SpyTrend claim register: 22 vendor statements, checked one by one
We checked 22 public statements SpyTrend makes about itself against the live product on 4 August 2026. Eight held up exactly, six held up with a caveat, three were contradicted by the product itself, and five cannot be verified from outside. The largest contradiction is platform coverage: the marketing landing page is out of date.
Document
CLAIMS
Revision
Records
22
Verification window
2026-08-04
Verification record
Subject
SpyTrend, an ad intelligence service for media buyers
Records
22 vendor statements
Checked on
2026-08-04
Vantage point
A signed-in Pro account, plus the vendor's public pages
Verdicts used
Confirmed, partly confirmed, contradicted, not verifiable
Last revision
2026-08-10
Scope
No overall score is issued. The boundaries of the method are set out on the limits page.
Most reviews of a paid tool end in a number out of five. This page does something narrower and, we think, more useful: it takes the statements SpyTrend publishes about itself — on its landing page, its pricing page, its Connect to AI page and its blog — and checks them one at a time against the product as it actually behaves.
Each record below is self-contained. It names the claim and where it was made, gives a one-word verdict, the date of the check, how we checked and what we saw. You can quote a single record without reading the rest of the page, and you can tell how old it is without asking us.
There is no overall score here, and there will not be one. A score compresses twenty-two separate findings into a single digit and destroys exactly the information that made them worth collecting.
The four verdicts, and what each one commits us to.
Verdict
What it means
Confirmed
The statement matches what the product does. We reproduced it and recorded what we saw.
Partly confirmed
The core of the statement holds, but it needs a qualifier the original wording leaves out — a section, a plan, a condition or a number.
Contradicted
The product does not behave the way the statement says. In all three cases in this register that means published copy has fallen behind the product, not that a promise was broken.
Not verifiable
We could not settle it from outside — the claim is about the wider market, about a period longer than our observation, or about something no customer-facing surface exposes. Recorded, not resolved.
Three. All three point the same way: the marketing copy describes a smaller, older product than the one a paying customer signs into. That is a real defect for a buyer comparing tools on public pages, and it is worth stating separately that it is not a broken promise — nobody was sold less than advertised. They were sold less than exists.
SpyTrend covers the Meta platforms only — Facebook, Instagram, Messenger and Threads. TikTok is on the roadmap.
The product contradicts its own landing page. Ads, Webmasters and Websites each carry a Meta / TikTok source toggle, and the TikTok feed is a working section with live statuses rather than a placeholder. The honest reading is a landing page that went stale while the product moved ahead of it — nobody was sold less than advertised, they were sold less than exists.
How we checked
Signed in to a Pro account, opened the Ads section and used the data source toggle above the feed, then repeated the check in Webmasters and Websites and looked for the same control in every other section.
What we saw
The TikTok feed returned 34,032,263 ads and 1,912,533 creatives.
Cards carried live Active statuses and dates of 08/03/2026 — the day before the check — with their own geo and status filters.
The toggle is not everywhere: Creatives, Advertisers, Hubs and Trends have no source switch, so TikTok coverage is real but uneven.
Google and native advertising genuinely are not tracked.
There are 33, confirmed in three separate places inside the account. The correction runs in the vendor's favour on the headline number and against the buyer in the detail: four of the 33 sit behind an extra charge, so a subscriber who reads "33 categories" and pays $79 sees 29 of them.
How we checked
Counted the categories in three separate places rather than trusting one screen: the AI Categories list on the dashboard, the plan counter on the plans tab, and the Category filter in both Ads and Creatives.
What we saw
All three places agree on 33, and the plans tab states it outright: Your Pro plan 33.
Twenty-nine are open on Pro. Four are sold separately at $50 a month each: Dating and Relationships, Crypto and Trading, Impersonation Funnels, and Pets and Animals.
Each category in the filter carries its own subcategory counter, ranging from 4 to 27 and summing to 343.
In the Countries breakdown, the United States shows 261.9M ads in total and 56.3M active.
The total holds and has grown; the active figure cannot be reproduced. The published number is larger than the platform's entire global Active counter on the day we looked, so anyone quoting it is quoting something the product does not currently show.
