Round 7 Investigation: The Reputation Machine, Measured
August 2026 — Every previous round listened to UltaHost's customers. This one listens to UltaHost. Using two public datasets — Meta's advertising archive and Trustpilot — we counted what the company tells prospective buyers and compared it, claim by claim, against the record. We also tested one specific allegation against the company and found it did not hold.
7.1 The 4.9 Claim, Running in 101 Paid Ads
Round 6 established that UltaHost displays a 4.9 rating while Trustpilot showed 3.4. Round 7 answers the question that follows: how widely is that number being distributed, and who is paying to put it in front of people?
Under the EU's Digital Services Act, Meta publishes an archive of advertising run on its platforms. On that archive held 166 UltaHost ads, placed between and , of which 43 were still running.
What the ads say
- ● 101 of 166 ads advertise a 4.9/5 rating — 33 of them still active
- ● 41 ads phrase it as “Rated 4.9/5 by 1900+ Users”
- ● 75 ads advertise “20x faster” NVMe storage
- ● 9 ads promise a 30-day money-back trial
The 4.9 is not presented as an internal figure. UltaHost's homepage attaches it to named review platforms with their own branding. The badge image file is literally called badge/trustpilot.png, and its alt text reads “Rated 4.9 from over 1000 reviews”. Two further badges carry badge/g2.png (“Rated 4.9 with over 300 reviews”) and badge/hostadvice.png (“4.9 star rating from 1,800 reviews”). The VPS page states “4.9 out of 5 based on 1,932 UltaHost Review”.
What Trustpilot actually shows
On , the Trustpilot profile for ultahost.com returned a TrustScore of 3.6, rendered as 3.5 stars, from 1,739 reviews. That is 1.3 points below the figure on the Trustpilot-branded badge. The score has moved up slightly since Round 6 recorded 3.4 in May 2026 — the gap has narrowed a little, and it remains a gap.
We are recording a discrepancy between two published numbers, both of which anyone can check. We make no claim about how it arose or whether it was intended.
7.2 Three Uptime Figures, All At The Same Time
This finding needs no external source at all. The contradiction is entirely internal to UltaHost's own advertising, and it is visible in a single pass over the archive.
| Claim | Ads | Still active | Running |
|---|---|---|---|
| 100% uptime | 58 | 17 | Mar 5 – Aug 3, 2026 |
| 99.9% uptime | 36 | 8 | Apr 10 – Aug 3, 2026 |
| 99.99% uptime | 3 | 0 | Sep 22 – Oct 1, 2025 |
| 99.99% Uptime Guarantee (own website) | — | current | As of Aug 4, 2026 |
Why this matters
No hosting provider can deliver 100% uptime, and UltaHost's own contractual promise is lower than what 58 of its ads assert. More seriously, the window in which those 100% ads ran — to — fully contains the Istanbul storage-node degradation of June 22 – July 3, 2026, logged by independent monitors and acknowledged on UltaHost's own status page as failed disks with a recovery running behind schedule.
For scale: the one independent month-long test on record measured 43.78% uptime with 17 consecutive days down, ranking UltaHost #3752 of 3,860 hosts.
7.3 Where The Ads Run — European Transparency Data
Because the Digital Services Act requires it, ads that reach the European Union carry reach reporting that ads elsewhere do not. That reporting makes the scale of distribution measurable rather than a matter of speculation.
- ● 122 of 166 ads carry an EU transparency block
- ● 34 ads explicitly declare that they target the EU
- ● Reach reported across 33 EU countries, most frequently Spain, Germany, Italy, Portugal, Poland, Bulgaria, France and Cyprus
- ● Summed declared EU reach across the archive: 410,641
- ● Placement spans Facebook, Instagram, Messenger and Audience Network
Reach figures are Meta's own reported numbers and are summed here across individual ads, so the same person may be counted more than once. Treat the total as an indication of distribution scale, not a count of distinct people.
7.4 How The Average Is Made
Round 6 identified the reputation gap. This is the mechanism that produces it — and it turns out to be more interesting, and more defensible, than the fake-review story people usually reach for.
