Archive

News

Microsoft Advertising Launches Product Explorer via @sejournal, @brookeosmundson

Microsoft Advertising has launched Product Explorer, a new Merchant Center feature designed to help advertisers better understand product status and performance.The tool provides a searchable view of product catalogs. Advertisers can quickly see which products are serving, which have issues, and which are driving results.
Product Explorer is currently available to U.S. advertisers with fewer than 100,000 SKUs.
Product Explorer Brings Product-Level Visibility To Merchant Center
Microsoft Advertising Ads Liaison Navah Hopkins announced the feature on LinkedIn.
According to Hopkins, Product Explorer was developed in response to advertiser feedback around feed management and product visibility.
Hopkins provided Search Engine Journal with a direct quote, stating:
We heard industry feedback that it was difficult to keep tabs on and manage feeds in Microsoft. With Product explorer, you can easily search for and understand which products are rejected, performing and which ones need optimization. This means less time manually hunting through reports, and more time making meaningful changes to your feed to ensure you’re reaching your desired outcomes.
The new tool helps advertisers identify products that are serving, rejected, or limited by feed issues.
It also connects to Microsoft’s Recommended Actions functionality. This gives advertisers guidance on how to resolve issues and improve product eligibility.
Search, Filter, And Export Product Data
Product Explorer includes filtering across feed attributes and performance metrics.
Advertisers can filter products using fields such as:

Title
Product ID
Brand
GTIN
Product Type
Custom Labels
and more.

Performance filters include:

Impressions
Clicks
Conversions
Spend
CTR
Conversion Rate.

Image credit: Microsoft Ads, June 2026
Advertisers can also combine feed attributes with performance data.
For example, they can identify products with low impressions inside a specific product category. They can also review performance across custom label groups.
Filtered product lists can be exported for offline analysis.
According to Hopkins, one use case is identifying products that are not serving. Advertisers can quickly find products with little or no visibility and investigate potential causes.
The report may also help advertisers evaluate feed taxonomy decisions.
Product types, categories, and custom labels are often used to organize campaigns. Product Explorer makes it easier to review how those classifications are performing.
The tool also provides a faster way to analyze product-level performance. High-performing and underperforming products can be identified without building separate reports.
Why This Matters For Advertisers
Product feeds continue to play an important role in Shopping campaign performance.
Feed quality can influence visibility, query matching, and overall campaign results.
Historically, troubleshooting feed issues was not always simple.
Advertisers often had to move between Merchant Center diagnostics, feed management tools, and campaign reports. Product Explorer brings much of that information into one location.
For advertisers managing large catalogs, the biggest benefit may be efficiency.
The tool makes it easier to identify rejected products. It also helps advertisers spot products that are not generating impressions or conversions.
That visibility can help teams prioritize feed updates and optimization efforts.
The addition of product-level performance filters may also help advertisers uncover trends that would otherwise be hidden inside campaign-level reporting.
What Comes Next
Product Explorer addresses a challenge many Microsoft advertisers have faced for years.
Understanding feed health and product performance often required multiple reports and workflows.
This update brings those insights together in a single interface.
The initial rollout is limited to U.S. advertisers with fewer than 100,000 SKUs. Microsoft says it is actively collecting feedback as it considers future improvements and expansion.
For ecommerce advertisers, Product Explorer provides a more direct way to monitor feed health and product performance. It may also make routine feed audits faster and easier to manage.
Featured image: Samuel Boivin / Shutterstock

Read More »
June 15, 2026 No Comments
digital marketing

What Matters In An AI Prompt? Intent or Keywords? via @sejournal, @maltelandwehr

This post was sponsored by Peec AI. The opinions expressed in this article are the sponsor’s own.Which prompts should I prioritize tracking for AI visibility?
Does exact wording change which brands AI engines recommend?
Do I need to track every way someone might phrase a prompt in AI search?
Marketers often panic about the infinite ways users might phrase questions to AI engines. But a recent study from Peec AI reveals a much more predictable reality.
In This SEO Guide1. How Prompt Wording Impacts AI Brand Visibility2. Methodology: How We Tested This3. Insight 1: Human Prompts Only Look Different On The Surface (Mostly)4. Insight 2: Changes in Wording Only Impacts Brand Mentions Past a Threshold5. Insight 3: Prompt Style Influences Brand Visibility6. Insight 4: Middle-Of-Funnel Prompts Are Where Wording Actually Decides Winners7. Insight 5: Answer Engines Don’t Behave The Same Way8. The Takeaway: 6-Step Measurement Playbook9. What This Study Doesn’t Prove10. How To Track AI Prompts Without Chasing Every Variation
How Prompt Wording Impacts AI Brand Visibility

Variation is limited, not chaotic: users phrase things differently. But over 90% of those variations have very similar meaning.
Wording matters less than intent: you don’t need to worry about the exact words used. Brand mentions hold steady as long as the core intention stays the same.
Style matters as much as meaning: concise keywords or “list” requests prompted the AI to surface up to 20% more brands in its answers compared to open-ended prompts.
Wording Variation Hits Hardest in the Middle-of-Funnel: top- and bottom-of-funnel queries are relatively stable against phrasing tweaks. Unbranded, commercial middle-of-funnel discovery is less. Because wording variation dictates winners here, capturing reality requires absolute phrasing precision and potentially a larger share of your tracking volume.

Two people can ask an AI the exact same commercial question using completely different words.
One asks for the “best noise-cancelling headphones under $200.” Another asks, “Which budget over-ear headphones have good noise reduction?” The wording changes. The underlying need mostly does not.
This distinction matters for AI brand visibility. On the surface, user phrasing looks chaotic. Under the surface, these questions are close in meaning – until they drift just far enough to trigger a completely different set of brands.
To find that breaking point, Peec AI analyzed 1,754 prompts, 37,804 AI responses, five sectors, and 18 sub verticals across ChatGPT, Gemini, Perplexity, Google AI Mode, and Google AI Overviews.
Methodology: How We Tested This
If your tracking tool says you show up for a specific query, does that visibility hold up when a real user types a variation with the exact same intent?To measure this drop-off, we ran two parallel studies.

