expect more Auto ML and fewer humans
Well, not fewer humans on the planet. Expect more of those. But when it comes to designing, building, and running predictive CX systems, count on more automation efforts that wring out manual tasks, such as data prep tasks, as well as streamlining model testing and tuning.
And count on more hard-to-pronounce algorithms surfacing that squeeze more insight juices from the data they feed on 24x7x365. Consider recent techniques that have risen to prominence like boosting, ensemble, generative adversarial, nearest-neighbor, and stacking. These techniques are yielding newer algorithms with eye-glazing names like XGBoost and GAN. That’s not slowing down.
And expect the plethora of self-driving ML platforms appearing (some sounding like college buddies of Brett Kavanaugh in a Star Trek episode) to expand. Examples: Big Squid, dotData, Datarobot, EdgeVerve, ElementAI, H2O.ai, Sagemaker, and Squark
Which leads to fewer data scientists, right? Wrong. Since there has been a shortage of those forever, they’ll be just fine. For you “full-stack” data scientists out there, that’s the good news. But hopefully, you really love working with data, because you will be drowning in it.
Robots will help with some of that prep work. By 2030, instead of you all reporting spending 80% of your time on data prep, you’ll proudly declare it’s now down to 79%. When asked why you still spend so much time on data prep, you’ll quickly point out that 80% of the circa-2020 big data is unstructured.
What does all this have to do with CX? Well, guess what most data scientists are doing? Marketing, Sales, and CX activities such as:
- Improving understanding of the customer – 46%
- Retaining customers – 37%
- Improving customer experience – 36%
- Cross-selling products & services – 35%
- Acquiring customers – 32%
Source: Rexer Survey of Data Scientists[iv]
Count on those allocations staying high.
If your robots commit crimes, you pay the fines
Because more brands are using machine learning models for regulated CX decision-making, such as ad personalization (regulated by laws such as GDPR), and loan determination (regulated by laws like the Fair Credit Reporting Act[v]), more cases are surfacing of models making seemingly unfair decisions. The latest glaring example being the Apple Card, whose algorithm, learning from bias data, made skewed credit decisions.
In this case, critics accuse Goldman’s algorithm (developed for the Apple Card) of extending lower credit limits to women with otherwise greater or equal credit scores (even Apple co-founder, Steve Wozniak, weighed in that he received 10x the credit as his wife even though they shared financials). And there are other cases of investigations underway, such as scrutiny on UHG’s algorithm used to prioritize patient care in hospitals. In addition, regulators are levying substantial fines on firms when they establish the algorithms make unfair, rigged, or biased decisions – regardless of whether the owners intended any harm. Examples:
- Google fined $2.7b in 2017, $5.0b in 2018, and $1.7b in 2019 by the EU for search bid placement bias
- Google penalized $21m in 2018 by India’s Competition Committee for bias in search engine results
In some cases, firms are getting squeamish and radically reigning in algorithmic targeting. For instance, in 2018 Facebook changed many of its targeting capabilities in its Custom Audiences ad tool.
The decade of witty connected platforms
Like giant stones that ancients somehow dragged and assembled into celestial observatories, the pillars of our modern CX technology will someday seem mysterious. Outsiders, trying to unravel their origins and logistical mysteries, will struggle without the luxury of having watched the process unfold.
Within 10 years, organizations will take the pile of platform stones and neatly stack them into place. Once done, enterprise and CX architects finally vindicated, will step back and marvel at their “Digital CX Enterprises,” aligning consumers and brands (hopefully more than twice per year).
Little technologies, like little stones, will play very small roles in this story. Unless you sell (or plan to play some part in developing) a major platform, you should be worried. In Martech, for example, there will be just a few major platforms. And those pillars will be data, content, and insight – which in turn will fuel the execution platforms. Consumers and firms alike will benefit from better-connected experiences and more valuable interactions.
Why is this important to CX? Because providing great customer experience is not episodic or transactional and it doesn’t begin and end with a customer service case or a single marketing offer. It needs to be perpetual, connecting, nurturing, and entertaining during the entire customer journey. And there is no way to do this at scale without intelligent and well-connected platforms.
What will those connected experiences look like? Thanks to better AI infused into the CX platforms, consumers can bank on:
- Their experience, when switching between self-service, automated, or human-assisted channels, being seamless and coherent
- Faster, relevant, and more convenient service
- More transparent access to their data, preferences, and interaction history
- Experiences that are personalized, fun, humorous, and story-worthy
On the last point, my bet is AI will play a bigger role in delivering a brand’s stand-up and improv. Why? Because AI is increasingly the brand voice (think voice assistants, chatbots, etc.) and it’s all about engaging experiences and the feelings they create. And though AI may not feel anything (or at least we better hope it doesn’t), AI needs to be an entertaining assistant. It’s happening already, so look out comedians, AI is coming for your job too.
For instance, ask Alexa to tell you a fat joke and she’ll respond in a clever, funny, and politically correct way. She steers away from the trap and tells you a joke about margarine or avocados. In my ‘20s, I didn’t get many predictions right, but I did write a final paper in one of my marketing classes about humor in advertising becoming a thing. At that point, it was rarely used. I think I nailed that one, although I probably got a C on the paper – professors – what do they know.
On personalization, I think it’s all about the hyper-personalization paradox – being relevant without crossing the creepy or too-edgy line. So if AI tries to be funny, it better learn what’s funny to one might be offensive to another.
People finally stop talking about AI in CX as if it’s new and cool
Sharp tools, wheels, electricity, steel, plastic – all huge innovations that changed history. But each eventually faded into the backdrop, becoming conventional technologies.
With a similar fate, AI’s novelty will dwindle, but its impact on the improved CX it powers will be real and evident. Chatbots, virtual assistants, VR, AR – all will be commonplace. The focus becomes the effectiveness of the products, services, and experiences, not the channels, technology, or algorithms used.
On to the ’20s
Truthfully, I love this life and career. I wake up every day and think, “I love the smell of AI in CX in the morning.” I know, weird huh?
But let’s face it. If
you are reading this and you made it this far, you love it too. And as we sail into the next decade, we all
have reason to be hyped-up. The future
of AI in CX is bright. And since we’ve
established that it’s time for the concept of AI on its own to go quietly into
the night, it’s probably better to just say the future of CX is bright.
[i] Mattturck.com, https://mattturck.com/data2019/, 2019
[ii] Researchgate.net, https://www.researchgate.net/publication/303562879_Setting_Up_a_Big_Data_Project_Challenges_Opportunities_Technologies_and_Optimization, 2016
[iii] Aanda.org, https://www.aanda.org/component/article?access=doi&doi=10.1051/0004-6361/201730787, 2017
[iv] Rexer Analytics, https://www.rexeranalytics.com/data-science-survey.html, 2017
[v] FTC, https://www.ftc.gov/enforcement/rules/rulemaking-regulatory-reform-proceedings/fair-credit-reporting-act, 1970