Trustpilot records how each review arrived: whether the company invited it through Trustpilot's automated invitation flow, or whether the customer navigated to Trustpilot on their own initiative. Splitting the 200 most recent reviews for ultahost.com by that field produces two populations describing what look like two different companies.
| How the review arrived | Reviews | Rated 4–5★ | Rated 1–2★ |
|---|---|---|---|
| Invited by UltaHost | 135 | 95 (70%) | 32 |
| Organic walk-in | 58 | 4 | 54 (93%) |
The rest of the shape
- ● Only 8 of 200 reviews sat at exactly 3★ — a 4% middle, with the mass piled at both ends
- ● 69% of reviewers had written exactly one review in their lifetime; among 5★ reviewers, 73%
- ● UltaHost replied to 100% of sampled reviews, at every star level
- ● 16 negative reviews concerned refunds or billing; 6 concerned downtime
What this is not
We want to be precise, because the obvious accusation is the wrong one. We found no evidence of fabricated review text: across 200 reviews, exactly one body of text appeared under more than one author. Soliciting reviews is explicitly permitted on Trustpilot and is not misconduct — thousands of legitimate businesses do it. What the data shows is narrower and harder to dismiss: UltaHost's public average is substantially a function of which customers get asked, and the customers who arrive unprompted report a different product almost unanimously.
7.5 What We Tested And Could Not Support
A recurring claim about UltaHost is that it buys comments on its Facebook advertising to manufacture the appearance of a happy customer base. We went looking for that, and we did not find it. We are publishing the negative result, because a site that only ever reports confirmations is not measuring anything.
First: ad comments cannot be inspected by anyone
All 166 archived ads are what the industry calls dark posts— creatives built inside Meta's advertising tools rather than published to the page timeline. Not one of the 166 references an existing page post, and none carries a reshared-post reference. A dark post has no public comment thread, and Meta's Ad Library deliberately exposes ad creative without ad engagement. So the specific claim is not merely unproven — on public data it is unprovable in either direction, for us and for anyone else.
Second: the comments we can inspect show the opposite
UltaHost's ordinary page posts are public, so we read all of them. Across 150 posts spanning to :
- ● 37 comments in total, from 16 distinct accounts
- ● Zero cases of identical comment text under different accounts — the signature of bulk-bought commenting
- ● Comments skew negative: 11 complaints against 8 positive remarks
- ● The only apparent clustering was one dissatisfied customer reposting the same grievance eight times
A page with roughly 20,800 likes producing 37 comments in seventeen months is evidence of low organic engagement, not manufactured engagement. If UltaHost is buying comments anywhere, it is not on the surface where anybody could see it.
We also could not confirm a claim carried in earlier rounds. The separately tracked ultahost.io Trustpilot profile did not resolve on : one retrieval redirected it to the .com profile, a second returned nothing at all. Accordingly we have withdrawn the combined 2,570-review figure previously shown on this site and reduced the review count to the 1,739 we can currently verify on the .com profile. Whether the .io profile was merged, removed, or is simply unreachable to us, we do not know, and we are not going to guess.
7.6 New Customer Accounts
One pattern cleared our two-source threshold during this round: the same product failure and the same refund outcome, reported independently on two different platforms ten weeks apart.
macOS server: billed at once, delivered unusable, refund refused
A comment on UltaHost's own Facebook page, dated , states the customer “purchased a macOS server from Ultahost”, that UltaHost “charged my payment immediately, but the service was not delivered in a usable working state”, and that instead of repairing or refunding, “they keep repeating policy language and asking me to wait.”
An unrelated Trustpilot review from a different country, dated , is titled “False Refund Policy + Unusable macOS VPS – Terrible Service & Unfair Terms”.
Two further 1★ accounts from the same window are worth noting individually. A Netherlands customer describes a domain “abducted for no clear reason” on . And on , a UK customer titled their review “Avoid Them, they have legally flawed terms”, arguing UltaHost refused a statutory UK refund while its own terms place its UK entity under UK law — which connects directly to the refund-clause analysis in Round 5.
We name no customers. These are published consumer reviews and public comments, cited as anyone citing a review would, and we have not verified the underlying transactions — only that these accounts were published, when, and what they say.
