The Local Pack Is Dying (Sort Of): What Comes Next
The Google local 3-pack isn't literally dying. It still sits at the top of most "plumber near me" searches, it still drives phone calls, and Google has every incentive to keep it there. But its share of local intent has been quietly bleeding out for two years, and that trend is accelerating. AI Overviews answer more queries before the pack is ever shown. Perplexity and ChatGPT handle the research phase in prose. Apple's on-device search means a growing slice of "near me" queries never reach Google at all. I'm Jake McCluskey, and I've spent 25 years watching local search change its shape. This is my take on where it's going, and what a realistic local visibility strategy looks like when the 3-pack is one surface among many instead of the only one that matters.
Is the Google local pack actually going away?
No, not literally. It's losing share of queries and share of attention, not its existence. For businesses whose strategy depended on ranking in the 3-pack as the whole game, that functional decline is what matters more than whether the pack still exists in 2030.
When Google introduced AI Overviews in 2024, the local pack started appearing below the generated answer for an increasing percentage of queries. For research-style local questions ("best family-friendly restaurants in my area that take reservations"), the Overview now often answers inline, and the user never scrolls to the pack. For transactional queries ("pizza near me open now"), the pack is still front and center.
That's the split most businesses haven't adjusted to. Your visibility for transactional queries might be fine. Your visibility for considered-purchase research queries, which are where the higher-value customers tend to start, may have quietly collapsed without you noticing.
Another shift worth naming: zero-click search keeps expanding. An AI Overview that answers the question inline, complete with a summary of reviews and hours, means the customer never visited your site or clicked a Maps listing. That's fine if the AI mentioned you favorably. It's a silent lost deal if it didn't.
Where are local searches actually going now?
Local searches are fragmenting across at least four surfaces: Google's organic local pack and AI Overviews, conversational AI tools like ChatGPT and Perplexity, Apple's on-device search (especially on iPhone), and category-specific vertical search engines like Yelp, Zillow, and Thumbtack.
No single channel owns more than half of any category's intent anymore. That's the real story. A decade ago, winning the Google local pack was 80% of local visibility. Now it's closer to 50%, and the other 50% is scattered across tools that each have their own ranking signals, their own data sources, and their own moats.
That fragmentation is why the old approach (optimize GBP, rank in the pack, done) is incomplete. You can be #1 in the pack and still be invisible in AI answers, invisible in Apple Maps, and missing from the top category directories. That stack of gaps adds up to a real revenue hole.
The good news: many of the signals overlap. Clean structured data, consistent citations, and review content help across most of these surfaces simultaneously. You're not running five different strategies. You're running one broader strategy with each surface as a distribution point.
The businesses that adjust early tend to find it cheaper than expected. Once the underlying entity signals are clean, each additional surface is mostly a matter of claiming and maintaining a profile, not building from scratch. The hard part is doing the cleanup work that's been deferred for years. Adding Apple Business Connect or updating Bing Places on top of a clean foundation is the easy part.
What does conversational local search actually look like?
Conversational local search is multi-turn, context-aware, and prose-based. A user asks an AI tool a vague question, the AI asks clarifying questions or makes assumptions, and eventually it recommends one or two specific businesses with reasoning attached.
An example I ran recently: "I need a dog groomer in my area who can handle a nervous rescue." ChatGPT asked for my zip code, then recommended two specific groomers with reasoning pulled from their website content and reviews. Neither of those groomers ranked #1 in the local pack for "dog groomer near me." Both had specific content on their sites about working with anxious dogs. That specificity is what got them cited.
That's the new shape. Not "who ranks for the category keyword," but "who's described in a way that matches this specific intent." It rewards businesses with depth, clarity, and real differentiation. It penalizes businesses with generic pages that could be describing any competitor in the category.
It also rewards businesses that answer the long-tail questions. Conversational queries are inherently specific. "Dentist" has one answer. "Dentist who takes Delta Dental and can see my seven-year-old on Saturday mornings" has maybe three, and if you answer that question directly somewhere on your site, you're likely to be one of them.
