blog

Artificial Intelligence in Retail: What to Look for When Evaluating Software Vendors

Written by Michelle Bacharach | Nov 12, 2024

Let me tell you what I'm seeing in retail tech right now; there’s so many vendors claiming they've built the next AI solution that will transform the industry forever. But with hundreds, if not thousands of vendors claiming to offer the best solution, how do you separate the real deal from the hype?

I've spent the last decade working shoulder-to-shoulder with retailers and brands, and I'll be honest – I've seen applications for artificial intelligence in retail that truly transformed businesses, but I've also watched just as many AI companies crash and burn. Through all of this, I've picked up some important lessons about what really matters when you're choosing retail software. Let me share what I've learned.

The Challenge of Vendor Claims

Perhaps the most immediate challenge facing retail decision-makers is the interpretation of vendor claims. We've all seen the impressive statistics: "700% higher conversion rates!" or "10x ROI!" While these numbers might be factually accurate, they often lack crucial context.

I call this "vendor math" - misleading calculations that can paint an overly optimistic picture of performance. When evaluating artificial intelligence in retail, or technology solutions in general, expect to encounter vendors who may employ some of these “shady” strategies:

  • Claiming massive percentage improvements from a tiny baseline
  • Cherry-picking results from ideal conditions or limited time periods
  • Combining multiple metrics in non-standard ways
  • Using non-standardized measurement methodologies

For example, a high conversion rate on a specific landing page with minimal traffic may not meaningfully impact overall business performance.

The Artificial Intelligence in Retail Reality Check

Artificial intelligence in retail has become both a powerful tool and a marketing buzzword. Many solutions claim to be "AI-powered," but the reality often reveals a heavy reliance on human effort — sometimes as much as 90% human input versus 10% genuine AI capability. This balance isn't inherently problematic, but it can significantly impact scalability and long-term efficiency.

It's easy to be seduced by impressive demos and cutting-edge features. However, the most crucial question remains: Does this solution solve a real business problem? Focus on technologies that demonstrably improve revenue, profitability, or brand value. To evaluate AI capabilities effectively, look beyond the marketing claims.

Team Analysis: Uncovering the Human-to-AI Ratio

One effective method is to analyze the vendor's team composition through LinkedIn or other public sources:

  • Look for red flags like large teams of stylists, data labelers, or content creators
  • Compare the ratio of technical roles (ML engineers, data scientists) to manual roles
  • Check for recent hiring patterns that might indicate scaling challenges
  • Evaluate the experience level of the technical team members

A truly automated AI solution should require minimal manual intervention to scale - if you see large operational teams, that's a sign the vendor may be relying more on human effort than true AI technology.

Building Your Implementation Blueprint: Key Planning Steps

Even the most promising technology can fall short during implementation. Success requires transparent communication about processes, expectations, and potential hurdles. Legal and security considerations often emerge during this phase, particularly regarding data privacy and intellectual property rights.

Before beginning any implementation, it's crucial to:

  1. Map Your Internal Processes:
    • Document your current technology adoption workflow
    • Identify key stakeholders and decision-makers
    • Review past implementation successes and failures
    • Outline your specific security and compliance requirements
  2. Address Common Legal and Security Concerns:
    • Clarify how vendors will use brand assets for AI training
    • Establish guidelines for handling Personally Identifiable Information (PII)
    • Define data access and storage protocols
    • Set clear boundaries for AI model ownership and usage rights
  3. Create an Implementation Timeline: 
    • Include specific milestones and deliverables
    • Account for internal review and approval processes
    • Build in buffer time for unexpected challenges
    • Plan for adequate testing and validation periods

Address these proactively in early conversations rather than allowing them to become roadblocks mid-implementation.

The Organizational Alignment Imperative

Perhaps the most overlooked aspect of software procurement is ensuring organizational alignment. Different teams may operate with conflicting KPIs — for instance, marketing focusing on conversion rates while merchandising prioritizing margin. These disconnects can undermine even the most powerful technology solution.

To achieve better alignment and better evaluate vendor metrics and claims:

  1. Organize Internal Education Sessions
    • Establish standard industry metrics and their typical ranges
    • Educate about common vendor manipulation tactics to watch for
    • Prepare questions to ask vendors about measurement methodology
    • Walk-through how to request and verify raw data
  2. Establish a Centralized Communication Channel:
    • Create a single point of contact for vendor interactions
    • Maintain a shared repository of feedback and requests
    • Regular cross-functional team meetings to discuss progress
    • Document and distribute key decisions and changes
  3. Implement Strategic Review Processes:
    • Monthly alignment sessions with key stakeholders
    • Quarterly performance reviews against strategic goals
    • Annual reassessment of objectives and metrics
    • Regular vendor relationship health checks
  4. Manage Team Focus and Priorities:
    • Identify potential conflicts in team objectives early
    • Create shared KPIs that support overall goals
    • Monitor for unintended consequences of metric focus
    • Adjust incentives to promote collaborative success

Success requires a shared vision and unified objectives across all stakeholders. This means getting specific about how each team's goals tie together. The key is making sure everyone — from your C-suite to your daily users — is measured based on the overall success of the technology, not just their departmental metrics. Be sure to understand the complete picture — including the baseline metrics, measurement methodology, and scale of implementation.

Understanding the AI Learning Curve

Working with true AI-powered solutions also requires a fundamental mindset shift. Unlike traditional software where "what you see is what you get," AI solutions improve over time through learning and optimization. Initial performance may not reflect the long-term potential, making patience and appropriate performance measurement crucial.

This learning curve applies not just to the technology but to the organization implementing it. Teams need time to adapt to new workflows and capabilities, making change management an essential component of successful implementation.

Artificial Intelligence in Retail: Building Lasting Success

As retail continues to innovate, technology will only grow in importance. Success requires moving beyond surface-level evaluations to truly understand how solutions will scale, integrate, and deliver value within your specific context.

Remember that the goal isn't to implement technology for technology's sake, but to solve real business challenges and create meaningful customer experiences. By maintaining this focus and applying a cohesive evaluation criteria, retailers can make technology investments that truly drive business transformation.

The future of retail technology is bright, but realizing its potential requires careful consideration, clear alignment, and a commitment to looking beyond the hype to find solutions that deliver genuine value. The most successful implementations will come from partnerships where both vendor and retailer are aligned in their commitment to driving meaningful business outcomes.

Author Bio: Michelle Bacharach is the CEO and Founder of FindMine, a retail technology company that inspires shoppers throughout their journey with on-brand, dynamic, and inventory-aware outfitting & styling. The world’s leading brands like Lululemon, Gap, and Anine Bing trust FindMine’s AI to amplify what their teams can do alone, increasing revenue and improving margins while creating loyalty that lasts.