Google SGE and Local Search: What the 2023 Tests Showed
Google's Search Generative Experience (SGE) was an experimental search format that placed an AI-generated snapshot above or alongside conventional results. For local queries, that snapshot included a map-style pack, business descriptions and a follow-up prompt. This guide explains what was observed during the 2023 US waitlist period, how the SGE local result differed from the classic local pack, and which listing and review tasks remain useful while the format changes.
What SGE was, and why local search mattered
SGE was Google's attempt to answer longer, more specific questions directly in the results page. Instead of returning only ten blue links and a map pack, it generated a summary using large language models and linked to sources. The example used at the time contrasted a broad 'hairdressers near me' search with a detailed request for a men's hairdresser offering a traditional wet shave and trim package in Boston. The point was not that short queries disappeared; the captured tests showed that a classic query such as 'brew pub near me' still produced a substantial SGE response.
Local search sat close to the centre of this change because local intent is often comparative. A searcher wants to know which businesses are nearby, how they differ and what other people say about them. Greg Sterling of Near Media was quoted as saying that local and products might be two of the most affected segments. That made the local pack a useful place to watch.
What the SGE local result looked like
The 2023 US tests described a coloured box carrying the caveat that generative AI is experimental and that information quality may vary. Inside that box, the local result had several recognisable parts:
- A five-item pack that resembled the existing local pack.
- A short description beside each entry, sometimes drawing on reviews.
- A carousel of cited results above a map with pinned locations.
- A prompt to ask a follow-up question, with suggested continuations.
- In some cases, the ordinary local pack remained, with an invitation to see an AI-powered response.
This is best read as a field map rather than a fixed template. The number of local results and the ordering were not identical to the classic pack, and the format was still experimental. The practical lesson for a business is that the same listing can surface in more than one place on the page, and each placement may show different text.
SGE versus the classic local pack
The clearest distinction from the early tests was that the SGE local result and the traditional local pack were related but not identical. Amanda Jordan of RicketyRoo observed that asking a query outright and conversing with the AI produced very different local business recommendations. Joy Hawkins of Sterling Sky noted that SGE gave visibility to businesses already appearing in the local pack, but that the two lists did not match exactly; sometimes the AI box used a wider map view that showed different businesses. Aleyda Solis described the SGE local output as largely a duplication of the map pack already shown below the first organic results.
Those observations point to a useful failure distinction. A business can be visible in the classic local pack and absent from the SGE snapshot, or appear in the SGE snapshot with a different set of neighbours. Neither outcome should be read as a permanent ranking verdict. The more reliable signal is whether the business appears consistently across the result types that matter for its category.
Reviews, descriptions and the content SGE reused
One of the more practical findings was that SGE could draw on Google Business Profile review text when answering follow-up questions. Andy Simpson documented an example of review content being used in that way. The captured tests also showed business descriptions, images and reviews appearing in results. That makes the quality and recency of listing content more than a cosmetic concern: if the AI summarises what is already on the profile, vague or outdated descriptions give it poor material to work with.
A sensible review of a profile therefore covers the name, category, description, images and the themes that recur in reviews. The aim is not to write for a particular AI model but to make the public listing accurate and specific enough that a summary has something true to say.
A worked example from the captured tests
Consider the two local searches used in the 2023 screenshots: 'brew pub near me' and 'arcade in Santa Cruz'. The first is a short, category-led query; the second names a place and a type of business. In both cases the SGE response presented a local pack with descriptions and a map, and offered a way to continue the conversation.
For a brew pub, the practical checklist would be: is the category correct, does the description name the kind of food and drink offered, are recent photos present, and do reviews mention the things a visitor would ask about? For an arcade, the same questions apply, with an added emphasis on location and family suitability. The example is useful because it shows that SGE did not require an exotic query. Ordinary local searches were enough to trigger the format.
What local businesses should keep doing
The early evidence did not justify panic or a wholesale change of strategy. It justified attention. A practical sequence is:
- Keep the Google Business Profile accurate: category, description, hours, images and services.
- Treat reviews as content: respond to them, and notice the themes that describe the experience.
- Monitor more than one result type: the classic local pack, the map results and any AI-generated local snapshot.
- Judge success by enquiries and revenue where possible, not by a single position number.
- Re-check the format periodically, because the 2023 observations describe an experimental product, not a settled interface.
Myles Anderson, the platform's CEO and co-founder, made a related point: ranking remains a valuable data point, but lead and revenue generation are the stronger proof of success, and those measures are less distorted by changes in Google's interface. The monitoring task is straightforward in principle; the harder part is deciding which result types deserve attention for a given business.
How this guide should be read
The observations here come from the 2023 US waitlist period, when SGE was experimental and not generally available at that time. The screenshots and expert comments describe that period, not the current state of Google search. Google's interfaces change, and a format that existed in 2023 may have been renamed, replaced or rolled into another product. Use this page as a dated reference for how generative local results were structured and what they implied for listing and review work, and check current conditions before making decisions that depend on a particular result layout.
Frequently asked questions
Did SGE replace the normal local pack?
Not in the 2023 tests. Some searches showed the SGE local result, some showed the traditional local pack with an option to see an AI response, and some showed both. The two lists did not always contain the same businesses in the same order.
Why did the SGE local result show different businesses from the local pack?
The early tests found that the SGE snapshot and the classic local pack were not identical. Different map views, different ordering and different recommendation logic could all produce a different set of businesses. A business appearing in one and not the other was a normal observation, not necessarily an error.
Do Google reviews affect SGE local answers?
In at least one documented case, SGE used text from Google Business Profile reviews to help answer a follow-up question. The broader lesson is that review content and listing descriptions are part of the material a generative summary can draw on, so keeping them accurate and current is worthwhile.
What is conversational mode in SGE?
Conversational mode was the follow-up prompt at the bottom of an SGE result. It suggested related questions and let the searcher continue the exchange rather than starting a new search. For local queries, that meant a searcher could move from a broad category to a more specific comparison without leaving the results page.
Should local businesses change their SEO strategy because of SGE?
The 2023 evidence supported continuing established local SEO work rather than replacing it. Accurate listings, relevant descriptions, current images and well-managed reviews were the practical levers. Monitoring additional result types and judging outcomes by enquiries and revenue were more useful than chasing a single position number.