Visual content is central to online purchasing decisions:

• 76% of shoppers are compelled to click because of great product images.
• 78% of shoppers find product images and descriptions extremely or very important to their buying journey.
• 95% of purchasing decisions happen when the subconscious brain is processing visual information.
• #1 reason that consumers don’t convert online today is bad product content.

Despite this impact, historically it’s been incredibly difficult to measure and optimize this content at scale.

One technology we’ve found value in is Vizit, an AI platform that learns from millions of organic online consumer interactions with imagery in order to replicate their visual preferences in an algorithm. Vizit’s software utilizes this large-scale learning to predictively score visual content on effectiveness for unique consumer audiences and provide a suite of visual attention, engagement, and optimization insights. In the world of digital shelf content, which is notoriously difficult to measure, this technology provides a data-driven approach to understanding the effectiveness of visual content before it’s ever syndicated, creating both confidence and buy-in from e-commerce and digital content leadership and teams.

These pre-launch proof points have proven invaluable, driving a stronger partnership between strategy and creative teams, and creating an objective consensus and dialogue about what “good content” really looks like - all from the point of view of who matters most - the consumer. Ultimately, the value we derive from AI tools is only possible because of the work we put into how and when to use them.

We’ve seen several tangible benefits using tools like Vizit: stronger creative, pre-launch moments of proof that validate our creative direction, the ability to tie predicted engagement to real digital shelf sales lift, and time saved in future revisions.

Internally, Vizit has pushed our creative team to rethink long-standing ways of working. One specific example of this comes to life in an image type commonly found across the digital shelf: promotion of subscription programs. Testing images in Vizit showed our team that maybe the way it’s always been done isn’t the best way to do it, and we can create more effective content by trying new ideas. CourtAvenue’s Executive Director of Creative, John Jacobsen , says it best:

“Historically, the subscription tile has been an industry standard, and the approach doesn't vary too much. We need to show the product clearly, and if possible in situ (on a front porch, for example). We checked all these boxes and felt good about it.

With Vizit and the help of AI/generative imagery, we're able to mock up our own front porch scenes and explore various angles. Vizit testing helped us minimize clutter found within typical home delivery compositions while enabling maximum label clarity and legibility within the context of a shipping box. The difference was undeniable.”

AI tools help our internal teams break down creative biases and rethink the status quo, pushing us to develop richer creative. Additionally, our optimization process ensures we’re confident in the effectiveness of our final product before it goes live.

The impact of using AI tools for image optimizations, whether they measure engagement, affinity, or another metric, can be tied to real sales lift on the digital shelf. For a major CPG client, we’ve been able to tie our full digital shelf optimizations (of which Vizit is a key component) to sales lift and predicted increase in conversion across the entire product set. This ability arms our clients with powerful data points that allow for real change.

When using Vizit with clients, we’ve found that although educating our partners on what the tool is and how it is used takes a level of effort upfront, it pays off tenfold. Clients understand and trust the AI tool used to guide optimizations, which allows for fewer rounds of feedback. By strategically focusing on education upfront, we’re able to foster a more collaborative, open relationship with our clients and more effectively use AI-based tools.

In almost 2 years of using Vizit internally for creative optimizations, we’ve gathered a list of 4 lessons we consider imperative to the effective integration of AI tools:

1. Strategic insight is key.

Without a strategic plan for how to use a tool, it often ends up under-utilized–another contract that sits on a shelf and gathers dust. How do we operationalize it?

• Vizit is a singular piece of Uplift, our 5-step content creation process. In addition to Vizit, we field category research, conduct SEO analysis, and validate hero image optimizations with real shoppers. Leveraging it as a piece of a bigger strategic process allows us to be purposeful with when and how it is used.

• We recommend using Vizit at the beginning of content generation, allowing you to gather learnings on what brand assets, layouts, and lifestyle imagery are most likely to engage your target audience and to get a baseline for how current imagery is performing.

• We think critically about the outputs and how to optimize based on the provided heat maps and engagement scores. Without a plan, you risk causing frustration and unnecessary work with changes that don’t move the needle.

• We’ve found that tools like Vizit work best when there is a small team of strategic power users that partner closely with creative. The strategic resource can guide the creative optimizations, allowing for more targeted changes and fewer rounds of optimizations. A strong partnership between creative and strategy builds a stronger end product.

2. Even small test & learns can have a big impact.

• AI-based tools like Vizit allow for quick test and learns, providing instantaneous feedback and allowing for quick-turn optimizations.
• This ability allows brand teams to test different versions of images or different brand assets in a short period of time, creating opportunity for increased testing compared to traditional research methods.
• Knowing how to extrapolate these quick-turn tests into larger trends is key, allowing teams to balance resources between garnering insights through testing and applying those insights across new channels and creative.
• “We’ve seen time and time again that even small, seemingly subtle changes to imagery can have a massive impact on consumer perception and content effectiveness. Something as simple as rotating a product or adding depth in a product hero image has the potential to drive double-digit lifts in conversion. Having the power to understand and predict the impact these changes have in real-time is a massive advantage for brands when it comes to increasing the efficiency of their content creation and testing processes across the digital shelf.” Eli Orkin, SVP of Marketing, Founding Team Member Vizit

3. Know your limits.

Going into testing with a clear set of guidelines is key to avoiding analysis paralysis. At CourtAvenue, we leverage the following guardrails when using a tool like Vizit:

• Limit the number of rounds – you could continue optimizing forever, but we’ve found limiting our optimizations to 3 rounds with a few options each is the most efficient.
• Limit the number of assets tested – similarly, you could invest in testing every image that goes live but still come out with the same few key lessons that are applicable across the board. We identify a few test SKUs before we start and focus our learnings on those specifically.
• Be flexible between aiming for a certain predicted engagement score vs. aiming for a certain level of improvement. Sometimes, the image’s message or specific brand assets may not test very well. In these cases, we focus on improvements instead of achieving a certain score.
• Understand that sometimes brand assets don’t drive engagement - and that’s okay. Brands are built on driving a variety of emotions, and not all of those result in highly engaging assets. Remembering that at the end of the day, brands mean more than their engagement levels is useful for level-setting among clients and internal stakeholders.

4. Impact is broader than digital shelf assets.

• While initially used for PDP imagery, all image-focused creative assets could be tested to find opportunities for further optimization.
• Beyond PDP content, AI image tools can also inform media creative and sizing. With FlowState, our proprietary e-commerce operating system, we’re able to combine these insights with media insights to drive more effective media placements that deliver true category growth.
• Sharing how the tool can be used opens up opportunities for additional testing. For example, when we shared Vizit’s capabilities, the physical packaging team for a large pet care brand asked to test variations of an upcoming pack redesign.

Tools like Vizit offer capabilities that allow for more effective and engaging creative that can help brands drive views, conversions, and better shopping experiences. As AI capabilities continue to evolve, we see opportunities to use tools like Vizit for validation of additional creative assets, including video, media creative, and campaign landing destinations.

Are your images working as hard as they need to in order to stand out on a crowded digital shelf? Check out our commerce capabilities on our website, or reach out to our Executive Director of CPG and Retail, Rory Foster, to test how AI-driven image optimizations can lead to richer creative for your brands.

Sources:

1. Salsify
2. Salsify
3. Inc.
4. Salsify