Study A: 288 human-written prompts from Rand Fishkin’s followers for two different intents, resulting in 17k+ chats. The authors thank Rand for making the dataset available to us.
Study B: 54 base prompts from 18 different verticals. For each we generated dozens of variations in tiny cosine-similarity steps, resulting in 1k+ total prompts and 20k+ chats.

Image created by Peec.AI, June 2026
Study A gives us a glimpse into how varied the prompting style of humans is. Study B allows us to observe the impact of tiny changes in prompts.
In study A we analyzed the difference between every pair of prompts (within each intent). In study B we analyzed the difference introduced by every small step (within each industry and intent).
Please note: we ran every prompt multiple times to account for the inherent variance of LLM responses.
Image created by Peec.AI, June 2026
Why Tracking Keywords Misses How People Actually Prompt
In AI search, exact keyword matching only plays a minor role. “CRM software” and “customer relationship management tool” share almost no characters but point at the same goal.
To measure this, we converted every prompt into a semantic embedding. We quantified the semantic distance using cosine similarity, which evaluates meaning rather than raw text length. Applying this to the human-written prompts yielded a precise similarity value between 0 and 1.
Image created by Peec.AI, June 2026
Instead of guessing how different two prompts are, we can quantify the semantic distance.
Insight 1: Human Prompts Only Look Different On The Surface (Mostly)
We used two different embedding models on the 288 human-written prompts (all-MiniLM-L6-v2 and all-mpnet-base-v2). Both showed the exact same pattern: most human prompts clustered tightly with high cosine similarity. People use different words to express the exact same intent. The percentage of prompts showing large semantic drift was surprisingly small – accounting for less than 10% of the variations.
Image created by Peec.AI, June 2026

~88% to 92% of human prompt pairs sat above a cosine similarity of 0.50.
~95% sat above 0.40.

The takeaway: People phrase the same commercial need in many different ways. But mathematically, most of those phrasings end up being fundamentally similar.
Insight 2: Changes in Wording Only Impacts Brand Mentions Past a Threshold
In study A we took all the brands mentioned during all the runs of the base prompt. We then observed how the average visibility of all these prompts changes when changing the prompt in tiny steps.
Against a near-identical reference group, the average probability of a brand being mentioned across our dataset was 4.9%. However, when prompts drifted into the lowest similarity bin (0.35 to 0.39), visibility dropped by 2.40 percentage points (pp) – a roughly 50% relative decrease.
Image created by Peec.AI, June 2026
That is a massive drop, but notice where it lives: entirely in the left tail.
As long as prompts stayed above 0.50 to 0.60 cosine similarity, depending on the AI Engine, brand visibility remained stable. While AI outputs inherently fluctuate, the largest wording-driven visibility losses only happen when a prompt’s core meaning drifts significantly. Because most humans naturally type well above that threshold, prompt tracking exposure to this risk is narrower than it seems.
The takeaway: Prompts with the same intent and same semantic characteristics largely lead to mentions of the same brands at the same frequency.
Beware Of The Semantic Blind Spot!
High similarity doesn’t equal matching intent. “Car rental Charleston” and “Car rental Charlestown” are 95% similar but serve entirely different commercial goals. If a core qualifier changes, treat it as a new intent. Typical qualifiers are locations, products, demographics, and brands.
For larger prompt sets, use an LLM-as-a-judge to check for these shifts automatically.
Insight 3: Prompt Style Influences Brand Visibility
Image created by Peec.AI, June 2026
What you prompt is only half the equation. How you prompt – the style, not just the intent – changes what the AI surfaces.

Format matters. Asking for a comparison, table, list, or ranking consistently surfaces more brands than open-ended questions. A ranking prompt leads to significantly more brand mentions in the answer (+20% average visibility).
Keywords beat conversations. Despite AI’s conversational interface, concise, keyword-style prompts (e.g., “best CRM small business 2026”) lead to more brand mentions (up to +25% average visibility). Keyword prompts preserve a sharp commercial retrieval anchor, whereas persona-engineered prompts (“You are an IT consultant…”) often broaden the query into educational paths that are less brand-dense.
Answer engines react differently to constraints. Adding budget or feature constraints leads to different outcomes depending on the model. In ChatGPT and Perplexity, constraints reduce the number of brands shown. In Gemini and Google AI Overviews, constraints actually increased the number of brands. Potentially by triggering additional fanout queries.
Length doesn’t matter. Typing more filler or conversational words has effectively zero impact on which brands are shown in the answer.

The takeaway: If you mix these styles in your prompt tracking, you should tag them by format.
Insight 4: Middle-Of-Funnel Prompts Are Where Wording Actually Decides Winners
Prompt wording doesn’t matter equally across the buyer journey (and which prompts you choose to track matters more than their exact phrasing):

Top-of-funnel (Low Sensitivity): Broad category questions like “What is a CRM?” are highly stable. Small phrasing differences rarely alter which brands appear.
Middle-of-funnel (High Sensitivity): Unbranded commercial queries (“best CRMs for a small remote team“) are highly sensitive to small details. We can observe significant changes of mentioned brands already in the 0.60 to 0.65 similarity bucket.
Bottom-of-funnel (False Stability): BOFU prompts are often branded. Their stability towards wording changes is probably a result of everything being anchored around the brand or product name(s).

