# How to Get Recommended by Gemini > How Gemini with Google Search grounding picks the pages behind a property-ops software answer, how the Index resolves its redirects, and what to fix first. Canonical: https://webpossible.com/get-recommended-by-gemini/ Source: https://webpossible.com/get-recommended-by-gemini/ Format: Markdown version for AI agents. The canonical HTML page is at the source URL above. Last verified: 2026-09-03 --- Shortlist · WebPossible’s property-ops practice # How to Get Recommended by Gemini Gemini answers a software question with Google Search behind it. When grounding is on, the model writes its own searches, Google runs them, and the answer comes back with the pages it stood on, down to which sentence rests on which page. That trail is the most detailed any of the four engines in the AI Shortlist Index provides. It is also wrapped in redirect URLs that hide the real page unless you resolve them, which the Index does. This page covers how grounding picks sources for a property-ops software question, what the Index has on Gemini so far, and what a vendor should do, in order. ## How Gemini picks sources Google’s Gemini API documentation for grounding with Google Search describes the mechanism. With the `google_search` tool enabled, the model decides whether a prompt needs a search. If it does, the response carries a `groundingMetadata` block with three parts: `webSearchQueries`, the searches the model wrote; `groundingChunks`, the web pages those searches returned, each with a URI and a title; and `groundingSupports`, which map segments of the answer text to the chunk indices they rely on, with confidence scores. Nothing else in the four engines gives you the query the model wrote from a buyer’s prompt. The same documentation says the URI on each chunk is a Google-hosted redirect rather than the publisher’s address, and asks apps that show grounded answers to display Google’s search suggestions alongside. The Index resolves every redirect URI to its final URL before classifying the page by type and linking it to the vendor whose domain it is, so the source map shows propertymeld.com rather than a Google redirect. Grounding supports are stored too, so a vendor can see which sentence of an answer rested on which page. Google’s crawler documentation says the Google-Extended token controls whether a site’s content is used to train Gemini models and for grounding in Gemini, and that it does not affect Search ranking or AI Overviews. A vendor that blocked Google-Extended in 2024 to keep content out of training may have cut itself out of grounding. Check that first. No Gemini runs are recorded in the September 2026 edition; the engine is marked missing and carries no mention rate for any vendor. The model set for the edition is gemini-3.8-flash. What the Index does have is a single ChatGPT run in the maintenance category. That run, captured through the OpenAI API on 3 September 2026 before ChatGPT capture moved to the app, consulted 124 pages, 38 of them on reddit.com, 16 on appfolio.com, and 12 on buildium.com, and quoted six: two Property Meld pages, AppFolio’s and Buildium’s maintenance feature pages, latchel.com, and rentvine.com. Because grounding runs Google Search, our hypothesis is that Gemini’s chunk set for the same prompt looks more like a Google top ten, with PMS vendor pages, listicles, and g2.com or capterra.com category pages, than like the Reddit-heavy list ChatGPT consulted. The October edition tests that. A grounding chunk is one page, and a support links a sentence to one or more chunks. The unit of evidence is a page that, on its own, backs a sentence about your product. That shapes everything below. ## What that implies you should do, in order First, fix Google eligibility. Grounding is Google Search, so the page has to be indexed, snippet-eligible, self-canonical, and readable without JavaScript. Then confirm Google-Extended is allowed in robots.txt. Run both checks before touching a word of copy. Second, write for the rewritten query. Gemini turns “What’s the best maintenance coordination software for a property management company with around 300 doors?” into searches that read like search queries: maintenance coordination software for property managers, maintenance software pricing per unit, maintenance software that integrates with Buildium. The `webSearchQueries` field shows the exact rewrites once the runs exist. Until then, title and head your pages the way those searches read. Third, make each page self-contained enough to be a chunk. A pricing page that says, in one paragraph with the product name in it, “$1.60 per unit per month, $160 monthly minimum” supports a sentence by itself. A page where the product name is in a header image and the price is in a table loaded by script does