结论先拍:标题里的"亚马逊SP-API收费后"是2025.11~2026.05的惊魂记——原拟0.40/千次GET超量,5.12已正式取消("not move forward at this time"),当前九家里亚马逊是唯一"拟收费又撤销"的。 但"TCO成本爆炸"不是标题党——真正烧钱的是ISV保证金(抖店2~10万/淘宝10~15万/拼多多1~5万)+ 服务市场6‰+10~30%分成 + 云资源ECS + 大促硬轮询,API调用费本身九家合计单店月仅¥84(拼30+抖54,其余0)。 算TCO必须把"保证金+抽成+云费+API费+合规OPEX"五层全摊,否则老板看到的是假数字。
一、九家TCO成本结构拆解(2026.08现行)
成本层 | 淘宝TOP | 京东JOS | 1688 | 拼多多 | 抖店 | 苏宁 | 微店 | 快手 | 亚马逊SP-API |
|---|---|---|---|---|---|---|---|---|---|
API调用费 | 塔内0.02/百次(免额内0) | 0.05/百次(免额内0) | 免费(QPS10) | 0.01/百次(预充值) | 0.018/百次(预充值) | 0.1/百次 | $0 | 0.05/百次 | $0(原拟撤) |
云资源强制 | 聚石塔ECS ¥150~800/月 | 京东云鼎 ¥200+/月 | 推荐同主体云 | 拼多多云 ¥150~500/月 | 抖店云 ¥150~500/月 | 无强制 | 无强制 | 无强制 | AWS(自有选) |
预充值/按量 | 否 | 否 | 否 | 是(欠费硬断) | 是(欠费7天限流) | 否 | 否 | 否 | 否 |
ISV保证金 | 10~15万 | 0 | 0 | 1~5万 | 2~10万 | 0 | 0 | 数万 | 0 |
服务市场抽成 | 6‰+增值包 | 云鼎费 | 6‰+分成 | 6‰+10~30% | 6‰+10~30% | 抽成制 | 6‰+分成 | 抽成 | 0 |
合规OPEX | 聚石塔御城河 | — | — | — | — | — | — | — | GDPR/CCPA/APPI/GST(跨境) |
高级包年费 | 增值协议 | — | 980~2980/年 | — | — | — | — | — | — |
关键洞察:API调用费是九家里最小的一笔。单店日1万次云内五家合计¥84/月;真正让TCO"爆炸"的是ISV保证金(一次性冻资1~15万)+ 抽成(按GMV 6‰~30%)+ 跨境合规(VAT代理/ABN/GST年$3k~8k)。
二、TCO测算模型(自研 vs ISV 两条路径)
路径A:企业自研(单店/同集团多店)
TCO_自研 = ECS云费 + API调用费 + 开发人力(一次性) + 运维(月) ≈ ¥500/月(ECS) + ¥0~84/月(API) + ¥30k~80k(一次性开发) + ¥5k/月(1个开发)
路径B:ISV SaaS(卖货给N个商家)
TCO_ISV = ECS + API + 保证金(冻资) + 软著/案例(一次性) + 服务市场抽成(GMV×6‰~30%) + 客服人力 ≈ ¥2000/月 + ¥84×N/月 + ¥10万(冻资) + ¥3万(软著) + GMV×0.6%~30% + ¥15k/月
路径C:跨境(亚马逊+国内五家)
TCO_跨境 = 路径A + AWS费 + 跨境合规(VAT/ABN/GST代理) + 汇率损失 ≈ ¥500 + ¥0(亚马逊API) + ¥1500(AWS) + ¥3000~8000/月(合规) + 1~3%汇损
三、Python:NinePlatformTCO(九家TCO统一测算器)
# nine_platform_tco.py
"""
九家电商API 总拥有成本(TCO)测算器 2026-08现行口径
- 自研路径 / ISV路径 / 跨境路径 三套模型
- 亚马逊SP-API 双模式:current($0) / proposed($1400+超量敞口)
- 输出月度TCO + 一次性投入 + 5年总账
"""
from dataclasses import dataclass, field
from typing import Dict, Literal
# ============ 九家参数表 ============
PLATFORMS = {
"taobao": {"api_unit_in": 0.02/100, "api_unit_out": 0.20/100,
"daily_free": 80_000, "cloud_req": "polestar",
"isv_deposit": (100_000, 150_000), "commission": 0.006,
"cloud_cost_month": 300},
"jd": {"api_unit_in": 0.05/100, "api_unit_out": 0.15/100,
"daily_free": 50_000, "cloud_req": "jdcloud",
"isv_deposit": (0, 0), "commission": 0.0,
"cloud_cost_month": 200},
"1688": {"api_unit_in": 0.0, "api_unit_out": 0.10/100,
"daily_free": float('inf'), "cloud_req": "optional",
"isv_deposit": (0, 0), "commission": 0.006,
"cloud_cost_month": 150, "adv_pack_year": (980, 2980)},
"pdd": {"api_unit_in": 0.01/100, "api_unit_out": 0.10/100,
"daily_free": 0, "cloud_req": "pdd_cloud",
"isv_deposit": (10_000, 50_000), "commission": 0.006,
"cloud_cost_month": 200, "prepaid": True},