The takeaway: To capture the full picture you should track more variations of your MOFU prompts. For TOFU and BOFU fewer prompts are enough. In practice that could mean 25% TOFU, 50% MOFU, and 25% BOFU.
Insight 5: Answer Engines Don’t Behave The Same Way
While the wording effect’s direction is consistent across all engines, the severity differs:

Gemini: The effect fades fastest, concentrated in the lowest similarity buckets.
Google AI Overviews: Show the most persistent middle-of-funnel sensitivity. Small wording changes impact visibility much more than in any other engine.
ChatGPT, Perplexity, & Google AI Mode: Visibility penalties span a wider range of variations. On ChatGPT, middle-of-funnel brand loss triggers the moment phrasing slips below the 0.60 to 0.64 bucket

The takeaway: Treat carefully when aggregating data across models.
The Takeaway: 6-Step Measurement Playbook

Segment by funnel stage early. Top-of-funnel queries provide a stable baseline for category awareness, and bottom-of-funnel prompts monitor branded retrieval environments. However, because wording variation actively dictates the winners in the commercial middle-of-funnel, capturing reality there requires absolute phrasing precision and a larger share of your tracking volume
Anchor on your buyer’s actual phrasing. There is no universally “perfect” base prompt. The right anchor matches your target intent and persona. Do a quick reality check: ask a few colleagues how they would naturally type that exact query. If their answers risk dropping below the crucial 0.50 similarity threshold, your phrasing is too narrow and you need to track an additional anchor.
Don’t mix prompt styles. Format, archetype, and constraint levels each shift the baseline – a list prompt and an open-ended prompt do not share the same starting line. Tag your prompts by format so you can compare apples to apples
Watch constraint details in the middle-of-funnel. Without a brand anchor, minor constraint shifts – adding an integration, team size, or budget limit – can completely change which brands surface. Track multiple prompts that capture these nuances within the same persona.
Don’t track the left tail. Human variation clusters naturally, and visibility only drops sharply when prompts drift into the 0.40 to 0.50 similarity range. Focus your tracking budget on the dense semantic middle where most real buyers actually type.
Report each AI engine separately. Get the per-engine picture before creating any blended views. That’s how you tell whether a visibility change is a broad market shift or an algorithm quirk in one system.

What This Study Doesn’t Prove
These patterns were consistent across 37,804 AI responses. But keep these caveats in mind:

Trends are not guaranteed. These percentages reflect the strong patterns we observed. They are not static rules for every query.
Regulated industries may vary. We tested 18 subverticals. It is possible that regulated categories like healthcare behave differently due to stricter AI safety guardrails.
Engines constantly change. The exact percentages will shift as models evolve or grounding systems change. Only the core mechanics (wording threshold, middle-of-funnel sensitivity, and style baselines) will remain.

How To Track AI Prompts Without Chasing Every Variation
If you are hesitant to track prompts because “every prompt is unique” and “you do not know how exactly your audience is typing”, you can relax. The wording space isn’t a flat, chaotic spread of random variations; it has shape and structure.
There is no need to monitor every single phrase or chase an endless list of variations. You only need to know the intent and the relevant contexts you want to monitor. Look at the true meaning, separate the style, segment by funnel stage, and read the AI engines one by one.

Try Peec AI

Image Credits
Featured Image: Image by Peec AI Used with permission.
In-Post Images: Images by Peec AI Used with permission.

Read More »
June 15, 2026 No Comments
Generative AI

Google’s Mueller Says llms.txt Can’t Help LLMs Differentiate Sites via @sejournal, @MattGSouthern

Google’s John Mueller argued that LLM systems can’t use files like llms.txt to decide which websites to surface for a given query.He made the comments on a recent episode of Search Off the Record, the podcast from Google’s Search Relations team.
His comment points to a broader signal problem, not just intentional gaming. Even a well-written llms.txt file is still self-reported information from the site that wants to be chosen.
For discovery, Mueller pointed back to normal HTML pages and internal links.
What Mueller Said
The conversation started with a question about whether publishers should convert websites to Markdown for LLMs. Mueller and co-host Martin Splitt agreed that HTML is still the foundation for crawling and discovery.
The discussion got specific when Mueller turned to llms.txt. He described the discovery use case as a dead end:
“It’s basically you’re telling these systems, like, I have the best website ever. And here are all of the pages that everyone must go to. And you must buy all of my products or whatever you put in there. So in LLM system, it basically, by design, can’t trust what is here as a way of differentiating between different websites.”
His argument comes down to differentiating. If sites use llms.txt to promote themselves, the files can make similar claims. An LLM deciding which site best answers a query still needs another way to differentiate between them.
What ‘By Design’ Might Mean
“By design” could mean two different things, and Mueller didn’t clarify which.
One reading is architectural. LLM systems evaluate web content and can’t use self-reported files when picking sources.
The other reading treats it as a signal problem. Self-reported signals lose value when everyone provides them. Meta keywords stopped working for the same reason. Every site stuffed them, and search engines couldn’t extract a useful ranking signal.
Both readings reach the same conclusion on discovery. But they imply different things about whether the limitation could change over time.
Where Mueller Sees A Role
Mueller didn’t reject all uses of llms.txt. He carved out one case where it could help:
“If someone is already on your website, maybe some kind of automated system is helpful.”
He used the example of an agent trying to buy a photograph from a specific site. The LLM would visit the site and look for instructions on how to complete the purchase.
The argument splits discovery from navigation. llms.txt can’t help an LLM choose which site to visit. But it could help once the agent is already there, like a store directory for someone who already walked in.
Beyond The Gaming Argument
Mueller has called building Markdown pages for bots “a stupid idea”. He’s also compared llms.txt to the keywords meta tag.
SEJ’s Roger Montti wrote that llms.txt is “inherently untrustworthy” because nothing stops site owners from adding self-serving content. SE Ranking’s analysis of 300,000 domains found no link between llms.txt adoption and citation frequency in LLM answers.
Those arguments focused on what happens when people game the files. Mueller’s podcast comment adds the nuance that there’s no mechanism within the files to help an LLM pick one site over another.
Why This Matters
The gaming argument against llms.txt has always had a counterargument available. Platforms could learn to penalize manipulation, the way search engines handled spammy structured data.
The differentiation argument leaves a harder problem. Penalizing manipulation may address abuse, but it doesn’t explain how self-reported files help an LLM choose one site over another. Your most accurate llms.txt file still can’t tell an LLM to pick your site over a competitor’s.
Looking Ahead
Standards for how agents navigate sites haven’t settled yet, Mueller acknowledged. He mentioned WebMCP alongside other file types under discussion.
None have become a standard. By his estimate, it could take six months to a year, or longer, for agentic systems to settle on a format. The discovery layer, where HTML and internal linking already work, isn’t part of that discussion.

