It’s probably not difficult to grasp that our customers’ purchase behaviors are deeply entangled with moods.

There’s a reason that we call shopping therapeutic. Purchasing things we want sends a serotonin surge to the brain that can temporarily make us feel better if we’re stressed, depressed or anxious. Moreover, according to widely-cited research by Gerald Zaltman, 95% of purchase decisions are made subconsciously and driven by emotions — so it’s no surprise that advertisers have been interested in understanding and evoking particular mood states for generations.

Now that data about internal states of mind is becoming more available, the stakes are higher when we consider how to act on this sensitive consumer information. For example, how far should brands go to utilize emotional data to encourage purchases?

Let’s take a look at where we’re at and how brands can take a human-centered approach to the use of this sensitive information.

Related: 5 Insights Into Human Behavior That Will Boost Your Sales and Marketing

How we gauge emotions

Let’s start with how we gauge emotions. Until recently, our data about feelings relied on self-reporting by consumers since it’s impossible to embody another person’s emotional experience. Self-reporting means that consumers answer direct questions about how they are feeling at a given time or in a given context. Usually, this happens via market research surveys.

Neuroscience is advancing to the point that we may be able to accurately predict emotional states without relying on overt consumer admissions. This type of emotional assessment may prove to be even more accurate than direct consumer reporting since many people struggle to predict how they’ll feel in particular contexts.

Technology that assesses activity in our brains is getting more advanced and better capable of predicting mood states. While most of this innovation is happening in research labs, we’re getting closer to realizing this technology as a marketing tool.

Neuroscience and wearables

The Art of Shopping, a subconscious shopping experience between art retailer Saatchi and eBay, is one of the most direct campaigns that aimed to utilize this technology in shopping.

During the experiential retail event, attendees browsed an art gallery while wearing headsets that were designed to track a consumer’s mental engagement. When the software suggested that viewers were inspired, eBay added similar items to the patron’s shopping cart.

While the activation was interesting, getting consumers to voluntarily and consistently wear mind-tracking headsets is far-fetched in our current environment. Although, it may become more common as more consumers adopt augmented and virtual realities.

Today, wearables like fitness trackers and smartwatches are becoming more ubiquitous and can aggregate mood data inferentially or from the self-assessment of consumers. The devices can assess everything from our heart rate and breathing patterns to our mindfulness activity. This can imply or correlate to stress levels or provide more direct mood data on apps like Calm and Halo that encourage emotional reporting.

Related: 4 Neuromarketing Hacks to Reach More People and Maximize Results

Inferring emotional data

There are other ways to gauge the mood of consumers, and some of them have a troubling history.

Meta, formerly Facebook, was famously under the microscope for conducting a large-scale emotion experiment aimed at understanding if emotions spread through networks.

It actively manipulated the algorithm of nearly 700,000 users without their informed consent, in order to serve them positive or negative content and to gauge the apparent mood in their resulting posts. Among other goals, the company was interested in how emotions might make the site more or less engaging.

The more engaged users are on the platform, the more valuable they are to Meta’s advertisers. Critics worried that the company wanted to understand how to manipulate emotions to bolster its bottom line and increase purchases for its advertisers, without apparent regard for the impact on the consumer.

Meta isn’t the only tech company making actionable inferences about emotions. Search engines like Google track emotional effects by utilizing software to assess language for positive and negative sentiments in search, among other tactics.

In conjunction with the rest of their consumer data, such as browsing and purchase history, these tech behemoths have real power to understand, contextualize and leverage consumer emotion without the use of neurological equipment.

Related: If You Want to Win Over Customers, Appeal to Their Emotions

How are we using this data?

Marketers are curious about how mood impacts purchases, and thereby interested in creating purchase paths that are aligned with particular feelings. Payment providers are paying attention as well. In fact, in their latest Future of Payments research paper, Worldpay from FIS identified personalization, including emotional engagement, as a trend that payment providers are attending to.

Creating payment journeys that utilize emotional information from consumers may sound troubling. But it’s worth noting that consumers increasingly expect these kinds of personalized experiences from brands — as long as they are additive to the consumer journey.

When an experience provides convenience to a consumer and helps the brand connect meaningfully with them, it can make the consumer feel supported and improve emotional engagement and loyalty.

Striking a balance between utilizing emotional data to offer mutual brand-consumer gains while respecting consumer rights and privacy is tricky. This is why we need to think deeply about creating consumer safeguards as we venture into the future.

Related: Personalization: A Perspective On The Future Of Targeting

Where do we go from here?

There’s no shortage of data, and we’re only going to get better at detecting and reacting to emotional states in various contexts. As advertisers and marketers, we need to be thoughtful about how all this emotional data is applied.

We’ve already seen social media companies exploiting negative emotional states like anxiety and depression to move users toward a purchase path of aspirational products in categories like beauty and fitness (What’s even more troubling is that the algorithms are likely contributing to the negative emotional state, but that’s a conversation for another day). We’ve seen the same algorithms promote negative headlines that are likely to elicit engagement, which results in exacerbated political polarization and negative societal impacts more broadly.

As an advertising community, we need to implement safeguards to protect consumers. These safeguards should come from regulators, as well as individual brands. Creating an ethics playbook prior to locking in uses of emotional data in the purchase path, conducting thought experiments for secondary and tertiary impacts of the use of mood-based information, and defining and acting in accordance with a brand’s values can help to ensure marketers are responsible brokers of mood data.

It’s worth remembering that understanding emotion can have powerful positive consequences as well. As humans, we’re emotional beings and brands that can meet consumers where they are in their internal experiences are likely to create better and more meaningful connections. It’s imperative for brands to think through how they’re using emotional information, not only to create lasting relationships with consumers but also to take a human-centered approach to innovation.