主题
推荐信是留学申请材料中少数来自第三方视角的评价,其客观性往往比申请者自述更有说服力。2026 申请季,推荐信生态出现两个显著变化:一是部分顶尖项目开始尝试视频推荐(Video Recommendation)和同行评价(Peer Review),传统纯文本推荐信不再是唯一形式;二是标化回归后,硬分数重新成为筛选门槛,推荐信在"够线"申请者之间的区分作用进一步放大——一封具体的、有比较性评价的推荐信,可能直接决定你能否从数千名"三维达标"的竞争者中脱颖而出。
本文将系统讲解推荐人选择策略、套磁方法、内容框架、2026 新趋势、waive rights 决策和提交方式,并提供完整模板和对比表。
一、核心概念:推荐信在申请中的真实权重
1.1 推荐信权重因项目而异
| 申请层次 | 推荐信权重 | 数量 | 组合建议 | 其他核心材料 |
|---|---|---|---|---|
| 本科申请 | 10-15% | 2-3 封 | Counselor + 学科教师 | GPA、标化、课外活动 |
| 授课型硕士 | 15-20% | 2-3 封 | 1-2 学术 + 1 实习 | GPA、标化、实习经历 |
| 研究型硕士 | 20-25% | 3 封 | 2 学术科研 + 1 补充 | GPA、科研产出、SOP |
| 博士申请 | 25-35% | 3 封学术 | 全部学术/科研推荐人 | 科研产出、SOP、导师匹配 |
| MBA 申请 | 20-25% | 2 封 | 1 学术 + 1 职业 | GMAT、工作经历 |
推荐信的"隐性价值"
在 GPA 和标化相近的竞争者中,推荐信是区分申请者的关键变量。招生官读推荐信时最关注两点:具体事例(不是"该生很优秀",而是"他在我的课上提出了一个我从未见过的解法")和比较性评价(如"在我教过的 500 名学生中排名前 3%")。空洞的赞美没有信息量,具体的对比才有说服力。
1.2 2026 推荐信的三个新变化
2026 推荐信生态变化
- 视频推荐兴起:部分商学院和硕士项目开始接受或要求短视频推荐(60-90 秒),推荐人面对镜头口述评价,更难"代写",真实性更高
- 同行评价试点:少数项目(尤其 MBA)引入 peer review,要求同龄人/同事评价你的领导力和协作能力
- 第三方平台验证:Interfolio 等平台的使用更普及,推荐信的提交和验证流程更规范化,"自己写自己交"的空间被压缩
二、推荐人选择策略 三类型对比
2.1 三大推荐人类型深度对比
| 维度 | 学术推荐人(教授) | 科研推荐人(导师/PI) | 实习推荐人(上司) |
|---|---|---|---|
| 适合项目 | 硕士/本科/博士 | 博士/研究型硕士 | 授课型硕士/MBA |
| 核心价值 | 证明学术基础和学习能力 | 证明研究能力和潜力 | 证明实践能力和职业素养 |
| 关系基础 | 上课、课程项目、助教 | 长期科研合作、论文 | 实习期间直接汇报 |
| 获取难度 | 中等 | 较低(已有合作关系) | 较低(实习已建立关系) |
| 内容深度 | 取决于互动频率 | 通常最深 | 取决于项目含金量 |
| 国际可信度 | 海外交换教授更高 | 有国际合作的更高 | 知名公司更高 |
| 常见风险 | 大课教授了解不深 | 导师太忙写得潦草 | 上司不熟悉学术评价标准 |
2.2 推荐人选择决策树
text
shiki
你申请的是博士/研究型硕士吗?
├── 是 → 优先选择:
│ ├── 1. 科研导师/实验室 PI(最了解你的研究能力,权重最高)
│ ├── 2. 核心课教授(最好与目标方向相关,有课程项目合作)
│ └── 3. 第二位科研合作者或海外交换教授(增加多样性)
│
└── 否(授课型硕士/本科)→ 你的方向是?