Read More »
June 15, 2026 No Comments
News

Google Ads Bidding Changes: What PPC Managers Need To Know About The 3 Updates via @sejournal, @brookeosmundson

Google Ads Liaison Ginny Marvin announced three bidding and budgeting updates on LinkedIn, including one change scheduled to begin rolling out on August 17.Two of the updates expand capabilities that were previously limited or unavailable to many advertisers. Smart Bidding Exploration is now available globally, while Promotion mode is entering beta for Search and Performance Max campaigns.
The third update affects how Google optimizes campaigns that are limited by budget. While Google expects only minor fluctuations during rollout, advertisers may notice temporary performance changes as campaigns recalibrate.
Here’s a closer look at what’s changing and what it could mean for advertisers.
Smart Bidding Exploration Goes Global
Smart Bidding Exploration (SBE) is designed to help campaigns find additional converting traffic beyond the queries they would normally pursue under existing bidding targets.
Google first introduced Smart Bidding Exploration ahead of Google Marketing Live 2025 as a way to help advertisers uncover additional conversion opportunities without significantly loosening bidding targets.
Marvin announced that SBE is now available globally across all languages for Search campaigns and Performance Max campaigns without a product feed.
Google is also opening a beta for Shopping advertisers, including both standard Shopping campaigns and Performance Max campaigns that use product feeds.
One reason the feature has generated interest is that it does not require advertisers to significantly loosen ROAS targets in order to pursue additional reach. Instead, Google attempts to identify incremental conversion opportunities while continuing to optimize toward existing campaign goals.
For advertisers that feel constrained by volume, this may provide another option to test growth without making major changes to campaign structure or bidding strategy.
Promotion Mode Enters Beta For Search And PMax
Google is also introducing Promotion mode in beta for Search and Performance Max campaigns.
The feature allows advertisers to temporarily increase budget flexibility and adjust ROAS tolerance around specific events such as product launches, seasonal promotions, or flash sales.
According to Marvin, Promotion mode can also be used alongside campaign total budgets.
Historically, advertisers often had to manually adjust budgets and bidding targets around promotional periods. Promotion mode appears intended to automate some of that process.
For advertisers that regularly make manual bidding adjustments around promotional periods, the feature could simplify some of that planning and give Google additional flexibility during short-term demand spikes.
Google has not yet provided details about beta eligibility or rollout availability. Advertisers should check their accounts before building campaign plans around the feature.
How Google Is Changing Budget-Limited Campaign Optimization
The third update is the one most likely to affect reporting.
Starting August 17, Google is making backend bidding target optimization changes aimed at budget-limited campaigns.
Per Marvin’s LinkedIn post, she stated:
Starting August 17, we’re making backend bidding target optimization updates to help campaigns limited by budget see more predictable performance in line with CPA and ROAS targets, especially when budgets increase.
Marvin added that when this goes into effect, Google expects a brief calibration period during which some advertisers will see minor performance fluctuations.
Google did not provide details on how long calibration may last or how significant the changes could be. 
Based on Google’s description, the goal is to reduce some of the volatility that can occur when budget-constrained campaigns receive additional budget while continuing to optimize toward CPA and ROAS targets.
To give advertisers lead time, Google will begin showing notifications in Google Ads accounts starting July 6. Those notifications will include historical campaign performance data and recommendations related to the upcoming changes.
Marvin also noted that advertisers may need to adjust CPA or ROAS targets to ensure they accurately reflect business goals before the rollout begins.
What This Means For Advertisers
While all three updates focus on bidding and budgeting, they address different challenges.
Smart Bidding Exploration is aimed at advertisers looking for additional volume without making major changes to existing bidding strategies. The Shopping beta will likely attract attention from advertisers that have been looking for more ways to expand reach beyond current query coverage.
Promotion mode is focused on a different problem. Many advertisers adjust budgets and bidding targets manually around launches, seasonal promotions, and peak demand periods. If the feature performs as advertised, it could reduce some of that management overhead.
The August 17 optimization update stands apart because advertisers do not need to adopt a new feature for it to affect campaign behavior.
That makes the July 6 account notifications particularly important for teams managing budget-limited campaigns. Google’s recommendation to review CPA and ROAS targets suggests that some advertisers may discover their existing targets no longer reflect current business conditions or business goals.
For agencies, this may also be a good opportunity to proactively discuss the upcoming change with clients before the rollout begins.
What Advertisers Should Do
Smart Bidding Exploration and Promotion mode are both optional features. In my opinion, the August 17 rollout deserves the most attention because it affects campaign behavior whether advertisers adopt the new features or not.
Here are a few areas worth reviewing:

Review the July 6 account notifications and historical performance data when they become available.
Revisit CPA and ROAS targets to confirm they still align with current business goals.
Identify campaigns that regularly operate under budget constraints and monitor them closely during the rollout period.
Evaluate Smart Bidding Exploration and Promotion mode if they become available in your account and align with your campaign objectives.

