Marketers have spent years obsessing over ranking reports wherever their products appear. A higher position meant more visibility, more clicks, and more revenue. That mental model made sense when everyone saw more or less the same results for the same query.
That world is disappearing. Between geography, purchase history, inventory, and platform-specific algorithms, two people can type the same query and get completely different top results.
Personalization is no longer a useful level of recommendation to have alongside search. It’s built into how products are discovered, making traditional rankings a directional measure of performance rather than a definitive measure.
Consider a simple, high-level question: “What are the most comfortable slippers?”
On Amazon, this question no longer refers to a universal shelf. Tools like Alexa for Shopping reorder and reshape results based on what the platform already knows about each shopper, including price sensitivity, past purchases, brand preferences, and even which products they’re most likely to keep.
Here is a possible trip:
- A value-focused shopper who historically purchased basics under $20 sees mass-market slippers at lower prices, with budget brands taking the top spots.
- A premium shopper who regularly buys high-end clothing sees wool, shearling and specialty brands priced above $100.
They both used the same words. Neither saw the same ranking.
The “most comfortable slippers” are not a single list. They are a customized set of candidates that flex around the buyer on the screen. As this model spreads across retailers and platforms, it weakens the idea of a single canonical position to optimize for.
Your customers search everywhere. Make sure your branding introduces himself.
The SEO toolkit you know, plus the AI visibility data you need.
Start your free trial
Start with
Why rankings are misleading in a personalized system
Most ranking tracking still assumes a stable baseline: pick a keyword, capture the best results from a location and device, and treat it as truth.
Personalization breaks this in several ways:
- The position moves the shelf: Local inventory, regional preferences and market-by-market offerings change, which products appear and in what order.
- History shapes relevance: Clicks, purchases and dwell time fuel future recommendations. Two shoppers with different stories effectively form two different sets of outcomes, especially among retailers.
- The platform logic diverges: Even within a company’s ecosystem, different surfaces foster different domains, formats or signals. Google’s AI Mode, AI and Gemini overviews vary significantly based on sources cited and frequency, according to Tinuiti’s AI Citation Trends research. (Disclosure: I’m the vice president for commercial media at Tinuiti.)
Layer in conversational AI, including Google’s AI overviews and AI mode, ChatGPT, and Alexa for Shopping, and the gaps widen. These interfaces summarize, personalize, and refine responses over the course of a dialogue, not a single query.
A snapshot of the “average location” from a geographic area on one device isn’t enough to describe what real shoppers see. It can look reassuring on a dashboard while still being out of sync with actual exposure in the wild.
Personalization is now the engine of discovery
Discovery no longer just happens on a static list of blue links. People search for information and products on TikTok, Reddit, AI overviews, reseller agents and LLM chat.
Much of this activity never shows up in traditional SEO reports:
- AI Recaps answer the question directly, often combining products, reviews and third-party comments.
- Retail and market research adjusts results in real time based on behavior, context and inventory, alongside on-site agents within walled gardens like Walmart’s Sparky and Target’s AI Shopping Assistant.
- Social and community content increasingly appears as sources cited in AI responses, determining which brands are recommended.
Personalization ties all of this together from the user’s perspective. To the buyer, it just seems like it’s about better results. For marketers, this creates a measurement problem: If everyone’s experience looks different, which position are you actually tracking?
From position to visibility and share of voice
Given all of this, “What is our average ranking?” is the wrong question. A better question is, “How visible are we in the many personalized journeys our customers actually take?”
For example, from a research perspective, our work with Profound uses AI visibility rate as its primary metric. Instead of looking at a single position for a single keyword, AI viewability rate measures how often your brand appears in AI-powered responses across a broad set of requests.
In practice, this means:
- Monitor whether your brand appears when shoppers ask about your category, not just when they search for your name.
- Gauge whether you’ll appear as a top recommendation with context, price, or pros and cons, versus a brief mention buried in a longer list.
- See how visibility changes over time based on category, audience and platform.
This is essentially synthetic share of voice for AI and personalized search – a view of how much response space you have across many scenarios, rather than a single best position.
Quote sharing: How platforms decide who to show
Visibility isn’t just about being listed. It’s also about who the system trusts enough to reference as a source. This is where sharing quotes comes in.
Citation share measures how often domains you own are cited in AI responses.
Quotes act as a signal of trust
- Indicates that your content helped shape the response the user sees.
- It strengthens your authority over the model, increasing the likelihood that you will be recommended again in similar scenarios.
- Generate direct referral traffic from AI platforms that pass through the links.
The results also show how uneven this landscape already is. Social platforms, particularly Reddit, account for a notable share of citations across many categories, with some AI products drawing a double-digit percentage of their sources from Reddit alone.
For e-commerce, Amazon remains one of the most cited domains in business inquiries on average, despite actively limiting some AI crawlers, while other retailers, including Walmart, Best Buy, Ulta and Home Depot, are leaders in specific verticals and platforms.
These models demonstrate how heavily AI systems rely on certain ecosystems. If your content and products aren’t featured in places they trust, your visibility will lag, regardless of what your old ranking report says.
What the future dashboard should look like
The teams that adapt fastest create reports that reflect how personalized search actually works. This often includes:
- AI visibility rate/share of voice: Frequency and visibility of your brand across a defined set of suggestions and platforms relevant to the category.
- Share of citations (owned and third-party): The frequency with which your domains and the top third-party sites that mention you serve as sources in the AI’s responses.
- Segmented visibility: Breakdowns by vertical, product line, and audience segment so you can see where personalization helps or hurts.
- Links to performances: Views that link visibility and mentions to downstream metrics like conversion rate, revenue, and incremental lift, based on first-party data.
Traditional rankings do not disappear completely, but move from the main title to a supporting role. The title is visibility through thousands of personalized experiences, linked to real business results.
How to get started
If your reporting and planning cycles still revolve around static rankings, a few practical steps can help you shift to a visibility-centric view:
- Check where you really show up: Use tools like Profound to understand how often your brand appears and is mentioned in AI overviews, AI mode, ChatGPT, and top retail search experiences in your category.
- Reformulate your KPIs: Increase AI visibility rate, citation share, and share of voice at the category level alongside (and not instead of) traditional metrics, so teams start thinking in terms of reach, not just position.
- Align content to real queries: Make sure your product detail pages (PDPs), FAQs, and category pages speak the language buyers actually use, including use cases and constraints, so personalization systems can match your products to the right people.
- See PDP updates recommended by Amazon’s AI: Non-media product titles are now limited to 75 characters, including spaces, branding, and style. Amazon will also add a new AI-powered “Highlight Items” section for mobile. Review and approve AI-recommended headlines and highlights in Catalog > Edit List > View Enhancements by the July 27 deadline.
Personalization is rewriting how research works, and measurement must reflect this reality. Moving from position to visibility keeps your reporting aligned with how customers actually discover your brand.
