Instagram user browsing photos.
These five ground rules can help marketers stay human when deploying artificial intelligence on social media. PHOTO: Erik Lucatero on Unsplash

Bot backlash is upon us. On Facebook, where companies once raced to implement time-saving bots, failure rates are reported to have hit 70 percent. 

Only three in 10 interactions go off without a hitch, according to recent reports. Customers complain that AI-powered bots bumble all but the most basic of queries, give nonsensical responses and waste more time than they save.

The irony, of course, is that all this botched customer service is playing out on social media, which not long ago promised to be the antidote to tiresome phone trees and subpar support. For brands on social media, this raises a critical — even existential — question.

Automation and AI are enabling companies to scale social media efforts in unprecedented ways — serving more customers and extending social media company-wide. Yet, social media’s strength and differentiator has always been the human element. In our quest to scale, what should we automate and what should we leave in human hands?

There are no easy answers here. But from the frontlines, I’ve seen some clear dos and plenty of don’ts. Success ultimately comes down to choosing tools that enable customers’ experiences to be more human, not less. This is true of all the best technology — from software to smartphones to the internet itself.

Without further ado, here are a few basic ground rules for preserving human connection at scale on social media in the age of AI.   

1. Smart Listening? Sign Me Up!

We all know the old , listen before you speak. Yet, too many brands still ignore this when it comes to social media. Networks like Facebook, Twitter and Instagram offer access to a global focus group, in real-time, with candid feedback from consumers and stakeholders on products, interests and problems. Data is raw and unfiltered but can be overwhelming and impossible to sort through at scale.

This is where technology absolutely has a role to play. The latest generation of smart social listening tools analyze sentiment, identify trends and surface important conversations, automatically. These tools aren’t foolproof, but they’re advancing quickly and, at the least, enable companies to take a quick pulse check on what their social following is saying and why.

Prime example: Monster, the job-site, adopted smart listening tools to continuously scan more than 20 social platforms for conversations involving people “going on an interview” or who “don’t like their job.” They successfully targeted thousands of job seekers and boosted their own social media mentions by 300 percent — all while spending 80 percent less time on task.  

2. Streamline the Creative Process

The behind-the-scenes logistics that go into creating and distributing social media content are ripe for AI and automation. Take something as simple as scheduling. Just as smart calendar apps now suss out the ideal time-slot in your day for a meeting, social scheduling tools automatically identify optimal times to post based on audience and goals.

But, increasingly, it’s not just when to post, but what to post that’s being automated. Once you understand your audience's interests, it makes sense to leverage tech to surface relevant content. Algorithm-driven social media suggestion tools (like Primal and UpContent) automatically identify trending articles, blog posts, images and more based around specific keywords. A human touch is still important to vet these suggestions, but the grunt work is handled by machine.

3. Automate Those Social Media Ads

Social media advertising also begs for automation. Recent years have seen most major networks evolve into pay-to-play zones. (Without ads, large companies currently reach as few as two percent of their Facebook fans.) Yet, the ad-buying process is time-consuming and… well… technical. Small changes in choice of copy or photos can dramatically impact how many people click on and see the posts. Identifying the right audience and making the right “bid” is decidedly hit or miss.

But new ad technology enables testing hundreds of ad variations, simultaneously. These tools automatically identify which variants are performing best, putting extra spend behind the winners, while withdrawing budget from duds. This kind of process could consume countless employee hours to get right — time better spent on the creative and strategic decisions that really require a human touch.

4. Customer-Facing Bots? Buyer Beware

For the moment, most consumer-facing interactions on social media and messaging platforms are best left to people. Bots have obvious cost-cutting appeal, but they generally fail the frustration test. As some observers have flagged, “the AI technology used to power chatbots simply isn’t mature enough to come close to replacing humans for anything but the most trivial tasks.”

These are early days, to be sure. And I’m confident that customer-facing chatbots on social media will evolve rapidly as AI advances, not unlike voice transcription software such as Siri. For the moment, however, I think a handy rule of thumb is: Does this bot get customers to a solution faster than other options they already know and expect (from FAQs to phone trees?) If the answer is no, tread with caution.

5. The Ultimate Test: Does It Make the Experience More Human?

To be clear, scale absolutely matters when it comes to social media. Companies serving thousands — or millions — of consumers can’t always afford the luxury of one-on-one interaction. Plus, social media now touches every phase of a customer’s journey — from consideration to advocacy. Tracking, reacting to and harnessing that information can be overwhelming. The desire to automate is understandable. The key is to do so selectively and strategically.

In the end, the right social media tech for a business, be that smart listening tools or content suggestion apps or even chat bots, should free up real, live people to spend more time engaging with real, live customers. Otherwise, you’re taking the “social” out of social media… and that’s a self-defeating proposition.