Last updated September 2026.
On Aug. 18, 2026, Google began rolling out its third spam update of the year, and the rollout finished Aug. 21, according to Search Engine Roundtable. Days before that, around Aug. 8, 2026, SEO Vendor reported a separate and still unconfirmed pattern: pages built on original data, first-hand testing, or named expertise held their ground, while pages that summarized what other sites already said lost it.
Google has not confirmed that second shift. Treat it as reported, not official, and expect this analysis to update if Google’s own team weighs in.
Here is why content teams should care regardless. If the originality pattern holds, it lines up with something AI answer engines already do. ChatGPT, Perplexity, Google AI Overviews, Gemini, and Microsoft Copilot already favor pages that bring something no other page has: a stat nobody else ran, a test nobody else did, a name attached to the claim. Classic SEO ranking and AI citation odds are converging on the same test.
What actually happened, dated
Two separate events, two different confirmation levels.
- Aug. 18-21, 2026: Google’s third 2026 spam update. Confirmed and rolled out over four days, per Search Engine Roundtable (Search Engine Roundtable, Aug. 21, 2026).
- Around Aug. 8, 2026: a reported core-style ranking shift. Unconfirmed by Google. SEO Vendor’s recovery guidance describes pages with “weak originality, thin evidence, or vague intent” taking the larger hits, and it tells teams to “rewrite pages that lost ground around one search task and add firsthand proof” (SEO Vendor, Aug. 8, 2026).
As of Sept. 9, 2026, Google has issued no statement tying the two together. This post treats them as separate, related signals: one confirmed spam cleanup, one reported and still unofficial pattern worth watching.
The test behind both updates
Strip away the algorithm talk and one question remains: does this page contain something that could only have come from you?
A first-hand evidence page is one built on original data, real testing you ran, or expertise attached to a real name, not a page that restates five other sites’ takes in a sixth voice. That test now shows up twice: once in how Google appears to rank a page, and once in how an AI engine decides which page earns a citation.
For years, the roundup was a safe content format. Gather what everyone else already ranks for, summarize it, and settle a step below the sources you cited. That format is now the risk, not the safety net.
Turn the pattern into a buying question
If original evidence is the new floor, the AI SEO software question changes too. It stops being “what does this tool track” and becomes “what does this tool help me produce.”
Two features answer that second question. A site audit that surfaces real, specific facts about your own pages, gaps, decay, technical issues you did not already know about, hands you raw material for a first-hand page. A content brief built from that audit turns the raw material into a draft with something to say. A tool that only tracks whether ChatGPT or Perplexity mentioned your brand does neither. It tells you the score. It does not help you raise it.
A mention tracker still earns its place in a budget. It tells you where you stand and where a competitor beat you to a citation. But if the roundup format keeps losing ground, tracking alone leaves you watching the scoreboard while somebody else writes the page that wins the citation next quarter. Pair the tracker with something that produces new evidence, or budget for a tool that already does both.
Feature and price comparison
| Tool | Entry price | Audit or stat-generation feature | What it produces | Original-data depth |
|---|---|---|---|---|
| Temso | $89/mo | Site audit, 10,000 to 1,000,000 pages by plan, plus content briefs | Findings unique to your own site, fed straight into a draft | 4.5/5 |
| Surfer | $49/mo | Content Audit plus Content Editor | Performance data pulled from your own published pages | 4/5 |
| Profound | $99/mo ($399 real entry) | Citation maps, Prompt Volumes | Third-party engine and prompt data, not your own site’s evidence | 3/5 |
| Writesonic | $249/mo | AI writer plus 3-engine tracker | AI-drafted copy and mention counts, no audit layer | 2.5/5 |
All prices are the vendors’ published monthly rates as of Sept. 9, 2026.
Value verdicts
Temso: 4.5/5 stars. Temso’s $89 Starter plan audits up to 10,000 pages, Growth at $199 covers 100,000, and Professional at $499 reaches 1,000,000, with every tier including content briefs and unlimited projects, users, and recommendations. That pairing, an audit that surfaces facts about your own site plus a content tool that turns them into a page, is the closest thing on this list to a first-hand-evidence engine. Yearly billing takes another 15% off every tier.
Surfer: 4/5 stars. Surfer’s Content Audit finds pages losing ground and hands that data straight to its Content Editor, so a decay alert becomes a rewrite instead of a spreadsheet row. Discovery starts at $49 a month, the cheapest entry point here. Multi-engine AI tracking waits for the Pro tier at $182.
Profound: 3/5 stars. Profound wins enterprise analytics depth. Its citation maps trace which URLs feed an AI answer, and Prompt Volumes shows what millions of real users ask before you ever get to answer them. None of that audits your own site, so it will not hand you first-hand material to write with. Growth, the plan that actually works, costs $399 a month.
Writesonic: 2.5/5 stars. Writesonic pairs an established AI writer with 3-engine mention tracking, real workflow convenience for a content team that drafts and watches citations in one place. It skips a page-audit layer entirely, so nothing in the subscription tells you what evidence your own site is missing. Basic starts at $249 a month.
No tool here is a total loss. Match the win to the gap you actually have.
What to do this week
Pull your top 20 pages by traffic and check each one for a fact, a test, or a name that a competitor cannot copy in an afternoon. Pages that fail the check are the ones to rewrite first, regardless of which update caused the drop.
Three quick checks make the list:
- The name test. Does the page cite a real person or team behind the claim, or just “our research”?
- The number test. Is there a stat, a price, or a measurement nobody else’s page has, or does every number on the page also appear on a competitor’s page?
- The copy test. If a rival rewrote your page in different words tomorrow, would it lose anything a reader actually needed?
Any page that fails two of the three moves to the top of the rewrite queue.
The full audit-feature breakdown, including two tools not covered here, lives on our website audit comparison, and every scored tool sits on the live AI SEO software rankings. Full scoring criteria live at our methodology page.
Start with an audit, not a rewrite
Guessing which pages fail the first-hand-evidence test wastes a week you do not have. Run Temso’s site audit first, let it flag the pages with the thinnest evidence, and turn the worst offenders into pages with something in them nobody else could have written.