│
├── 理工科(CS/DS/工程)
│ ├── 1. 科研导师或项目指导老师(有成果产出优先)
│ ├── 2. 核心课教授(有课程项目合作的优先于纯大课)
│ └── 3. 实习技术主管(有大厂/知名公司实习时)
│
├── 商科(金融/BA/MBA)
│ ├── 1. 实习直管上司(投行/券商/咨询等,能评价软技能)
│ ├── 2. 核心课教授(量化课程优先于纯管理课)
│ └── 3. 第二位实习主管或社团指导老师
│
└── 人文社科
├── 1. 论文导师/研究项目指导教授
├── 2. 核心课教授(有写作样本可引用)
└── 3. 实习/田野调查指导者2.3 选择核心原则
推荐人选择五大原则
- 了解程度 > 头衔:一位深度了解你的副教授,远胜于一位只见过你一面的院士
- 学术相关性:推荐人研究领域应与申请方向相关,申 CS 找算法教授比找操作系统教授更对口
- 维度多样性:3 封推荐信应覆盖不同维度(如 1 学术基础 + 1 科研能力 + 1 实践能力),避免三封都说"成绩好"
- 比较性评价能力:优先选择教过/带过很多学生的推荐人,他们能给出"top X%"的对比
- 可靠性:选择守时、负责任的推荐人,推荐信晚交是常见且致命的失误
最大的推荐信错误
很多申请者执着于找"大牛"——院士、院长、公司高管。但如果这位大牛并不了解你,推荐信只会是:"该生在我课上表现不错,是个好学生。"这种泛泛评价不仅无益,反而暴露你没有真正的学术关系。一位了解你的年轻教授写的具体推荐信,远胜于一位不了解你的诺奖得主写的模板信。 招生官能一眼看出推荐信的"含金量"——具体的、有细节的、有对比的,才是真推荐。
三、套磁推荐人策略与邮件模板
3.1 套磁时间线
| 时间节点 | 任务 | 目的 |
|---|---|---|
| 申请前 3-4 个月 | 初次联系推荐人,面谈或邮件请求 | 确认推荐人意愿 |
| 申请前 2-3 个月 | 提供完整资料包,沟通推荐信重点 | 帮推荐人有素材可写 |
| 申请前 1 个月 | 发送各校截止日期和提交链接 | 确保推荐人知道时间节点 |
| 截止前 1 周 | 第一封提醒邮件 | 温和提醒 |
| 截止前 3 天 | 第二封提醒邮件(如未提交) | 紧急提醒 |
| 截止前 1 天 | 电话/微信提醒(如仍未提交) | 最后补救 |
3.2 首次请求邮件模板
shiki
Subject: Recommendation Letter Request - San Zhang (Fall 2026 Application)
Dear Professor Chen,
I hope this email finds you well. I am San Zhang, a senior in your
Computer Science program at Peking University. I took your Advanced
Algorithms course (Fall 2024, grade: A+, ranking: 1/180) and your
Machine Learning course (Spring 2025, grade: A), and I am currently
working on an independent research project under your supervision
on efficient attention mechanisms for long-context NLP.
I am applying to MS in Computer Science programs for Fall 2026
enrollment, with a focus on natural language processing. My target
schools include CMU, Stanford, and UIUC, with application deadlines
ranging from December 15, 2025 to January 5, 2026.
I would be honored if you would be willing to write a strong letter
of recommendation on my behalf. During your courses and our research
collaboration, I believe I have demonstrated both my technical
abilities and my research potential — particularly through my
algorithm design work and the sparse attention project we are
currently pursuing (which has improved inference speed by 40% on
sequences over 4K tokens).
To assist with your letter, I have prepared a packet including:
- My current CV (attached)
- My transcript (attached)
- A draft of my Statement of Purpose (attached)
- A summary of my research progress and key results (attached)
- A list of application deadlines and submission links
I understand that writing recommendation letters is a significant
time commitment. If you feel you cannot provide a strong
recommendation, I completely understand. Please let me know either
way, and I would be happy to discuss this in person at your
convenience.
Thank you very much for your time and consideration, and for your
guidance throughout my undergraduate studies.
