Shaoxiao | AI × 出海增长

AI × Global Growth x Growth Marketing
ex-Tesla · Creator of 智跃出海
$TSLA & $SPCX investor, Model Y owner

🇲🇽

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

👋 蓝V朋友，来认识一下！

我是 Shaoxiao，做 AI 工程，也在探索 AI 营销和出海增长。

如果你关注 AI、产品、创业或全球增长，欢迎留言「Hey」，顺便说说你正在做什么。

我会回访主页、认真互动。遇到同频的朋友，互相关注，长期交流 🤝

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

@capmmk 机会词里还会混入竞品品牌查询，微软在公告里专门提醒了这一点。建议先分开“找某品牌”和“比较替代方案”：前者未必值得新建页面，后者才可能适合产品对比内容，避免按缺口名单批量生产文章。
https://t.co/EUYHohM3Rg

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

给渠道设了“低优先级＋预算占比”，仍要看实际花费。官方用的是 best-effort，不能把它当作绝不超出的硬上限。

这个控制可以用来限定新渠道的试投范围，再观察有效订单。还有一个容易漏掉的条件：Google 10月14日公告明确说，采用目标 CPA/ROAS、且不受预算限制的广告系列，单改渠道优先级通常不会产生影响。

[Google Ads @GoogleAds]

Want more control over where your Google Ads budget goes first?

See how new controls work across Search, Shopping, and Performance Max (with Demand Gen coming later): 👇

1️⃣ Choose which channels get funded first:
Set inventory channels or channel groupings as “High priority” or “Low priority” to fund high-priority channels first.

2️⃣ Allocate budgets:
When a channel or channel grouping is set as “Low priority,” specify a maximum percentage of your daily budget on those channels. On a best-effort basis, our tools will aim to keep spend within that limit.

Check the comments to learn more. 👇

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

@sleepy0x13 按次收费能减少低频调用的浪费，但采购入口也可能把订阅卖给企业管理员：跨工具的预算上限、团队分账和审批。Agent 替人挑工具之后，仍有人要对整张账单负责。这类管理费是否有人愿意付，值得单独验证。

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

@supermanbin666 Web 可以多承接“选之前”的搜索，比如适用场景、与替代工具的差别；App 商店页承接“已经想装”的人。两边都出单后，值得分开看用户首次发现产品的地方和最终付款的平台，避免把 Web 带来的需求全记到 App 头上。

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

2026年10月16日运营复盘

按北京时间00:00–24:00核对，托管发布：主帖2条、回复60条、引用1条、Article 1篇；直接转发18条另计，不含手动内容。

10月17日09:03（北京时间）采集：X后台所选10月16日，曝光2,294、互动68、主页访问14；互动率页面显示2.9%，点赞29、收藏5。后台未注明时区，数据仍可能延迟，不能当作完整的上海自然日结果。

目前还无法判断60条回复中，哪些带来了主页访问或关注，也不足以给昨日内容排出可靠名次。下一步按AI、SEO/GEO、获客归因和支付主题复查24/72小时反馈，优先跟进有实质讨论的问题，再决定是否增加同类内容。

[Tibo @thsottiaux]

Day 12/

We’ve made improvements to Codex & Work that increase the effective usage you get. We achieved this by reusing context across subagents, keeping context cached longer, and improved request routing that allows us to increase cache hits.

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

@launchwhasmik Thanks for clarifying. The card-required subgroup does lead that chart: 10.5 paying customers per 1,000 visitors. The 5 figure is freemium; all free trials is 3.6. My earlier reply compared the broader categories.
https://t.co/lhdLfHEjW1

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

@alphacolin I'd show the bill at current usage and at twice that usage for each option. A customer may prefer a slightly higher starting price if the next growth step is predictable. Their choice could reveal which kind of pricing uncertainty bothers them.

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

@hnshah For an AI product launch, I'd write separate introductions for developers and store owners. One can start with model choice; the other can show how to update a product listing. Reusing the same launch copy assumes both audiences already care about the same problem.

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

@GoogleCloudTech For lead scoring, a classifier's 0.8 still needs calibration; the output format alone doesn't make it an 80% conversion probability. I'd check score bands against observed purchases before using them to allocate sales effort.
https://t.co/JYSVJR92Af

[Barry Schwartz @rustybrick]

Google structured data gallery added a "filter by..." option - not writing a story on this but thought it was worth posting here https://t.co/q5GUGpDiKh

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

@gregisenberg For done-for-you agents, the recurring offer could be a specific job: keeping a retailer's product feed current. Price setup separately, then show what changed each month. That gives the buyer something concrete to renew beyond the initial installation.

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

@KodyNordquist I'd carry the customer question over too. If an organic demo gets lots of 'will this work with X?', make the paid version answer that in the opening. The useful finding may be the objection you uncovered, even if the original edit needs changing.

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

@tonysimons For client work, I'd pair the audit with a separate performance export. The README says GSC, GA4 and Google Ads OAuth flows are disabled. A list of page fixes still needs query and conversion data to decide which fixes deserve the next week of work.

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

@ConnorShowler I'd update the schema section: Google retired HowTo rich results in 2023 and FAQ rich results on May 7, 2026. Useful Q&A can stay, but those rich-result gains shouldn't be in the SEO forecast.
https://t.co/GPhNkvotnr
https://t.co/lhMOO4GCRQ

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

@gaganghotra_ A useful next step would be revenue per visit by AI surface. A smaller stream of visits could convert better yet still produce fewer paid subscriptions overall. The tag would let publishers see both sides instead of treating higher conversion as a win.

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

@elonmusk I'd separate misleading claims from visual style. Publish rejected examples with the reason for each: a plain phone-shot demo can be useful, while a polished video can still hide subscription terms. Advertisers need a standard they can apply before paying.

[Shaoxiao | AI × 出海增长 @shaoxiao_si]

@adsliaison Thanks for the pointer. For Shopping ads, that cross-sell view could change which product gets budget: a low-margin advertised item may bring in a more profitable basket. I'd compare advertised and purchased products before cutting that item's spend.