Most advertisers will likely spend more time monitoring this update than making major account changes. However, campaigns that frequently hit budget limits deserve a closer review before August 17.
Final Takeaways
The Smart Bidding Exploration expansion and Promotion mode beta give advertisers additional tools to test.
The August 17 rollout is different because it affects how Google handles optimization for budget-limited campaigns behind the scenes.
Google is providing advance notice through July 6 account notifications, giving advertisers time to review existing CPA and ROAS targets before the change takes effect.
For most accounts, the update will likely be something to monitor rather than something that requires immediate action. Still, any campaign that regularly operates under budget constraints deserves a closer look before August arrives.
Featured image: Prostock-studio / Shutterstock

Read More »
June 15, 2026 No Comments
Local Search

The Review Gap: Finding Client Opportunities In Competitor Feedback

SEOs all know how important reviews are as a local SEO ranking factor and decision maker for users. But how many SEOs are actually using the review content to help with their roadmapping and content updates?Reviews are typically looked at as a reputation task, and the focus is on the quantitative data (number of reviews, star rating, review velocity). The work that’s done with reviews is more reactive, where we make sure reviews are responded to, or we notice that reviews are missing, so we figure out what happened there. While all of that is important, SEOs often forget they are sitting on a goldmine of information that comes directly from users: the review text.
Reviews are where customers who felt very strongly one way or another left their feedback and experience out for the business and other potential customers to see. It happens with our clients, and it happens with our competitors. 
Why Competitor Reviews Are The Data You’re Missing
Google Business Profile reviews are essentially a free, always-updating focus group. The real opportunity is knowing why your client’s top competitor has 56 one-star reviews about pricing opacity. It’s an opportunity to turn that into a conversion lever.
Here’s what competitor review analysis surfaces:

Customer language: The exact phrases real customers use to describe their problems. These complaints are a positioning opportunity for your client.
Service delivery failures: No-shows, communication gaps, pricing surprises, rushed jobs. This is a public record of what frustrated customers wish someone else had offered them.
Trust gaps competitors haven’t addressed: The anxieties showing up in reviews that aren’t being answered anywhere in a competitor’s messaging.
What “good” actually means in that market: What customers praise tells you the standard they’re measuring against.

The Framework
The framework is straightforward: Export competitor reviews → Analyze sentiment → Cluster.
Use competitor shortfalls to your advantage by highlighting the things your client does well in that area.
But why should you do this? AI systems in local SEO can summarize based partly on the specific language in their GBP reviews and business descriptions. Think of Ask Maps. Ask Maps about this place, and know before you go, all of these new AI features on Google Maps pull from review text. Review patterns shape how these AI features characterize a business.
We’ll go through how to get started with this framework.
Step 1: Pick The Right Competitors
Don’t pull every business in the local pack. You want the two to three competitors your client is actually losing jobs to, the ones showing up consistently for your client’s core services/products.
The easiest way to identify them: Run your client’s three to four highest-value searches in Google Maps and note which names keep appearing. Check with your client, too. They usually know exactly who they lose bids to.
Step 2: Export Reviews
Once you’ve identified your targets, decide how you want to pull the data. You can definitely vibe code your own tools to pull competitor reviews if you’d like. Or you can use the GBP Reviews Sentiment Analyzer Chrome extension (full disclosure: I built this). Or any other tool that will allow you to pull competitor reviews.
Step 3: Run Sentiment Analysis
No matter how you grab the reviews, you’ll want to use AI to help you run the sentiment analysis on them. This will help you categorize reviews into positive, negative, and neutral buckets, which makes it easier to filter through in sheets.
Screenshot by author, June 2026
You can approach running the sentiment analysis in many ways. One would be using the Google Cloud Natural Language API if you’re comfortable working with APIs to set it up or you can use a custom GPT to help you out.
(A note on privacy: You’re working with publicly available reviews only. So the typical privacy concerns of giving LLMs access to proprietary data should not apply here.)
If you use the Chrome extension, the sentiment analysis is run during your data pull and is part of the XLS export. If you prefer starting from scratch and just running prompts in your LLM of choice, you can get started with this:

I’m going to paste a CSV of Google reviews for [Competitor Name], a [business type] in [city].

Please:

Identify the top 5-7 recurring themes (both positive and negative)
Count how many reviews mention each theme
Flag any patterns in the negative reviews that suggest operational failures or unmet customer needs
Pull 3-5 direct quotes that best represent each theme
Summarize the biggest gap between what customers praise and what they complain about

Here is the data: [paste CSV]

Adjust as needed for your client’s situation, but the core task stays the same: themes, counts, language, gaps.
Step 4: Build Your Topic Cluster Map
Once you have the analysis output, organize recurring themes into clusters. It can be based on the following credibility factors:

Quality (workmanship, results, expertise).
Communication (responsiveness, updates, follow-through).
Pricing (transparency, value, billing surprises).
Speed (arrival times, turnaround, scheduling).
Trust (reliability, honesty, doing what they said).
Staff/Team (professionalism, friendliness, knowledge).

The gap between “what customers love about my client” and “what customers hate about competitors” is where the real opportunity lives.
What To Look For In The Data
Having the data is one thing, but knowing how to read it is another.
Start with review velocity and volume. A competitor with 129 reviews at 5.3 reviews per week tells a completely different story than one with 28 reviews at 0.9 per week, and that’s before you’ve read a single word. Velocity signals active, ongoing trust-building. Volume signals a business customers feel compelled to talk about.
When sentiment scores are close between two competitors, differentiation has to come from messaging specificity, not star ratings. A 4.7 vs. a 4.8 isn’t a meaningful difference to a customer. The words you use to describe what you do, and whether those words reflect what customers actually care about, that’s the difference.
Ask these four questions of every competitive review set:

What do customers consistently praise about this competitor that your client also does well, but doesn’t say anywhere in their messaging?
Where do customers express frustration that your client’s operations genuinely solve?
What language do reviewers use that your client’s website doesn’t reflect at all?
What’s the underlying fear or desire behind the complaint?

That last one is the most important. Negative reviews are a map of customer anxieties in that category.

“They overcharged me”: fear of being taken advantage of.
“They said they’d come between 9 and 11. They showed up at 3.”: fear of wasted time.
“I had to chase them for updates”: fear of being ignored or dismissed after signing.

Each of those anxieties is a conversion lever, if your client genuinely resolves it and their messaging says so directly.
Turning The Gaps Into Real Deliverables
Here’s how to translate what you found into actual client work.
USP Extraction
The language customers use to praise your client is the raw material for H1s, meta descriptions, GBP descriptions, and homepage hero copy. Language real customers used, unprompted, to describe their experience.
Competitor Gap Messaging
For every recurring competitor complaint, write a direct-response positioning statement that’s a clear, specific answer to the anxiety.