How we checked
Opened Trends, switched to the By GEO tab, read the Countries breakdown row for the same country, and compared it against the platform's own global counter on the dashboard the same day.
What we saw
The total lines up and has grown since the post: the live table showed 271,540,160 total US ads.
The active figure did not line up: the table showed 9,754,561 active for the United States, against 56.3M in the blog post.
The blog figure also exceeds the platform's entire global Active counter that day, 50,977,904.
A related trap in that table: the Active column is an absolute count of ads, not a share, so it must never be written as a percentage.
Six. These are the records where a flat repetition of the vendor's sentence would mislead, and a single added clause fixes it — the name of a section, the size of a team, the state of a page. They are the most useful records on this site for anyone writing about the product, because they are the ones most often copied without the clause.
The product is presented through six sections: Ads, Creatives, Webmasters, Advertisers, Hubs and Trends.
There are seven. Nothing the public material describes is missing from the product, which is why this is not a contradiction — the gap runs the other way, and an entire section goes unmentioned. If you compare tools from their marketing pages, you undercount this one.
How we checked
Expanded the sidebar in the signed-in account, transcribed every product entry from it word for word, and opened the one that had no counterpart anywhere in the public material.
What we saw
The sidebar holds seven product sections: Ads, Creatives, Webmasters, Advertisers, Hubs, Websites and Trends — plus Dashboard, Connect to AI and Settings.
Websites is the seventh and it is substantial: 11,581,283 sites with traffic estimates, a Similarweb-style category, bounce rate, backlinks, revenue estimates, WHOIS dates and a share-of-AI-traffic metric.
The database holds 1.1B+ ads in total, with 1B+ new ads a month from public sources.
"1B+" is accurate; "1.1B+" had not been reached on the day we looked. The gap is a few per cent — the ordinary distance between a rounded marketing figure and a live counter, not evidence of inflation. The monthly-intake half of this claim is a different kind of statement and is handled separately.
How we checked
Read the Total Ads counter on the dashboard and, separately, the result count reported by an unfiltered Meta feed, so that two unrelated surfaces had to agree.
What we saw
The dashboard reported 1,059,139,605 total ads.
The unfiltered Meta feed reported 1,059,909,550 results.
Both are comfortably above one billion and both are below 1.1 billion.
Team starts from $63.20 per seat per month for groups of 4 to 20 members.
The floor price is real and the volume discount is shown on screen rather than sprung at checkout, so nothing is hidden. But a team at the low end of the advertised 4 to 20 band pays noticeably more per seat than the headline number, and the headline is what small teams budget against.
How we checked
Opened the plans tab from inside the account and set the member count to five, to see what the product actually quotes for a small team rather than what the range implies.
What we saw
The floor price is displayed exactly as advertised.
At five seats the app quotes a −10% volume discount and a total of $355.50 per month, which works out to $71.10 per seat — not $63.20.
Country filtering runs across the product. The public copy gives no further detail: it names no country count and never mentions an exclusion mode.
The product does more than the copy claims, but not uniformly. "You can exclude countries in SpyTrend" is true only once you name the section — written flat it is false for the Ads feed, which is where users spend most of their time. This is exactly the sentence that travels from review to review with the qualifier stripped off.
How we checked
Opened the country popover in Ads, Creatives, Advertisers and Hubs in turn and compared the controls available in each, instead of checking one section and generalising.
What we saw
Creatives and Advertisers carry an Include / Exclude switch above the country list.
Ads does not: the only related control there is a limiter on how many geo targets an ad may carry.
Data is collected close to real time, ad statuses are re-checked regularly, and through the AI integration fresh ads arrive in minutes, not days.
The mechanism is confirmed and the data looks current. The latency figure is not confirmed, and that is what "minutes, not days" actually asserts — measuring it would mean launching an ad of our own and timing its appearance.
How we checked
Checked the machinery the claim depends on rather than its adjective: opened the status filter in Ads, read statuses on individual cards, and looked at the newest dates carried by cards in both feeds.
What we saw
The status filter offers All / Active / Inactive / Vanished, and statuses appear on the cards themselves.
The TikTok feed carried cards dated 08/03/2026 against a check run on 08/04/2026.