Best regards,
San Zhang
Computer Science, Peking University
san.zhang@pku.edu.cn | +86 138-xxxx-xxxx请求邮件的关键技巧
- 给推荐人"退路":明确说"如果您觉得无法写一封 strong letter,我完全理解"。这比强求一封勉强推荐更好——一封"还行"的推荐信可能伤害你的申请
- 主动提供素材:不要让推荐人从零开始回忆,附上简历、成绩单、SOP 草稿和研究进展总结
- 点明你希望突出的 2-3 个要点:如"希望您能重点提到我在算法课上的创新解法和科研中的问题解决能力"
- 用"strong letter"而非"letter":这是行业暗语,提醒推荐人你要的是有力推荐而非泛泛之谈
3.3 推荐信资料包清单
为每位推荐人准备一个完整的资料包,大幅提升推荐信质量:
| 资料 | 作用 | 必要性 |
|---|---|---|
| 最新简历 | 让推荐人了解你的全貌 | 必备 |
| 成绩单 | 提供具体的课程和成绩数据 | 必备 |
| SOP/PS 草稿 | 让推荐信与你的叙事呼应 | 必备 |
| 截止日期与提交链接清单 | 确保推荐人按时提交 | 必备 |
| 你希望突出的 2-3 个要点 | 引导推荐信重点 | 强烈推荐 |
| 你与推荐人合作的回忆笔记 | 帮推荐人回忆细节 | 强烈推荐 |
| 目标项目说明 | 让推荐人针对性写作 | 推荐 |
四、推荐信内容框架
4.1 推荐信结构占比
一封高质量推荐信应包含以下要素,比例分配如下:
| 结构部分 | 占比 | 内容要点 | 常见错误 |
|---|---|---|---|
| 推荐人自我介绍 | 10% | 姓名、职位、与申请者关系、认识时间 | 过长,喧宾夺主 |
| 总体评价 | 15% | 推荐力度 + 比较性定位(top X%) | 只说"优秀"无对比 |
| 具体事例 1 | 20-25% | 学术/科研能力的具体展示 | 流水账无细节 |
| 具体事例 2 | 20-25% | 另一维度能力(领导力/创新/韧性) | 与事例 1 重复 |
| 对比分析 | 10-15% | 与同龄人对比,提供参照系 | 缺失对比 |
| 总结推荐 | 5-10% | 重申推荐力度 + 对目标项目的信心 | 结尾无力 |
4.2 推荐信"金句"示例
学术推荐信金句库
比较性评价类:
- "In my 15 years of teaching at Peking University, San ranks among the top 3% of the approximately 800 students I have taught."
- "She is the strongest undergraduate researcher I have mentored in the past decade — comparable to my best PhD students at the same career stage."
具体能力类:
- "His senior thesis demonstrated a level of analytical sophistication that I typically see in second-year PhD students."
- "When her initial implementation failed, she methodically diagnosed the issue and designed a hybrid approach that achieved 94% of dense performance with 40% less computation."
个人品质类:
- "What distinguishes San is not just his technical skill, but his willingness to help peers — he organized weekly study groups that improved the performance of several struggling students."
- "She persisted through three failed experiments before arriving at a novel solution, demonstrating the resilience essential for doctoral research."
实习推荐信金句库
超出预期类:
- "San consistently delivered work that exceeded expectations, taking on responsibilities typically given to full-time analysts."
- "I stopped double-checking his work after the first month — a level of trust I have extended to only one other intern in my 12-year career."
具体影响类:
- "His analysis identified a 15% valuation gap that saved our client approximately $75M — a remarkable contribution for a summer analyst."
- "She built a recommendation framework that was adopted by 15+ engineers and reduced API response time from 200ms to 50ms."
软技能类:
- "San has a rare ability to translate complex analysis into actionable business recommendations — his final presentation was at the level of a second-year analyst."
- "She coordinated teams across Hong Kong, New York, and London, often working until 2 AM but never compromising on accuracy."
推荐信的"比较性评价"原则
招生官最看重推荐信中的比较性评价——推荐人将你与其他学生/实习生对比。"He is a good student"没有信息量;"He ranks in the top 3% of 800 students I have taught in 15 years"则极具说服力。如果你帮推荐人起草推荐信,务必包含这类比较性数据。推荐人教过的学生越多、教龄越长,"top X%"的含金量越高。
五、完整推荐信模板
5.1 学术推荐信模板(CS 方向,核心课教授)
shiki
To: Admissions Committee, Computer Science Department
From: Prof. Wei Chen, Department of Computer Science, Peking University
Date: November 15, 2025
Re: Recommendation for San Zhang (MS in Computer Science, Fall 2026)
Dear Admissions Committee,
It is with great enthusiasm that I recommend San Zhang for admission
to your MS in Computer Science program. I have known San for two
years as his professor in two core courses — Advanced Algorithms
(Fall 2024) and Machine Learning (Spring 2025) — and as his academic
advisor for an independent research project on efficient attention
mechanisms. In my 15 years of teaching at Peking University, San
ranks among the top 3% of the approximately 800 students I have
taught.