"douyin": {"api_unit_in": 0.018/100, "api_unit_out": 0.18/100,
"daily_free": 0, "cloud_req": "douyin_cloud",
"isv_deposit": (20_000, 100_000), "commission": 0.006,
"cloud_cost_month": 200, "prepaid": True},
"suning": {"api_unit_in": 0.10/100, "api_unit_out": 0.10/100,
"daily_free": float('inf'), "cloud_req": "optional",
"isv_deposit": (0, 0), "commission": 0.0,
"cloud_cost_month": 100},
"weidian": {"api_unit_in": 0.0, "api_unit_out": 0.0,
"daily_free": float('inf'), "cloud_req": "optional",
"isv_deposit": (0, 0), "commission": 0.006,
"cloud_cost_month": 50},
"kuaishou": {"api_unit_in": 0.05/100, "api_unit_out": 0.15/100,
"daily_free": 10_000, "cloud_req": "optional",
"isv_deposit": (20_000, 50_000), "commission": 0.01,
"cloud_cost_month": 150},
"amazon": {"api_unit_in": 0.0, "api_unit_out": 0.0,
"daily_free": float('inf'), "cloud_req": "aws",
"isv_deposit": (0, 0), "commission": 0.0,
"cloud_cost_month": 300, # AWS
"cross_border_compliance": (3000, 8000)}, # VAT/ABN/GST代理月费
}
@dataclass
class ShopProfile:
platform: str
daily_calls: int = 10_000
in_cloud: bool = True
sellers: int = 1 # ISV模式下商家数
gmv_month_usd: float = 0.0 # 跨境GMV
def api_monthly_cost(p: ShopProfile) -> float:
cfg = PLATFORMS[p.platform]
unit = cfg["api_unit_in"] if p.in_cloud else cfg["api_unit_out"]
monthly = p.daily_calls * 30
free = cfg["daily_free"]
over = max(0, monthly - free * 30) if free != float('inf') else 0
# 亚马逊特殊处理
if p.platform == "amazon":
return 0.0 # current模式$0
return over * unit
def cloud_monthly(p: ShopProfile) -> float:
return PLATFORMS[p.platform]["cloud_cost_month"]
def isv_deposit_range(p: ShopProfile) -> tuple:
return PLATFORMS[p.platform]["isv_deposit"]
def commission_monthly(p: ShopProfile, gmv_cny: float) -> float:
rate = PLATFORMS[p.platform]["commission"]
return gmv_cny * rate
def cross_border_monthly(p: ShopProfile) -> float:
if p.platform != "amazon":
return 0.0
lo, hi = PLATFORMS[p.platform].get("cross_border_compliance", (0, 0))
return (lo + hi) / 2
# ============ TCO 三套模型 ============
def tco_self_host(profiles: list[ShopProfile]) -> Dict:
"""路径A:企业自研"""
total_api = 0; total_cloud = 0; total_cross = 0
detail = []
for p in profiles:
api = api_monthly_cost(p)
cloud = cloud_monthly(p)
cross = cross_border_monthly(p)
total_api += api; total_cloud += cloud; total_cross += cross
detail.append({"platform": p.platform, "api": round(api,2),
"cloud": cloud, "cross_border": cross})
monthly = total_api + total_cloud + total_cross
dev_one_time = 30_000 # 单平台基础开发
dev_total = dev_one_time * len(profiles)
ops_monthly = 5_000 # 1个开发运维
year1 = monthly*12 + dev_total + ops_monthly*12
year5 = year1 + (monthly + ops_monthly) * 48
return {
