GEO's First National "Report Card" Is Out: CAICT Delivers Its Inaugural Service Ratings — and "What Tier Are You?" Becomes the Only Question That Matters
Two weeks ago, we covered the launch of China's first Generative Engine Optimization service capability assessment by the China Academy of Information and Communications Technology (CAICT). At the time, we called it the first national-level capability baseline for GEO — and said the first cohort would carry the most weight.
Now the first report card has been handed out.
On October 21, CAICT's Telecommunication Terminal Labs announced that Baisou AI, a GEO service under Shanghai Yibaisou Technology, completed the "GEO Capability Completeness" evaluation on October 10 — and earned Excellence tier, the highest grade in the current assessment system. It is the first GEO provider in the industry's history to receive a top rating from a national-level third-party evaluation.
Don't underestimate this piece of paper. It marks the industry's formal entry into the "certified practice" era: providers used to win you over with decks and case stories; from now on, you only need to ask one question — "What tier did CAICT rate you?"
What the Exam Actually Tests
The "GEO Capability Completeness Assessment Specification," drafted under CAICT's terminal lab, is not a rubber-stamp registration. It is a full-chain practical examination spanning six core dimensions:
1. Knowledge base construction — has the brand knowledge system actually been built, or is it just a pile of placement records? This foundation item alone filters out pseudo-GEO firms that only distribute press releases.
2. Intent recognition and generation — can the provider decompose how users actually phrase questions to AI (long-tail queries like "which is best" or "how do I choose"), and produce content AI is willing to cite — rather than keyword-stuffed articles?
3. Distribution channel management — where does content get placed, what weight do those channels carry, and is everything traceable? The gap between a provider with 12,000 channels and one with a dozen becomes painfully visible here.
4. Monitoring and effect analysis — is there 24/7 real-time monitoring tracking first-recommendation rates, sentiment fluctuations, and anomaly alerts, with weekly and monthly attribution reviews? Firms that deliver a one-off screenshot report fail this dimension.
5. Strategy iteration — does monitoring data feed back into content direction and source structure, closing the loop? GEO is not a one-shot deal; a service without iteration capability decays rapidly as AI models upgrade.
6. Full-chain security controls — content safety, data governance, and user rights protection, from technical foundation through delivery. This is the dimension regulators care about most — and the death line for black-hat tactics.
The six dimensions add up to a clear logic: technical foundation, business delivery, risk control, value closure — a full-chain examination with no dimension that money can fudge.
Three Tiers: Far Harsher Than Pass/Fail
The assessment returns three grades: Basic, Complete, and Excellence. The genius of this design is that instead of issuing a one-time "admission ticket," it makes the entire industry's capability gaps visible.
In the past, choosing a GEO provider meant facing a market where everyone claimed to be a leader, everyone had success stories, and every deck was prettier than the last. How bad was the information asymmetry? As we documented in a previous article, the industry can't even agree on its own market size — four conflicting estimates with a 30-fold spread.
Three tiers change the game. You say you're a market leader? Show your rating. Basic means you barely passed. Complete means your system works end to end. Excellence means all six dimensions survived full-chain scrutiny. Going forward, the first page of any procurement memo can simply be a tier comparison table.
Zoom Out: Standards, Certification, and Platforms Are Closing the Triangle
Connect the dots across the past month and a clear governance structure emerges:
The rule-makers — the China Advertising Association of Commerce's GEO group standards (T/CAACCHINA 001–005), published in September with a launch event in Shanghai on October 16, answering "what does a compliant GEO service look like."
The certifiers — CAICT's capability completeness assessment, which released its first results today, answering "who is qualified, and how good are they."
The gatekeepers — platform-side GEO manipulation filters, announced by DeepSeek on October 17, answering "which content never even reaches the AI's answer."
Rules, certification, and access control — three lines closing in. For black-hat operators it's a noose; for legitimate brands and providers, it's a guardrail. The era of bad money driving out good is ending. What comes next is good money competing on speed.
Three Actionable Takeaways for Brands
First, if you're selecting a provider now, write the tier into your tender requirements. The selection process can be simplified to: ask for the CAICT rating first, review industry cases second, negotiate price last. Don't reverse the order — price talks belong after qualification screening, not before.
Second, if you're already working with an agency, ask for their "exam plan." When do they intend to sit the assessment? If they have no intention of participating at all, reconsider what that says about their confidence in their own capabilities. The assessment is openly enrolled — the consultation email (zjjh@caict.ac.cn) is public. "Can't take the exam" is not an available excuse.
Third, if you're building in-house, use the six dimensions as a self-audit checklist. Not every company needs to outsource GEO. If you're doing it yourself, the six dimensions are a ready-made capability map: knowledge base, intent recognition, channel management, effect monitoring, iteration mechanism, security controls. Score yourself on each; your weakest dimension is next month's improvement priority.
The Bottom Line
The true sign of an industry's maturity was never the size of its market numbers. It's when the power to judge who's good transfers from the practitioners themselves to an independent third party.
GEO completed the full journey — from concept hype, to standard-setting, to a national assessment's first results — in just over a year. For brands, this is the best possible news: you can finally replace a hundred pages of unverifiable slides with one checkable certificate.
The next time a provider pounds their chest and claims "we're an industry leader," just smile and ask: "Did you get Excellence?"
FAQ
Q: Is the CAICT assessment mandatory? Can I still use a provider that hasn't participated?
A: It's not a license requirement — participation is voluntary. But it is currently the only GEO capability evaluation backed by a national research institution, and now that the first results are public, the market will quickly stratify into "certified" and "uncertified." Not participating doesn't mean incapable — it means you bear the full cost of verifying capability yourself, while assessed providers have had all six dimensions examined end to end. Pragmatic advice: prefer providers who have sat the assessment; if you must use one who hasn't, demand checkable evidence for each of the six dimensions.
Q: Baisou AI is the only Excellence-tier company — does that mean everyone else falls short?
A: No. This is the first cohort. Enrollment is continuous, and more providers will complete the assessment and publish their tiers over time. The first certificate establishes a benchmark, not a final verdict. Judge any provider by "did they participate, and what tier did they get" — not by whether they were first. Also note: ratings carry time pressure. AI platforms and industry practice evolve fast, and the need to maintain a tier will force providers to keep investing.
Q: How does this assessment relate to the CAAC group standards? Do they conflict?
A: They complement each other. The group standards (T/CAACCHINA 001–005) solve the "shared language" problem — what GEO is, how services are priced, what makes corpus trustworthy, how processes stay compliant. The CAICT assessment solves the "capability verification" problem — empirically testing a provider's real delivery across six dimensions. Think of it this way: the standard is the textbook; the assessment is the exam. Only when textbooks are unified do exam scores become comparable.