Why do reviews matter more now than 3-pack ranking ever required?
Reviews now function partly as training and retrieval data for AI models, not just as a ranking signal for the pack. The language in your reviews shapes how the AI describes you in generated answers, which means review quality (and specifically review content) has a second-order effect it didn't have before.
A 4.6-star average used to be a strong signal and a human-legibility factor. Now, on top of that, the words inside those reviews get read by models. When the AI says "Smithson Plumbing is known for transparent pricing and showing up when they say they will," it's composing that sentence from language it pulled out of reviews, site copy, and citations.
That changes the review ask. Encouraging customers to describe what you actually did, in their own words, is worth more than asking for stars alone. Prompts like "What surprised you most about working with us?" or "Would you mind describing what we did for you?" tend to generate the kind of specific language AI tools can quote.
It also means bad reviews are worse than they used to be. A specific, articulate negative review ends up in AI descriptions the same way a positive one does. The response you write matters. A calm, accountable reply partially neutralizes a bad review in the AI's eye, because the AI reads the thread, not just the review.
What is a 'local business entity' and why does it matter?
A local business entity is the aggregate of everything the web knows about your business, and it matters because AI tools increasingly reason about entities, not just pages. A clean, consistent, well-described entity gets recommended. A fragmented one gets skipped or described inaccurately.
Entity signals include:
- One consistent business name across every surface
- One address, one phone, one canonical domain
- Structured schema markup that declares what you are
- Citations from authoritative local and industry sources
- Reviews that reinforce the entity's identity and specialties
- Content that clearly describes your services, your service area, and your distinct approach
- Links to and from adjacent entities (industry associations, partners, local publications)
When those signals align, AI tools can confidently recommend you for queries that match your actual strengths. When they're fragmented (different phone numbers, slightly different names, thin content, sparse citations), AI tools hedge, skip you, or confuse you with a competitor.
The shift is subtle but real. Ranking for "one query" is being replaced by being recognized as "the right entity for a class of queries." That reframe changes where you invest your SEO effort.
What should a local business actually do differently now?
Do three things differently: build for the diversified stack, write for conversational and AI retrieval, and invest in entity clarity rather than chasing individual keyword rankings. None of this is exotic. Most of it is what strong local brands have always done, just applied with the AI era in mind.
What I'd actually do this quarter if I ran a local business:
- Run a manual test of 20 real customer queries across Google, ChatGPT, Perplexity, Gemini, and Apple Maps. Document where you show up and where you don't.
- Tighten entity signals: one business name, one phone, clean NAP across the top 30 citations, LocalBusiness schema on every location page.
- Rewrite the service pages to answer real customer questions with real specifics. Not keyword-targeted, question-targeted.
- Upgrade the review request flow to encourage descriptive language, not just stars.
- Add depth content on the long-tail queries your actual customers ask.
- Build category-specific presence on 3 to 5 vertical platforms that matter in your category.
That's a quarter of focused work. Done right, it compounds across every AI surface simultaneously, not just Google.
What happens to local SEO as AI keeps evolving?
The fundamentals stay. The surface layer gets more varied, and the reward shifts toward businesses that make it easy for AI tools to describe them accurately. That's the throughline I keep coming back to: clarity, depth, consistency, and evidence.
The businesses I'm watching over the next two years are the ones that treat local SEO as reputation engineering across a distributed network of AI-readable surfaces. Not Google alone, not GBP alone, not AI alone, but the stack of all of them with one coherent business entity underneath.
The ones still treating it as "rank in the 3-pack or lose" are going to be caught off-guard when the 3-pack shows on half as many queries and their share of revenue from the other half is invisible because they never optimized for any of those surfaces.
If you want an honest look at where your local business stands across the diversified stack, the free local SEO audit checks most of the surfaces that now matter. Or book a discovery call and I'll tell you, in plain terms, which of these moves would pay off fastest for your specific situation. I'd rather give you three actionable observations than a 40-page deck you'll never read.