San first distinguished himself in my Advanced Algorithms course,
where he achieved the highest score (97/100) among 180 students.
More impressively, during the algorithm design competition, San
proposed a novel approach to the weighted interval scheduling
problem that reduced time complexity from O(n log n) to O(n) for a
special case — a solution I had not seen in any textbook and later
incorporated into my lecture notes.
In my Machine Learning course, San's final project stood out for
its technical depth. He implemented a Transformer architecture from
scratch using only NumPy, and systematically analyzed how different
positional encoding schemes affected performance on long sequences.
His analysis revealed that sinusoidal encodings degraded
significantly beyond 512 tokens — a finding that led to his
independent research project on efficient attention mechanisms,
which I now supervise.
In this ongoing research, San has demonstrated the key qualities of
a successful graduate researcher: the ability to identify meaningful
problems, design rigorous experiments, and persist through setbacks.
When his initial sparse attention implementation failed to match
dense baseline performance, he methodically diagnosed the issue
(insufficient gradient flow through sparse connections) and designed
a hybrid approach achieving 94% of dense performance with 40% less
computation. I expect this work to result in a workshop paper
submission within the next few months.
Beyond technical abilities, San is a natural leader and mentor. He
voluntarily organized weekly study groups for struggling students,
and his patient explanations helped several students improve their
grades by a full letter. This combination of intellectual ability,
research potential, and interpersonal qualities makes me confident
San will thrive in your program.
I recommend San Zhang with my strongest possible endorsement. He is
the type of student who elevates the quality of any program he
joins. Please contact me at wchen@pku.edu.cn for further information.
Sincerely,
Wei Chen
Professor of Computer Science
Peking University
wchen@pku.edu.cn5.2 实习推荐信模板(商科方向,实习主管)
shiki
To: Admissions Committee, MIT Sloan School of Management
From: Michael Liu, Vice President, Goldman Sachs Hong Kong
Date: November 20, 2025
Re: Recommendation for San Zhang (MBAn Program, Fall 2026)
Dear Admissions Committee,
I am writing to strongly recommend San Zhang for admission to the
MIT Master of Business Analytics program. As Vice President in the
TMT Investment Banking group at Goldman Sachs, I supervised San
directly during his summer analyst internship from June to August
2025. In my 12 years at Goldman, San is one of the top three
undergraduate interns I have worked with — and the only one I have
trusted with building financial models for live client presentations.
San's most impressive contribution was his work on a $500M
acquisition in the EdTech sector. When our existing DCF model proved
inadequate for valuing the target's subscription-based revenue, San
independently researched and implemented an LTV/CAC framework
typically used by growth equity investors. His analysis revealed
that the target's projected customer retention rate was overly
optimistic — a finding that prompted our MD to renegotiate the offer
price downward by 15%, saving our client approximately $75M. This
level of analytical initiative is remarkable for a summer analyst.
San also demonstrated exceptional ability under pressure. During the
final week of a cross-border M&A deal, he maintained six financial
models simultaneously while coordinating teams across Hong Kong, New
York, and London. His attention to detail was such that I stopped
double-checking his work after the first month — a level of trust I
have extended to only one other intern in my career.
What sets San apart is his ability to translate complex analysis
into actionable business recommendations. In his final presentation,
he articulated a clear strategic narrative about the EdTech sector's
unit economics, supported by comparative analysis of 12 comparable
transactions. Several senior bankers commented his presentation was
at the level of a second-year analyst.
San's combination of quantitative rigor, business acumen, and work
ethic makes him an ideal candidate for the MIT MBAn program. I
recommend him without reservation. Contact me at michael.liu@gs.com
or +852-xxxx-xxxx for any questions.