not. Integrations get the same treatment: a page that names Yardi, RealPage, Entrata, AppFolio, Buildium, and Rent Manager in plain text, with one line on what each integration does, is a chunk that can support an integration claim. The [prompt library](/ai-shortlist-index/prompt-library/) has three integration prompts per category, mvn-int-03 among them, “Which maintenance and vendor management platforms integrate with Yardi Voyager for a multifamily portfolio?” Fourth, get corroborated. A sentence naming your product that rests on two chunks from two domains, your pricing page and a reddit.com thread or a PMS marketplace listing, is one we expect the model to keep; a sentence resting only on your own site is one we expect it to hedge. In this category the corroborating domains were reddit.com, appfolio.com, buildium.com, PMS help centers, and the listicles on buildium.com and lula.life. The category [source map](/ai-shortlist-index/maintenance-vendor-network-software/sources/) lists them with counts and will add Gemini’s resolved domains as the runs come in. Fifth, make the entity agree with itself. One name, one category, one integration list across your site, Crunchbase, LinkedIn, g2.com, capterra.com, and the PMS marketplaces. Grounding pulls chunks from several of those at once, and a model handed two categories for one company picks one and moves on. Sixth, if you monitor Gemini yourself, resolve the redirects. Without that step your source report is a column of Google URLs and you will not know whether the page behind the mention was yours or a competitor’s comparison of you. ## What the Audit finds The [Shortlist Audit](/ai-visibility-audit/) is $3,500 and takes seven days. It runs the full 25-prompt set for your category across all four engines and returns your position, plainly; the cause, as ranked findings labeled verified, inferred, or hypothesis with evidence links; the source gap, meaning which of the top ten cited sources in your category you are absent from; a 90-day fix scoped to a [Sprint](/answer-engine-optimization-for-proptech/); and what we won’t promise, which is rankings, traffic, leads, or revenue. If we can’t name the cause, you don’t pay for implementation. On the Gemini column the Audit shows the searches the model wrote for each prompt, the resolved pages it grounded on, and which sentences about you rested on which page. A vendor whose mention rests on a competitor’s comparison page has a different fix from one whose pricing page was never retrieved, and the Sprint is scoped to the one you have. If you want to talk before requesting anything, [book a 20-minute call](/contact/). The mechanism in more depth is in the guide to [how search grounding shapes Gemini’s answers](/gemini-seo/). The other engines have their own pages: [getting your software into ChatGPT’s answers](/get-recommended-by-chatgpt/), [earning Perplexity’s citations](/get-cited-by-perplexity/), and [appearing in Google AI Overviews](/appear-in-google-ai-overviews/). All of it rolls up to the [Shortlist practice page](/ai-visibility-for-property-management-software/). ## Questions ### Which Gemini model does the Index use? For the September 2026 edition, gemini-3.8-flash through the Gemini API with the Google Search grounding tool enabled. The model identifier returned by the API is stored on every run, so a month-to-month move that follows a model change is visible as one. The [methodology page](/ai-shortlist-index/methodology/) keeps the engine and version record. ### Why are Gemini's source URLs redirects? Google's Gemini API documentation returns each grounding chunk with a Google-hosted redirect URI rather than the publisher's address. The Index follows every redirect to its final URL before classifying the page and linking it to a vendor, so the source map shows real domains. If you run your own checks, resolve the redirects or your source list is a list of Google links. ### Is the Gemini app the same as the API with grounding? Close, not identical. The app adds personalization, memory, and other Google surfaces the API does not expose. The Index treats the API with grounding as the repeatable proxy and says so on the methodology page rather than claiming parity. ## Get your row explained — $3,500 Seven days. Twenty-five prompts across four engines. The cause behind your row and a 90-day fix. Ryan replies with an agreement and an invoice; no call required. Leave this empty Name Work email Company domain Request the Audit Prefer a call? [Contact](/contact/). Goes to ryan@webpossible.com with this page’s URL attached. Never shared, never sold. See the [privacy policy](/privacy/). Part of [Shortlist, the practice](/ai-visibility-for-property-management-software/). Method behind this page: [the methodology guide](/gemini-seo/).