Competitor complaint pattern
Direct positioning response

“They never gave me a price up front.”
“Upfront pricing on every job – no surprises on your invoice.”

“They said they’d be here at 9. They came at 3.”
“Exact arrival windows, not four-hour guessing games.”

“The work looked rushed, and they just left.”
“We don’t leave until the job is done and you’re satisfied.”

Website Copy And Structure Updates
Once you have your topic clusters and gap analysis, you have a clear brief for website copy changes:

H2 variants: Grounded in top review clusters, run some SEO A/B tests and see how they affect user behavior data and conversions.
Testimonial selection: Don’t just pick the most enthusiastic reviews and start picking the ones that speak directly to the gaps competitors are failing on.
FAQ content: Proactively neutralize the anxieties surfaced in competitor negatives. If 200 reviews across your competitors mention pricing surprises, your FAQ should include “How is pricing determined?” before a customer even has to ask.

GBP Profile Updates
Your client’s GBP description, posts, and services list are all conversion touchpoints, and they can all be updated to reflect what you’ve learned:

Description: Pull language directly from top positive review clusters, the words real customers used.
Posts: Feature the specific trust signals competitors are consistently failing on. If competitors have a communication problem, post about your client’s same-day callback guarantee.

Content Series Opportunities
Review clusters often point directly to content gaps. If tons of reviews across your analysis mention customers feeling confused about the process, that’s a “What to Expect” video and informational page creation waiting to happen. If “explained everything clearly” shows up repeatedly as praise, that’s a signal that the category has a clarity problem, and your client can own it.
Measuring What Changes
You can measure the impact of your changes in a few ways:

If you run an H2 SEO A/B test, consider also tracking scroll depth past the hero section, CTA click rate, and dead clicks before and after you swap in review-language-based copy.
If you update your client’s GBP, track call volume, direction requests, and website clicks in your GBP Insights dashboard before and after profile changes.
For new content changes, track organic visibility for the informational queries tied to the review themes you called attention to.
You can also consider looking at AI citations and grounding queries in Bing’s AI Performance Dashboard to see if anything new appears after including the new language on your website and GBP

Re-run the analysis periodically. Are the same complaints showing up in competitor reviews, or have your client’s updates shifted how customers compare them? Are any new patterns emerging that you should address?
The Opportunity Most SEOs Are Leaving On The Table
Reviews are a research and strategy layer with a reputation management component.
The competitor who dominates local search isn’t necessarily the one with the most reviews or the highest rating. It’s the one whose messaging reflects what customers actually care about, the one who answers the anxiety before the customer even has to voice it.
You have free, public, always-updated customer research sitting inside every competitor’s GBP right now. It’s telling you exactly what customers in your client’s market are afraid of, what they value, and what language they use to describe the experience they’re looking for. That list is your client’s next positioning opportunity.
More Resources:

Featured Image: Master1305/Shutterstock

Read More »
June 15, 2026 No Comments
ecommerce

Agentic Commerce And The New Rules Of Google Ads via @sejournal, @tonyadam

Your Google Ads account is going to start serving a buyer you cannot see. That buyer is an AI agent that compares products and goes through checkout on your customer’s behalf.Agentic commerce is going to be the biggest structural change to ecommerce paid media since mobile, and it is already moving real money. During Cyber Week 2025, Salesforce attributed roughly $67 billion in global sales, about 20% of all orders, to AI and agents. That wasn’t an estimate or forecast for next year; that was last holiday season.
Your product feed is now turning into a bidding signal instead of a catalog, and a new paid surface opens inside Google’s AI Mode. Performance Max and Shopping start placing into that surface directly, and your conversion tracking breaks in ways that depend on how the agent checks out. This is going to be a bumpy ride.
None of this means your current campaigns stop working. It means the inputs that decide whether you win are shifting, and the accounts that adjust early get a real edge.
What’s The Latest In Agentic Commerce
A quick grounding, because this space moves fast and the headlines blur together.
Google’s agentic checkout, branded Buy for Me, is live in AI Mode with launch partners including Wayfair, Chewy, and Quince. At NRF 2026, Google introduced the Universal Commerce Protocol, an open standard built with Shopify, Etsy, Walmart, and Target, and endorsed by more than 20 companies, including Visa and American Express. OpenAI shipped Instant Checkout in ChatGPT on its own protocol built with Stripe. Perplexity paired with PayPal. Visa, Mastercard, and Stripe all rolled out agent-ready payment rails.
When discovery, checkout, and payments all reorganize around agents within 12 months, this is not a pilot you wait out.
→ Read more: Selling To AI: The Complete Guide To Agentic Commerce
Your Feed Is Now A Google Ads Bidding Signal
In Shopping and Performance Max, your product feed already drives matching and bidding. Agents push that further. When an AI agent evaluates products, it does not read your ad copy or your creative. It reads structured data, the price, availability, shipping, returns, and specs in your feed, and it decides whether you make the shortlist before a human sees anything.
OpenAI’s own evaluation of its shopping research tool makes the point. The tool hit 52% product accuracy on multi-constraint queries against 37% for standard ChatGPT search, where product accuracy measures how well results match requirements like price, color, material, and specs. Buyers are handing agents hard constraints, and the agent is matching those constraints against your feed fields.
Google has noticed where the lever sits. It released new Merchant Center attributes specifically to help products get surfaced in conversational shopping.
The takeaway for a paid team is uncomfortable but simple. Feed quality is now a bidding issue, not a hygiene issue. If your feed is owned by whoever set up Merchant Center two years ago, while your budget and attention go to creative, you have it backward for this surface. We treat the feed as a media asset now, with the same rigor we give a creative testing plan.
Direct Offers Is A New Google Ads Paid Surface
The part most paid media coverage has not caught up to is that agentic commerce arrived with an actual ad product.
Direct Offers is a Google Ads pilot that drops merchant-funded promotions directly into AI Mode when the system reads a shopper as high intent. You set the offers in your campaign settings, and Google decides when to surface them. Google’s own ads liaison described the format as less like a standard ad and more like a salesperson negotiating a deal on the shopper’s behalf.
Sit with what that means for a media buyer.
You are no longer only bidding for a placement. You are deciding how much margin you will give up at the exact moment of decision, inside an interface Google controls.
That cuts two ways. The risk is obvious. If discount depth is the only lever, this surface becomes a margin race, and the wrong brands win it. The opportunity is that Google has already said it will expand Direct Offers beyond price to value, naming loyalty benefits and product bundles. The brands that build a non-price offer strategy early get to compete on something other than how much they will bleed.
Decide your posture before you opt in. Which products, what margin floor, and whether you lead with price or with value.
PMax & Shopping Ads Now Place Into AI Mode
Here is the development that makes this concrete for anyone running Performance Max. As of February 2026, Google began serving shopping ads inside AI Mode, and those placements are served from your existing Shopping and Performance Max campaigns, marked as sponsored.
So your workhorse campaigns are already feeding the agent-mediated surface, whether or not you planned for it. The catch is visibility. More of the journey now happens inside AI Mode, where you see less of what is going on, and Performance Max was already the most opaque campaign type Google offers.
This is the same widening gap showing up with AI Max, where query expansion stretches the distance between what you bid on and what actually converts. Agents stretch it further.
The good news is that Google handed back real controls over the last year, so use them. Channel-level reporting shows where budget goes across Search, Shopping, YouTube, and the rest. Campaign-level negative keywords are no longer a support request. And search terms visibility in Performance Max finally approaches what Standard Shopping always gave you. If you are not using these to keep brand and non-brand legible, you are flying blinder than you need to be.
Agentic Checkout Breaks Tracking Two Ways
Your attribution was already imperfect. Agents break it in two specific ways, and which one hits you depends on how the buyer checks out.
The first path is Buy for Me, where the agent completes the purchase on your own site and you stay the merchant of record. Google’s documentation is clear that the transaction happens on your site, so your conversion tag can still fire. What breaks is the link back to the campaign that earned the sale, because the agent session does not carry an ad click through to checkout the way a normal visit does. You keep the conversion, but you lose the attribution. Better than losing both, I guess?
The second path is UCP-powered checkout, where the purchase happens directly on Google’s surface inside AI Mode or Gemini. You are still the merchant of record, so you still get the order, but the sale never happens in a browser session on your domain. That means your client-side tracking goes blind, your own pixel, and any Meta or other platform tags included, because there is no on-site event for them to catch. You lean on conversion data coming back through Merchant Center instead. The worst of the bad options.
I am not going to tell you exactly how those UCP conversions show up in Google Ads, or whether other platforms see anything at all, because Google has not documented that cleanly yet. I am also not going to tell you that you shouldn’t do this because you lose attribution and lose pixel tracking without a customer hitting your website.
What I will tell you to do is get it set up, watch that space really closely, and don’t trust a platform OR a random person that claims to know. Test and verify yourself.
What you can do now is concrete:

Get server-side tracking and enhanced conversions in place, so you capture everything capturable.
Set up the native commerce attribute and your feed for UCP.
Put more weight on blended efficiency and incrementality, because in-platform numbers are going to tell you less of the truth than they used to.

This is the time to move fast, adapt, break things, and adopt these changes at the very beginning because chances are, you will be ahead of your competition. And, as things become less chaotic, you will have gone through it while others are at the starting line.
Agentic Commerce PPC Playbook: What To Do Now
None of this is a reason to panic or to tear down what works. It is a reason to get a few things in order while the surface is still young.

Treat your product feed as a bidding asset. Fill every constraint field, keep it accurate, and refresh it often. Inclusion is won or lost here.
Make price, shipping, returns, and availability machine-readable and correct. These are the fields agents read first.
Decide your Direct Offers posture before you opt in. Pick the products, set a margin floor, and choose whether you lead with price or value.
Tighten Performance Max and Shopping controls. Use channel-level reporting and campaign-level negatives, and protect your brand traffic.
Shore up the measurement now. Server-side tracking and enhanced conversions for capture, incrementality, and a blended efficiency metric for the truth.
Confirm your eligibility on the surfaces that matter. Buy for Me needs Google Pay and a guest checkout option, and Shopify merchants have a faster path in.
Do not pull budget from Search and Meta yet. This is additive. The overwhelming majority of your revenue still flows through the campaigns you already run.

The Real Agentic Shift In Ecommerce
The advertisers who win agentic commerce will not be the ones with the cleverest ads. They will be the ones whose product data, margin posture, and measurement are ready for a buyer who never sees the ad. This is not something you should be planning for anymore; you should be moving on with this because Agentic Commerce is here.
The agent is becoming the customer you optimize for, and it judges you on inputs most accounts still treat as an afterthought. This is the real shift in ecommerce you should be paying attention to.
More Resources:

Featured Image: Ao Zaa Studio/Shutterstock

Read More »
June 15, 2026 No Comments
News

Google’s Unexpected Take On Site Folder Structure And SEO via @sejournal, @martinibuster

Google’s John Mueller answered a question about website category structure in URLs. The context was about a website that offered the same content in multiple languages and the person asking the question wanted to know what was the best way to structure the category URLs.The main site is in English, representing their primary market.
Their current category structure resembles:

example.com/blog
example.com/en-us/blog

and the internationlized versions have category structures like this: site.com/fr-fr/blog
A person asked the following question on Reddit:
“Do we need localized folders with duplicate content for our home market on our site?Hi all,
I’m familiar with hreflang tags and setting up alternate folders and references for different countries and languages, but I have a specific question for our home market. My client has a large site serving many international clients with localized content, but they’re a US-based company and that’s where the majority of their user base is.
At the moment they have 25+ international localizations across all of their core folders, including a /en-us/ folder for all their main pages.
The issue is, the content on the main site and in these /en-us/ folders is the same, so we’re splitting page authority and creating potential duplicate content issues which (as far as I can see) provide no discernible benefit….
Am I missing anything in my understanding of this?
Is there any specific benefit to the /en-us/ folders we’d be losing?
Are there other considerations or factors I should be thinking about?”
John Mueller’s Answer
Mueller’s answer focused on the impact to analytics and being able to accurately track the different visitors. His answer was indifferent to common SEO concerns like keywords or topical themes for the category names. That in itself is interesting and worth considering.
Here’s his answer:
“I’d generally recommend just one, but this likely isn’t going to make or break your site.
IMO the advantage of using /en-us/blog/ instead of /blog/ for US content (on an internation site that uses /LL-CC/anything URL patterns) is that it’s easier for you to filter & slice your metrics by country/language. I don’t think you’d see a practical SEO difference between using /blog/ or /en-us/blog/ for your US content. /blog/ looks nice, but /en-us/blog/ is also not super-weird.”
Many SEOs are rightfully concerned about the site architecture, with the words used in category folders, with the main concern being if keywords are being used and how that might influence the ability to rank, a concern the Redditor shared.
Mueller’s answer didn’t mention keywords but rather focused on the practical angle of how that will look for analytics. In fact, he said that in this specific case he didn’t expect any kind of SEO difference.
Featured Image by Shutterstock/SNEHIT PHOTO

Read More »
June 15, 2026 No Comments
Generative AI

Google Cloud Announces The Open Knowledge Format via @sejournal, @martinibuster

Google announced the Open Knowledge Format (OKF), a new open specification for organizing and exchanging the knowledge that AI systems need in order to perform useful work.The announcement explains the reason for developing this new specification:
“As foundation models continue to improve, the lack of relevant context often limits what they can do, especially as they are used to build agentic systems. While these models can help you write code, summarize documents, or analyze a dataset, they still need the right information to produce accurate and actionable results. “
AI Agents Need Context
AI systems often need knowledge that exists outside the model, including how data is structured, how systems work, how metrics are defined, and how internal processes operate.
That knowledge is usually scattered across catalogs, wikis, documentation, repositories, shared drives, and other internal systems, forcing AI agents to assemble context before they can complete a task.
Google says OKF is meant to solve that problem by turning scattered knowledge sources into a common format that can move between humans, AI agents, tools, and organizations.
What Open Knowledge Format Is
OKF is a format for representing organizational knowledge in a way that can be shared between different AI agents, tools, and organizations.
The format organizes concepts such as datasets, metrics, APIs, tables, and runbooks into documents that can be read by both humans and AI systems.
Google designed OKF to be simple and independent of any specific platform, allowing the same knowledge to be shared between different AI agents, tools, and organizations.
The announcement explains:
“To make the format concrete, we’re publishing reference implementations at both the producer and consumer ends:

An enrichment agent that walks a BigQuery dataset, drafts an OKF concept document for every table and view, then runs a second LLM pass that crawls authoritative documentation and enriches each concept with citations, schemas, and join paths.
A static HTML visualizer that turns any OKF bundle into an interactive graph view in a single self-contained file; no backend, no install on the viewing side, no data leaves the page.
Three ready-to-browse sample bundles: GA4 e-commerce, Stack Overflow, and Bitcoin public datasets, produced by the reference agent and committed to the repo as living examples of conformant OKF.

These are proofs of concept, deliberately. The agent demonstrates one way to produce OKF; nothing about the format requires a specific agent framework or LLM. The visualizer demonstrates one way to consume it; nothing about the format requires HTML or a graph view. We expect (and want!) the ecosystem of producers and consumers to grow far beyond what we’ve shipped.”
Who OKF Is For
OKF is designed around a producer-and-consumer model. Some users create, edit, and maintain the knowledge. Others consume it through AI agents, LLMs, software systems, or internal tools.
AI Agents and LLMs
AI agents and LLMs are the primary consumers of OKF. They use the format to access the structured context and curated knowledge needed to perform tasks and produce accurate results.
Useful For AI Agents And LLMs

Coding agents
Data analysis agents
Research agents
Internal enterprise assistants
Agentic workflows

Humans And OKF
OKF uses markdown files and YAML frontmatter, making the format readable and editable by people using standard tools.
People Who May Find OKF Useful

AI developers
Software engineers
Data engineers
Analytics teams
Technical writers
Business teams

Organizations And OKF
Organizations can use OKF to package and share institutional knowledge that would otherwise remain scattered across documentation systems, metadata catalogs, repositories, and internal tools.
Organizations That May Find OKF Useful

Organizations building AI agents
Data teams
Engineering teams
Knowledge management teams

Availability
Google is proposing a common format for representing organizational knowledge rather than a new platform for managing it.  The OKF specification, reference implementations, and sample bundles are available on GitHub. The announcement makes a point of saying that it is a starting point:
“OKF v0.1 is a starting point, not a finished standard. The format will evolve as more producers and consumers emerge and as we collectively learn what knowledge representations agents actually need in practice.
We’re publishing in the open from day one because that’s the only way a knowledge format earns its name, whether you’re building a knowledge catalog, an enrichment pipeline, a wiki tailored to AI agents, or anything in the AI knowledge domain.”
An explainer tweet by Tech With Mak shared why this solves a problem:
“The most underrated idea in agent tooling this year might be a gist Andrej Karpathy wrote about “LLM Wikis” – markdown libraries that agents read, update, and maintain on their own.
What followed was predictable. Teams everywhere started building their own version – AGENTS[.]md, CLAUDE[.]md, Obsidian vaults wired into coding agents, folders of index[.]md and log[.]md files agents consult before doing anything.
…Google just tried to close that gap with the Open Knowledge Format – a spec that says = > here’s the one field every concept needs (type), here’s a small set of optional fields if you want them queryable, and otherwise, write however you want.
It’s not a new tool or platform. It’s an agreement on shape, which is exactly what Karpathy’s pattern needed to stop being a hundred incompatible reinventions of the same idea.”
Read the original announcement here:
Introducing the Open Knowledge Format
Featured Image by Shutterstock/Poetra.RH

Read More »
June 15, 2026 No Comments