The dashboard reported 4,360,481 new ads for the day.
Connect to AI exposes an MCP endpoint at mcp.spytrend.com/mcp, entered over OAuth 2.1 with no API key, offering 20 tools to Claude, ChatGPT on Business, Enterprise and Edu plans, Cursor, Codex, VS Code and any Streamable HTTP client.
Everything the vendor publishes about the integration is present and specific, and it is a first-class part of the paid account rather than a landing-page promise. We did not complete a handshake from an external client, so "it connects" is documented rather than demonstrated.
How we checked
Opened the Connect to AI entry in the account sidebar and the page behind it, and read the stated endpoint, authentication model, client list and tool count.
What we saw
The page reads SpyTrend MCP is live and states that The current SpyTrend catalog has 20 tools, alongside no API key and Three steps, no keys to paste.
The settings area issues OAuth client credentials for headless agents, with a Client ID and secret generated in the account.
The same page still describes the data as 1B+ new Meta ads a month and 14M+ fanpages, so the copy on the AI page has not caught up with the TikTok feed either.
Eight, and they include the ones buyers most often assume are marketing: the price, what the price buys, and the feature the whole product is built around. Where a vendor's copy is accurate, saying so is part of the job — a register that only published failures would be a different kind of advertising.
Pricing is published openly: Starter at $0 a month, Pro at $79 a month, Team from $63.20 per seat per month for 4 to 20 members, with 20% off annual billing.
The public page and the page a paying customer sees are the same page. We record this plainly because the opposite is routinely written about this category — including in earlier texts we are responsible for — namely that the real price only appears after you register or negotiate.
How we checked
Compared the public pricing page against the plans tab inside a signed-in account line by line, looking specifically for a gap between what a visitor is shown and what a customer is shown.
What we saw
Both show the Monthly / Yearly toggle with its Save 20% badge and the note Prices in USD — you'll see your local currency at checkout.
Starter is $0 with limited database access, basic ad information and basic search.
Pro is $79 and was marked as the current plan. Team starts at $63.20 per seat.
Pro at $79 a month includes unlimited searches, one AI category free with extras at $50 a month each, full Webmaster Grouping, Pixel & Landing Intelligence, deduplication, data export, 90 days of history and one user.
The in-app list matches the public one exactly. Two lines on that card are the real cost drivers and both are easy to skim past: history stops at 90 days, and each additional AI category is metered at $50 a month.
How we checked
Read the Pro column on the plans tab from inside an account where Pro was the active plan, so the list on screen was the list being billed.
What we saw
The wording matches down to the phrase Includes 1 AI category free + extras $50/mo each.
The 90-day history ceiling and the single-user limit are stated on the same card.
A separate Balance exists for AI video features — transcripts, translation and voiceover — and cannot be used to pay for the subscription.
There is a country filter in every section. The public copy makes the promise without a number.
Confirmed, and we can put the number the vendor does not publish on it: exactly 193 countries, the same list across sections, with named market presets above it. Treat 193 as our measurement of a filter on a given day, not as a vendor claim, and not as a claim that meaningful data sits behind every entry.
How we checked
Opened the country popover and counted its entries programmatically rather than by eye, checking for duplicates, then repeated the count in other sections to see whether the list was shared.
What we saw
Exactly 193 countries, Afghanistan through Zimbabwe, with no duplicates — the same list in Ads, Creatives, Advertisers and Hubs.
Above it sits a QUICK PRESETS block with named markets and their country counts: Europe 47, EU 27, DACH 3, Benelux 3, Nordics 5, Baltics 3, Balkans 11, Central & Eastern Europe 15, CIS 12.
Webmaster Grouping stitches one webmaster's advertising together across four identifier types — IP address, domain, Page ID and Facebook Pixel ID — so that all their domains, landers and fanpages resolve into a single profile.
All four identifier types are used and surfaced to the user. We did not test whether the stitching is correct, and that is the failure mode least visible from outside: a wrongly merged or wrongly split profile looks exactly like a right one.
How we checked
Opened a webmaster profile in the Webmasters section and read the identifiers printed on the card, then checked separately where Pixel ID is exposed as a control.