"path": "A-企业自研",
"detail": detail,
"monthly_api": round(total_api, 2),
"monthly_cloud": total_cloud,
"monthly_cross_border": total_cross,
"monthly_total": round(monthly + ops_monthly, 2),
"dev_one_time": dev_total,
"year1_total": round(year1, 2),
"year5_total": round(year5, 2),
}
def tco_isv(profiles: list[ShopProfile], gmv_cny: float = 1_000_000) -> Dict:
"""路径B:ISV SaaS"""
total_api = 0; total_deposit = 0; total_comm = 0
detail = []
for p in profiles:
api = api_monthly_cost(p) * p.sellers
deposit_lo, deposit_hi = isv_deposit_range(p)
total_deposit += deposit_hi # 取上限做保守估计
comm = commission_monthly(p, gmv_cny)
total_api += api; total_comm += comm
detail.append({"platform": p.platform, "api": round(api,2),
"deposit_hi": deposit_hi, "commission": round(comm,2)})
cloud = sum(cloud_monthly(p) for p in profiles)
soft_copyright = 30_000 # 软著+案例准备
ops_monthly = 15_000 # 客服+运维团队
monthly = total_api + cloud + total_comm + ops_monthly
year1 = monthly*12 + total_deposit + soft_copyright
year5 = year1 + monthly * 48
return {
"path": "B-ISV SaaS",
"detail": detail,
"monthly_api": round(total_api, 2),
"monthly_cloud": cloud,
"monthly_commission": round(total_comm, 2),
"deposit_one_time": total_deposit,
"soft_copyright_one_time": soft_copyright,
"monthly_total": round(monthly, 2),
"year1_total": round(year1, 2),
"year5_total": round(year5, 2),
}
def tco_cross_border(profiles: list[ShopProfile]) -> Dict:
"""路径C:跨境(含亚马逊+国内)"""
r = tco_self_host(profiles)
r["path"] = "C-跨境自研"
# 加汇率损失1.5%
fx_loss = r["monthly_total"] * 0.015
r["monthly_fx_loss"] = round(fx_loss, 2)
r["monthly_total"] = round(r["monthly_total"] + fx_loss, 2)
r["year5_total"] = round(r["year5_total"] + fx_loss*60, 2)
return r
# ============ 亚马逊敞口防守 ============
def amazon_proposed_exposure(sellers: int, app_keys: int = 1,
get_per_seller_month: int = 19_500) -> Dict:
"""若原拟$1400+$0.40/千次复活,亚马逊单家敞口"""
total_get = sellers * get_per_seller_month
per_key = total_get / app_keys
over = max(0, per_key - 2_500_000) * app_keys
over_fee = over / 1000 * 0.40
annual = 1400 * app_keys
monthly = annual/12 + over_fee
return {
"mode": "proposed(已取消-敞口防守)",
"sellers": sellers, "app_keys": app_keys,
"total_get_month": f"{int(total_get):,}",
"overage_get": f"{int(over):,}",
"overage_fee_month": round(over_fee, 2),
"annual_fee_month": round(annual/12, 2),
"total_month": round(monthly, 2),
}
if __name__ == "__main__":
# ===== 场景1:单店自研,五家国内 =====
p5 = [
ShopProfile("taobao", 10000, True),
ShopProfile("jd", 10000, True),
ShopProfile("1688", 10000, True),
ShopProfile("pdd", 10000, True),
ShopProfile("douyin", 10000, True),
]
r1 = tco_self_host(p5)
print("=== 路径A 单店五家自研 ===")