Sincerely,
Michael Liu
Vice President, TMT Investment Banking
Goldman Sachs (Hong Kong)六、Waive Rights 详解
6.1 什么是 FERPA Rights
根据美国《家庭教育权利和隐私法》(FERPA),学生在入学后有权查看自己的教育档案,包括推荐信。但如果在申请时签署 waive rights,就放弃了这个权利。
6.2 Waive vs Not Waive 对比
| 维度 | Waive(放弃查看权) | Not Waive(保留查看权) |
|---|---|---|
| 推荐信可信度 | 高——招生官认为是真实评价 | 低——可能怀疑推荐人因压力不敢说实话 |
| 推荐人态度 | 更愿意写真实评价 | 可能不愿写或写得更保守 |
| 招生官印象 | 正面——申请者自信 | 负面——申请者可能不信任推荐人 |
| 推荐信质量 | 通常更具体、更坦诚 | 可能更模板化、更保守 |
强烈建议 Waive Rights
始终选择 waive your right to view. 招生官明确表示,未 waive 的推荐信权重会大幅降低——因为他们无法确定推荐人是否因知道你会看到而"手下留情"。即使你与推荐人关系很好,也请 waive。Waive rights 传递的信号是:"我信任我的推荐人,我对我的申请材料有信心。"
6.3 担心推荐信质量的正确做法
如果你担心推荐人写得不好,正确做法不是不 waive:
- 请求阶段就确认:问推荐人"您是否愿意为我写一封 strong letter of recommendation?"——对方犹豫就换人
- 提供详尽素材包:让推荐人有足够信息写出具体有力的推荐信
- 提前沟通要点:告诉推荐人你希望重点突出的 2-3 个方面
- 提供草稿:在中国学术环境中,为推荐人准备草稿是普遍做法,推荐人修改后署名提交
- 而不是:不 waive rights 然后偷偷查看——这会损害信任
七、2026 推荐信新趋势 2026趋势
7.1 视频推荐(Video Recommendation)
2026 部分顶尖商学院和硕士项目开始尝试视频推荐。推荐人需录制 60-90 秒短视频,口述对申请者的评价。
| 维度 | 传统文本推荐信 | 视频推荐 |
|---|---|---|
| 真实性 | 可代写,难以验证 | 难代写,面部表情和语气真实 |
| 信息量 | 可详细展开 | 受时长限制,需精炼 |
| 推荐人负担 | 撰写耗时 | 录制+表达压力 |
| 适用项目 | 全部 | 部分商学院、MBA、创新硕士项目 |
| 2026 普及度 | 主流 | 试点中,预计扩大 |
视频推荐的应对
如果你的推荐人被要求录制视频推荐:
- 提前告知推荐人项目有此要求,确认其愿意出镜
- 为推荐人准备 3-5 个 talking points,便于其组织语言
- 建议推荐人在安静、光线充足的环境录制
- 提醒推荐人自然表达即可,无需过度修饰——真实比完美更重要
7.2 同行评价(Peer Review)
少数 MBA 项目引入同行评价,要求申请者的同龄人(同事、同学)评价其领导力、协作能力和影响力。这是对传统"上级推荐"的补充,从平级视角提供不同维度的评价。
同行评价的选择策略
选择同行评价人时:
- 选择与你深度合作过的同龄人,而非泛泛之交
- 选择能评价你"软技能"的人(领导力、协作、沟通)
- 避免选择下属——同行评价强调"平级视角"
- 提前与同行评价人沟通你希望突出的具体事例
7.3 第三方平台与流程规范化
Interfolio 等第三方推荐信管理平台使用更普及。推荐信一旦上传,申请者无法查看或修改,流程更规范化,"自己写自己交"的空间被压缩。
| 平台/方式 | 机制 | 2026 趋势 |
|---|---|---|
| 网申系统直发 | 填推荐人邮箱,系统发链接 | 主流,最普遍 |
| Interfolio | 统一管理推荐信,可发多校 | 普及度上升 |
| 视频推荐 | 推荐人录制短视频 | 试点扩大 |
| 纸质邮寄 | 签名密封后寄出 | 越来越少 |
八、提交方式详解
| 提交方式 | 操作流程 | 优点 | 注意事项 |
|---|---|---|---|
| 网申系统推荐链接 | 填写推荐人邮箱,系统自动发链接 | 最主流,流程简单 | 提前确认邮箱正确,提醒查垃圾邮件 |