7.7 Verify It Yourself — And Our Method
Every number in this round comes from a dataset you can open yourself.
| Source | What you can verify |
|---|---|
| Meta Ad Library | Every ad, its creative text, run dates, platforms and EU reach reporting |
| Trustpilot — ultahost.com | Live TrustScore, review count, and each review's arrival channel |
| UltaHost homepage | The three 4.9 badges and the 99.99% uptime guarantee |
| StatusGator | Independent outage history, including the June–July 2026 cluster |
| WebsitePlanet | The independently measured 43.78% uptime and #3752/3,860 ranking |
Method, including its limits
- ● Retrieval.Meta's Ad Library and Trustpilot both block automated reads — the Ad Library returns a bot challenge, and Trustpilot sits behind a web application firewall that rejected every direct request we made, including from residential addresses. We therefore retrieved both datasets through commercial scraping services on , and cross-checked the ad archive with two independent collectors.
- ● Sample limits. The Trustpilot analysis covers the 200 most recent reviews, spanning to — not the full 1,739. The company-level TrustScore and review count are Trustpilot's own published aggregates. Rating percentages in section 7.4 describe that 200-review sample and should not be read as lifetime figures.
- ● What we could not reach. Reviews before January 2026 were beyond the retrieval cap, so several older customer reports this site has tracked remain uncorroborated and unpublished. The ultahost.io profile did not resolve. We report these as gaps rather than filling them.
- ● No comparison group.We did not measure other hosting companies' advertising or review patterns, so nothing here should be read as a claim that UltaHost is worse than its competitors on these specific metrics. Every finding above is an absolute comparison against UltaHost's own published claims and Trustpilot's own published score.
- ● Archiving. We submitted every URL cited on this page to the Internet Archive so the claims stay checkable if the originals change. Nine of thirteen now hold a snapshot, including UltaHost's homepage with the 4.9 badges. Four could not be archived: the Ad Library search URL and the three individual Trustpilot review permalinks, all of which return an error to the Archive's crawler because Trustpilot and Meta block it — the same firewalls that blocked our own direct reads. For those four we retain our own dated copies of the retrieved records, and we flag here that a reader checking them later may find them altered or removed with no public snapshot to compare against.
7.8 Editor's Conclusion
Six rounds documented what happened to UltaHost's customers. Round 7 documents what UltaHost says to the next ones — and the two do not reconcile.
A 4.9 rating attributed to Trustpilot on a Trustpilot-branded badge, running in 101 paid advertisements, against a Trustpilot score of 3.6. A promise of 100% uptime in 58 ads whose run window contains an eleven-day storage failure the company itself acknowledged. Three different uptime numbers advertised simultaneously. And underneath the headline average, two distinct populations of customer: the 135 who were invited to review and rated the company 70% positive, and the 58 who showed up on their own and rated it 93% negative.
We should be equally clear about what we did not find. There is no evidence of fabricated review text. There is no evidence of purchased Facebook comments — we looked specifically, and the comment volumes are far too low and too critical to support it. Asking happy customers for reviews is legal and ordinary. The uncomfortable finding is not that UltaHost fakes its reputation; it is that UltaHost's reputation is assembled from the customers it chooses to ask, then advertised at a number the underlying platform does not show.
Round 6 concluded that the 4.9 is bought and the 3.4 is what you get. Round 7 can put it more precisely: the 4.9 is selected, the 3.6 is measured, and the 100% uptime was being advertised while the disks were still rebuilding.
What you can do right now
I bought based on the 4.9 rating or the uptime promise
Chargeback statutes by country + 4-step safe migration order
I'm about to buy UltaHost — should I?
Comparison with 6 reputable alternatives
I want to report misleading advertising
7 ready-to-send templates (ICANN, BBB, FTC, UK Action Fraud, and more)
I'm a journalist working on this story
Press kit: headline stats, pull quotes, verified facts, source documents
Round 7 Sources
Compiled August 4, 2026 from Meta's Ad Library and Trustpilot, both retrieved that day. Advertising counts and review-channel percentages are our own measurements of those public datasets; the method and its limits are set out in section 7.7. All information publicly sourced for accountability-journalism purposes. We document patterns and make no allegation of intent.