Joe Della Mora

Growth at @GroundScoreAI — we check whether ChatGPT recommends your business, and fix it when it doesn't.

[Joe Della Mora @jdellamora]

We ran this live across 7 cities last week. The answers are real — and most businesses have no idea which side of them they're on. https://t.co/o3pbJEm4xe

[GroundScoreAI @GroundScoreAI]

1/ Google ranks pages. AI recommends businesses.
Different game, different rules. What actually gets a business cited by ChatGPT — a thread.

[Joe Della Mora @jdellamora]

@achille610 That reframe holds up in practice. Most sites we scan are not badly written, they are unlabeled. The facts a model needs sit in an image, a widget or a PDF, never as text it can lift. Understandable is a lower bar than ranking, and more sites miss it.

[Joe Della Mora @jdellamora]

@brodieseo Interesting that the feed is the lever here. Services and local have no feed to submit and no ad unit, so the recommendation answer stays entirely earned. Two different games forming inside the same product, and only one of them has a paid entry point.

[Joe Della Mora @jdellamora]

@screamingfrog @Cloudflare From ranking to recommended is the right frame. The part that lands hardest locally: there is no page two. Ten positions and a next button collapse into two or three names and a phone number. Position four is not further down the list, it is simply absent.

[Joe Della Mora @jdellamora]

@GoogleSmallBiz The step most owners skip is the boring one: write the answers as plain text on the page. Across 433 local business sites we scanned, 89 percent had no FAQ schema and 36 percent had no structured data at all. Usually it is not quality holding them back, it is labeling.

[Joe Della Mora @jdellamora]

@wwardenn @perplexity_ai @GeminiApp The one I would add for local and services: often there is nothing to be cited. The answer names two businesses and gives a phone number, no sources. Citation tracking reads zero while the outcome you care about already happened. Worth tracking named-or-not separately.

[Joe Della Mora @jdellamora]

@RickyDPR Love it!

[Joe Della Mora @jdellamora]

@publisherinabox The opt-out framing is the hard part. Publishers are being asked to choose between the AI Overview and the placement that now sits inside it, which is not really two decisions. Worth watching how many stay opted out once they see what it costs above the fold.

[Joe Della Mora @jdellamora]

@vadimk_77 @tibo_maker Good tactic, and the underrated half is the criticism. A page of only glowing reviews reads as marketing to a model. Names, dates and a few honest complaints are what make it look like a source rather than a landing page.

[Joe Della Mora @jdellamora]

@arthuryuzbashew 304 is probably the floor, not the ceiling. Plenty of people get the recommendation from ChatGPT and then type your name into Google or the App Store, which lands in analytics as something else entirely. The referral is real, the attribution is not.

[Joe Della Mora @jdellamora]

@RickyDPR The gap you are describing is real and invisible from the outside. A site can look completely fine and still have nothing labeled for a machine to read. We scanned 433 local sites: 89% had no FAQ schema, 36% had no structured data at all. Looks ok, reads as nothing.

[Joe Della Mora @jdellamora]

@BrianEDean The local version is even starker. Ask for a plumber in a city and the answer often cites no sources at all, it just names two and hands over a phone number. Nothing to click because nothing is shown. Being on the list is the whole game.

[Joe Della Mora @jdellamora]

@InferonLabs Right, and the reason is usually stated right there in the answer, which almost nobody goes and reads. That is what I built GroundScore to surface. Free to start at https://t.co/ec2qRLlDCk if you want to run it on your own site or a client's. Would value your take.

[Joe Della Mora @jdellamora]

Founders who built a "check your X for free" tool: did giving the result away with no signup, and earning the account later, actually work? Or did you gate it behind an email from day one? Genuinely torn on where that line belongs.

[Joe Della Mora @jdellamora]

@jay_neyer Worth adding the other blind spot: it is Google only. Nothing in there covers what ChatGPT or Perplexity said about you. On "who should I hire near me" questions the answer is two names and a phone number. No click, no impression, nothing to log anywhere.

[Joe Della Mora @jdellamora]

@Prosperitiv @neilpatel Agree, and locally it is starker: ask for a recommendation and the answer often cites nothing at all, just two names and a phone number. Citation tracking reads zero while the entity is doing fine. Being returned to is the measure; the citation is a side effect.

[Joe Della Mora @jdellamora]

@Josh_Lowry The jump from one to two is steeper than the sequence suggests. "Can they find you" had ten slots and a page two. "Will AI recommend you" returns two or three names and stops. Whatever the agents choose from, they inherit that shortlist.

[Joe Della Mora @jdellamora]

@brandshareio Fair on the vocabulary, but Google can only speak for Google. The answer that names two businesses and hands over a phone number is often ChatGPT, and nothing in Search Console tells you what it said. Same craft, second surface nobody reports on.

[Joe Della Mora @jdellamora]

@nickswan The line that stands out is a public body putting "AI search visibility" in a procurement doc. Once it is a tender requirement it stops being a conference topic. Curious what they will accept as proof of delivery: impressions, or whether the answer names them.

[Joe Della Mora @jdellamora]

@lightsilver323 The problem: your customers now ask AI who to hire, and most businesses have no idea whether they get named or are invisible. GroundScore shows you, across ChatGPT, Perplexity, Gemini and Claude, and what to fix.

[Joe Della Mora @jdellamora]

@lightsilver323 GroundScore. It shows whether AI like ChatGPT, Perplexity and Gemini actually names your business when buyers ask who to hire, and what to fix if it doesn't. Think "are we ranking" for AI search.

Free to try: https://t.co/ec2qRLlDCk