Cloud Stack does not do anything we can measure

Cloud Stack is sold as a link building engine. You give it a target page and a few keywords. It writes a branded article about the business, then publishes that article to a list of major cloud hosting providers. Each published copy links back to your site.

The pitch rests on one idea: a link from Amazon, Microsoft Azure, or Google grade infrastructure carries the trust of that infrastructure. Stack up 10 to 24 of them and the target page gets a meaningful lift in authority, which lifts rankings, which brings calls.

1. 1You pick a target page and keywordsThe tool auto derives the rest from the client's Brand Vault and knowledge graph.
2. 2It writes a unique article per providerVariations are used so the copies do not read as duplicate content.
3. 3It publishes to 10 to 24 cloud providersAmazon, Azure, Google, Cloudflare, DigitalOcean, Backblaze, Wasabi, Vercel, and more.
4. 4Each copy links back to the target pageThis is the deliverable: a bundle of new followed backlinks.
5. 5Authority flows to the target pageWhich is supposed to move rankings, traffic, and eventually booked jobs.

Cost is roughly 25 to 35 dollars in Hyperdrive credits per stack, and about 20 minutes of run time. We have run it 19 times.

The claim makes four testable promises. Each one was checked against live data, not against what the tool reports about itself.

One deliberate choice here. Search Atlas reports its own index status for each published page, and it reports 5 of 14 as indexed for Texas Strong. That number was not taken on trust. Every test below measures an outcome, not a status flag.

This part of the claim holds up. The tool does what it says at the mechanical level, which is why the failure further down is easy to miss.

Texas Strong's stack (ID 36165) published 14 pages. Fetching the Wasabi copy directly returns HTTP 200 and 44,820 bytes of real HTML. The page carries <meta name="robots" content="index, follow">. It has 7 outbound links, one of which points at texasstrongplumbing.com/services carrying rel="noopener", which is not nofollow. That is a followed link by every technical definition.

Worth flagging separately: all 19 stacks report status Published but deployed_provider_count: 0. In the Cloud Stack lifecycle, Published and Deployed are different states. Not one stack we have ever run reached Deployed.

This is the root cause, and it is structural. The pages are fine. The address is the problem.

That Wasabi page does not sit on its own domain. It sits in a shared storage bucket called lg-cloud-stack, alongside every other Search Atlas customer's stack pages. The bucket allows public listing, so it can be counted exactly. I walked all 32 pages of the listing.

Here is a random sample of what shares the bucket with your Katy plumber:

Search Atlas marks 5 of those pages as indexed. Search Atlas's own backlink index counts zero of them. Both numbers come from the same vendor.

I pulled the complete referring domain list for two clients, every page of it, and searched for all 14 provider hosts: wasabisys, contabostorage, blob.core.windows.net, backblazeb2, scw.cloud, cloudfront, vercel.app, marketer.report, exo.io, upcloudobjects, myqcloud, b-cdn.net, digitaloceanspaces, and git.searchatlasseo.com.

Mayday is the strongest possible test case. It has had a stack since well before the current one, two separate runs, and its profile is 196 domains deep. Still zero.

Separate issue worth its own look: both profiles are full of junk we did not build. backlinkstree.com, backlinksbank.com, cheapsmmprovider.online, goooogla.com, and a pile of .sbs and .cfd domains at authority 1 to 2. That is auto generated scraper spam, not ours, but it is sitting in the client's profile.

A link that is not counted might still send a human. This is the cleanest test in the review because it needs no interpretation. Either someone clicked or they did not.

GA4, every traffic source, 10 client properties, 1 August 2025 to today.

Not low. Not noise level. Zero. In twelve months, across 1,192 distinct traffic sources, not one visitor arrived from a page we paid to publish.

One near miss checked and cleared: Curated Events shows 8 sessions from themeadowsatfireflyfarmpreserve.b-cdn.net. That is a venue partner's own CDN. Our Bunny host is lg-cloud-zone-v2.b-cdn.net. Different property, not a stack referral.

This was the one I expected to find something in, and it came back the cleanest zero of the four.

A thin, keyword dense, indexable page full of business facts is exactly the shape of thing an AI crawler might pick up. AI visibility is also what we actually sell, so if Cloud Stack helped anywhere, this is where it would show. I pulled every domain the six AI platforms cite when discussing each client, then searched the full lists.

When ChatGPT or Gemini recommends a plumber in Katy, it reads katyplumbers.com, Yelp, Reddit, Google, the BBB, and the client's own site. When it discusses Mayday, it reads Orkin, Terminix, This Old House, Reddit, and the client's own site. It has never once read a page we published through Cloud Stack.

Search term check on the full lists: "wasabi" returned 0 results on both clients. "contabo" returned 0. "cloud" returned 0 on Texas Strong and 3 on Mayday, all of which are unrelated real businesses (McCloud Services, a competing pest control company, plus CloudKitchens and Cloudy Pet Care).

This matters because it rules out the easy excuse. It is not that nothing moves for these clients. Things move. Cloud Stack is just not one of them.

Mayday Pest, same period, same account, with a stack running the whole time and contributing nothing to it:

Impressions over the same five months went from 6,772 to 120,630. That is real compounding visibility, and every bit of it traces to the work that is not Cloud Stack: the profile work, the content, the citations, the technical fixes, the review engine.

Cloud Stack does exactly one thing. It creates pages that exist. Every channel where existing turns into value reads zero.

What this has cost

At roughly 25 to 35 dollars in Hyperdrive credits per stack, 19 stacks is 475 to 665 dollars in credits, plus the operator time to run and check each one, plus a link scheme footprint sitting on client backlink profiles for no return.

The honest caveat

Absence from Search Atlas's backlink index is not proof of absence from Google's index, and a ranking contribution too small to separate from noise cannot be fully ruled out. But something that reads zero on backlinks, zero on traffic, and zero on AI citations is not something we can report to a client or build an offer on. If it were working, it would surface somewhere in 34,751 citations or 1.5 million sessions. It does not.