| Interfolio | 通过 Interfolio 统一管理,可发多校 | 推荐人只需上传一次 | 需提前注册,部分项目不支持 |
| 视频推荐 | 推荐人按链接录制视频上传 | 真实性高 | 提前告知推荐人准备 |
| 纸质邮寄 | 推荐人签名后密封寄出 | 传统项目接受 | 需提前邮寄,确保截止前到达 |
| 邮件发送 | 推荐人通过邮箱发给招生委员会 | 少数项目接受 | 需提前确认是否接受此方式 |
推荐信提交的常见问题处理
- 推荐人没收到链接:检查邮箱拼写,让推荐人查垃圾邮件,必要时联系学校重新发送
- 推荐人忘记提交:截止前 1 周和 3 天分别发提醒,截止前 1 天电话/微信提醒
- 推荐人提交晚了:及时联系学校招生办说明情况,部分学校允许宽限几天;但顶尖项目通常严格截止
- 推荐人临时反悔:这就是为什么要准备 4-5 位推荐人作为备份
九、推荐信撰写中的伦理与合规
9.1 "自己写推荐信"的边界
在中国学术环境中,申请者帮推荐人起草推荐信草稿是普遍做法。但必须遵守以下边界:
推荐信的伦理红线
- 内容必须真实:草稿中的所有事例、数据、评价必须真实发生,不得编造
- 推荐人必须认可并修改:推荐人不能只签字不阅读,必须对内容进行实质性修改和认可
- 不得虚构推荐人评价:不能写推荐人从未说过的话
- 推荐人必须知情并自愿署名:不能在推荐人不知情的情况下使用其名义
9.2 AI 辅助推荐信的合规性
AI 辅助推荐信的风险
2026 各校对 AI 参与文书的政策趋严,推荐信同样适用。AI 可用于:
- 检查语法和拼写(合规)
- 优化句式表达(合规,但最终文字需人工确认)
AI 不可用于:
- 生成整封推荐信(违规)
- 编造推荐人未提及的事例和数据(严重违规)
- 模仿推荐人语气生成虚假评价(严重违规)
最稳妥做法:推荐信内容由推荐人或申请者基于真实经历撰写,AI 仅用于语法检查。
十、常见问题
Q1:推荐人可以用中文写推荐信吗? 大多数英文授课项目要求英文推荐信。如果推荐人无法用英文写作,可以用中文撰写后由翻译公司翻译(需附翻译声明),或你先帮推荐人拟定英文草稿,推荐人修改后签字提交。
Q2:可以自己写推荐信然后让推荐人签字吗? 这是普遍做法(尤其中国学术环境)。你可以为推荐人准备详细草稿,推荐人修改、补充、调整语气后署名提交。但必须确保推荐人认可内容并愿意署名。关键原则:内容必须真实反映推荐人的真实评价。
Q3:推荐信数量多少合适? 本科 2-3 封,硕士 2-3 封,博士 3 封。部分项目最多 4 封。建议准备 3-4 位推荐人,根据不同项目灵活选择。不要提交超过要求数量的推荐信——多余的信不会被阅读。
Q4:推荐人提交晚了怎么办? 提前与推荐人确认截止日期,在截止前 1 周和 3 天分别发提醒。如果推荐人确实延误,及时联系学校招生办说明情况,部分学校允许宽限几天。建议在选推荐人时就考虑可靠性。
Q5:应该选副教授还是正教授? 了解程度比头衔更重要。如果正教授对你了解不多(如只上过大课),推荐信内容会很空洞。一位了解你的副教授的强力推荐远比一位不了解你的正教授的模板化推荐有价值。但如果有两位同等了解你的教授,选头衔更高的。
Q6:推荐信和 PS 内容可以重复吗? 可以,但角度应不同。PS 中你描述自己的经历和感受,推荐信从第三方角度提供佐证和评价。例如 PS 说"我领导了一个研究项目",推荐信应该从导师角度说"San 主动提出了这个研究方向,并在执行中展现了出色的领导力"——同一件事,不同视角。
Q7:26Fall 标化回归对推荐信有什么影响? 标化回归后,硬分数重新成为筛选门槛,但"够线"的申请者数量也增加。这意味着推荐信在"三维达标"的竞争者之间的区分作用进一步放大。一封具体的、有比较性评价的推荐信,可能直接决定你能否从数千名"够线"申请者中脱颖而出。
Q8:视频推荐信怎么准备? 2026 部分项目开始尝试视频推荐。提前告知推荐人项目有此要求,准备 3-5 个 talking points,建议推荐人在安静、光线充足的环境录制,自然表达即可。真实比完美更重要。
十一、相关阅读
- PS/SOP 写作完全指南(2026深度版) — 文书写作全方位指南
- AI 辅助文书写作政策与合规指南 — AI 使用红线
- 简历 CV/Resume 模板与技巧(2026版) — 简历写作全面指南
- 录取案例与数据分析(2026版) — 各项目录取三维参考
- 申请材料清单与时间节点(2026版) — 完整材料清单
- 网申系统操作指南(深度版) — 推荐信提交流程详解