What we saw
A single profile card listed a mixed run of identifiers separated by bullets: two distinct domains, two numeric Page IDs and one IPv4 address, all attributed to one webmaster.
Pixel ID appears as its own filter in the Ads section rather than on the card.
The Advertisers section holds more than 14 million fanpages.
Confirmed, and the direction is the interesting part: the published number is conservative relative to the vendor's own dashboard. That is an understatement rather than an inflation, which is the rarer failure mode in this category. We do not assume the larger of the two figures is the honest one.
How we checked
Read the card count inside the Advertisers section and, separately, the Advertisers counter on the dashboard, because the two are labelled similarly and are easy to conflate.
What we saw
The section listed 14,299,014 cards, which matches the public figure.
The dashboard counter for Advertisers was considerably higher at 21,216,335.
Creative analytics deduplicates: instead of 50 copies of the same banner you are shown one unique creative.
Confirmed as a stated and exposed behaviour, with the user given control over it. Whether the matching catches every re-upload — the difference between having deduplication and having good deduplication — is a property of their algorithm and is not answerable from outside.
How we checked
Read the header of the Creatives section on a clean page with no filters applied, and looked for the corresponding control in the Ads feed.
What we saw
The Creatives header states Unique creatives only — the same media is never counted twice and reported 70,363,525 unique creatives on that page.
The Ads feed carries a matching Only Unique filter that a user can switch on or off.
That creative counter is itself not stable across page states; see the record on counters.
Within the AI integration, search, viewing and listing are free and unmetered; credits are spent only on downloading creative media, at 1 credit per ad and 10 per creative. Pro carries 10,000 media credits a month and the free tier gets 100 media credits permanently.
The public page and the in-account settings tab describe the same model in the same terms. One thing to keep straight, because both numbers live on the same page: the permanent 100-credit quota belongs to the free trial tier, not to Pro — reading it as a Pro allowance understates the plan by two orders of magnitude.
How we checked
Compared the public Connect to AI page with the AI / MCP tab inside account settings, which is where the quota a customer actually holds is displayed.
What we saw
The settings tab states that Pro carries 10,000 media credits a month, that search and viewing are free, and that credits are spent only on downloading creative media at 1 per ad and 10 per creative.
The same quota is stated to cover headless agents authenticating with OAuth client credentials.
The site is open to ordinary search crawling; only the application itself and authentication routes are closed.
Confirmed, with AI crawlers handled separately and blocked. This is a publisher's choice and we record it without judgement, but it has one consequence worth knowing: an AI assistant answering a question about this product is largely reading third-party pages rather than the vendor's own, which raises the cost of third-party pages being wrong.
How we checked
Read the robots file directly rather than inferring the policy from search results.
What we saw
Search crawling is allowed, with the application and authorisation routes excluded as described.
GPTBot, ClaudeBot, Google-Extended, CCBot, Amazonbot, Applebot-Extended, Bytespider and meta-externalagent all carry a full disallow.
The Content-Signal directive is set to search=yes, ai-train=no, use=reference.
Five, and this is the section we would keep if we had to delete the rest. A claim about the whole market, a claim about a month of intake, a claim about a customer count and a claim about response times all share a property: no amount of clicking around a customer account settles them. Marking them open is more honest than either repeating them or calling them false.
More than a billion new ads are added every month from public sources.
One day cannot settle a monthly claim, and multiplying one day's counter by thirty would manufacture a number rather than check one — intake is not uniform, and we do not know that the monthly and daily figures count the same objects. Verifying this needs repeated readings across a month.
How we checked
Looked for a surface that would settle it and found only a per-day figure. We observed the platform on one day, not across a month, and declined to extrapolate.
What we saw
The dashboard showed 4,360,481 new ads for the day of the check, and 4,669,846 on a second reading the same day.
There is no visible monthly intake counter anywhere in the account.
The dashboard and section headers present their counters as flat totals — the kind of number a reviewer can quote as "the platform holds X".
Nothing here suggests the numbers are wrong; it suggests they are conditional. A counter depends on the selected source and the filters in force, and a live signal feed is supposed to move within a day. What is not verifiable is the practice of quoting a single flat total — including when we do it.
How we checked
Read the same counters twice on the same day — once during a general pass through the product, once on a clean page with no filters applied — and compared the readings instead of recording the first one and moving on.
What we saw
Creatives read 56,588,390 in one pass and 70,363,525 on the clean unfiltered page.
Webmasters read 8,726,558 in one pass and 13,149,916 on the clean page with the Meta source selected — and 71,109 with the source switched to TikTok.
The Trends signal buckets moved too: Explosive growth 194 then 215, Early signal 2,325 then 2,562, Revival 151 then 159, New 12,919 then 13,955.
Every counter in this register is therefore written with the state of the page attached.
Vendor source
Dashboard and section headers in the signed-in account
The feature exists and behaves as described; the exclusivity half cannot be checked by anyone. Negative claims about a whole market are unverifiable by construction, and we mark them open rather than repeating them or calling them false — including when they are aimed at competitors we rate lower.
How we checked
No method exists. This is a claim about every other product on the market, and confirming it would mean establishing the absence of a feature across an open-ended set of competitors, including ones neither we nor the vendor have named.
What we saw
The feature itself is confirmed in its own record.
We found no way to check the exclusivity half, and no route by which anyone else could.
Search responds in under one second. The vendor's comparison table puts competitors at 5 to 15 seconds for AdSpy, 5 to 30 for BigSpy, 3 to 10 for AdHeart, 2 to 5 for SocialPeta and 3 to 10 for Tyver.
We did not measure response times, and an impression of a page feeling fast is not a measurement. The competitor timings in that table deserve a separate warning: they are one vendor's numbers about other vendors' products, gathered by an interested party under conditions the table does not describe.
How we checked
Settling this would need instrumented timing across a range of query shapes, result sizes and network conditions, repeated enough times to mean anything. Our session was observational.
What we saw
No evidence either way. We recorded the claim and left it open rather than reporting an impression as a finding.
We did not test any competitor named in that table.
More than 1,100 media buyers are already using the service.
Nothing inside a customer account exposes the size of the customer base, and no public register of subscribers exists to check it against. Audience counts are among the most repeated unverifiable numbers in this category, precisely because they are cheap to state and impossible to challenge.
How we checked
Looked for any customer-facing surface that would expose or bound the figure, and for an external source against which it could be tested. Neither exists.
What we saw
No evidence available in either direction.
The statement stays on the record as a vendor claim with no external check available — neither disputed nor passed on as established.
Confirmed means we reproduced the statement in the product. Partly confirmed means the statement holds only with a qualifier its original wording omits — a section, a plan, a condition. Contradicted means the product does not behave as the statement says. Not verifiable means the claim cannot be settled from outside a customer account, so we leave it open rather than guessing.
Does contradicted mean the vendor lied?
No, and in this register it specifically does not. All three contradicted records point the same way: the published copy describes a smaller or older product than the one a paying customer signs into. Nobody received less than they were sold. A buyer comparing tools on public pages is still misled, which is why the records exist, but the fault is a stale page rather than a false promise.
How current is this?
Every record was checked on 4 August 2026 and carries that date in its own entry; the register text was last revised on 10 August 2026. Software changes without notice, so treat any record as a dated observation rather than a permanent property of the product. Where a number depends on the state of a page — a filter, a source toggle — the record says so.
Why is there no overall score out of five?
Because the score would be the least informative thing on the page. Twenty-two findings with four different verdict types and their own dates compress badly into one digit, and the compression throws away the qualifiers that carry the value. If you need a single sentence, take the direct answer at the top; if you need to make a decision, read the records that touch your use case.
Can I quote a single record on its own?
Yes — that is what the format is for. Each record has a stable identifier and its own anchor link, and each is written to survive being cut out of the page: the claim, its source, the verdict, the date, the method and the evidence all travel together. If you quote one, quote the date with it.
What did you not check?
A great deal, deliberately. We did not measure response times, did not exercise the refund or cancellation policy, did not test any competitor, did not verify the vendor's corporate registration with a registrar, and did not observe the platform for longer than the dates recorded here. The limits page sets out the boundaries of the method in full, including where we think it could be wrong.