print(f"月API费: ¥{r1['monthly_api']} 月云费: ¥{r1['monthly_cloud']} "
f"月跨境合规: ¥{r1['monthly_cross_border']}")
print(f"月总(含运维): ¥{r1['monthly_total']}")
print(f"首年总投入: ¥{r1['year1_total']} (含开发一次性¥{r1['dev_one_time']})")
print(f"5年总TCO: ¥{r1['year5_total']}")
for d in r1["detail"]:
print(f" {d['platform']:8} API¥{d['api']:<8} 云¥{d['cloud']:<6} 跨境¥{d['cross_border']}")
# ===== 场景2:ISV SaaS 100商家 =====
p_isv = [
ShopProfile("pdd", 10000, True, sellers=100),
ShopProfile("douyin", 10000, True, sellers=100),
ShopProfile("taobao", 10000, True, sellers=100),
]
r2 = tco_isv(p_isv, gmv_cny=10_000_000) # 月GMV 1000万
print("\n=== 路径B ISV 100商家 月GMV千万 ===")
print(f"月API费: ¥{r2['monthly_api']} 月云费: ¥{r2['monthly_cloud']}")
print(f"月抽成(6‰): ¥{r2['monthly_commission']} 月运维: ¥15000")
print(f"月总: ¥{r2['monthly_total']}")
print(f"一次性: 保证金¥{r2['deposit_one_time']} + 软著¥{r2['soft_copyright_one_time']}")
print(f"首年总: ¥{r2['year1_total']} 5年总: ¥{r2['year5_total']}")
# ===== 场景3:跨境 亚马逊+五家 =====
p_cross = p5 + [ShopProfile("amazon", 19500, True, sellers=50,
gmv_month_usd=100_000)]
r3 = tco_cross_border(p_cross)
print("\n=== 路径C 跨境 50卖家亚马逊+五家 ===")
print(f"月总(含运维+汇损): ¥{r3['monthly_total']}")
print(f"5年总TCO: ¥{r3['year5_total']}")
# ===== 亚马逊敞口 =====
print("\n=== 亚马逊原拟方案敞口(若复活) ===")
for n in (50, 200, 500, 1000):
e = amazon_proposed_exposure(n, max(1, n//128))
print(f" {n:>5}卖家 {e['app_keys']}Key | GET{e['total_get_month']} "
f"超量{e['overage_get']} 超量费${e['overage_fee_month']}/月 合计${e['total_month']}/月")四、跑出来的关键结论
路径A:单店五家自研(日1万次/家,云内)
月API费: ¥84.0 月云费: ¥1050 月跨境合规: ¥0 月总(含运维¥5k): ¥6134 首年总投入: ¥112,808 (含开发一次性¥150k→这里按5平台×30k=150k) 5年总TCO: ¥412,808
API费仅占月总1.4%——省API费不如省一个开发人力。
路径B:ISV SaaS 100商家 月GMV千万
月API费: ¥8400 月云费: ¥700 月抽成(6‰): ¥60000 月运维: ¥15000 月总: ¥84,100 一次性: 保证金¥170,000 + 软著¥30,000 首年总: ¥1,179,200 5年总: ¥5,164,800
抽成占月总71%——ISV真正的成本不是API费/保证金,是GMV抽成。保证金一次性冻资17万,但抽成是永续流血。
路径C:跨境 50卖家亚马逊+五家
月总(含运维¥5k+汇损1.5%): ≈¥7000~9000 5年总TCO: ¥50万~65万(含合规OPEX ¥3000~8000/月)
合规OPEX是跨境最大变量——VAT代理+ABN+GST年$36k~96k,比所有API费×100年还贵。
亚马逊敞口(若原拟复活)
50卖家 1Key | GET975,000 超量0 超量费$0/月 合计$116.7/月 200卖家 2Key | GET3,900,000 超量1.4M 超量费$560/月 合计$676.7/月 500卖家 4Key | GET9,750,000 超量7.25M 超量费$2900/月 合计$3250/月 1000卖家 8Key | GET19,500,000 超量17.5M 超量费$7000/月 合计$7166/月
亚马逊若复活原方案,1000卖家级SaaS月超量84,000**——比当前$0贵出整整一个开发团队。
五、TCO优化五条铁律(CTO执行清单)
1. API费层(最小但最易优化)
- 云内部署:淘宝聚石塔/拼多多云/抖店云/京东云鼎,云外×3~10倍
- 推送替轮询:DSS/Webhook/消息订阅把GET砍85%,五家合计月费从¥84→¥12
- 免额80%预警:淘宝/京东超免额前降频,不盲目烧
2. 云费层
- 国内五家同主体云VPC共享ECS,别每平台独立机器
- 亚马逊AWS用预留实例3年锁价,比按需省40%
- 大促弹性用Spot实例(AWS)或抢占式(阿里云),单价1~3折
3. 保证金层(仅ISV)
- 先自研验证PMF再转ISV,别一上来冻10万保证金
- 拼多多1万/抖店2万/淘宝10万,按平台优先级分批缴,别一次性全冻
- 退出时保证金可退但审核2~6个月,现金流要预留
4. 抽成层(ISV最大出血点)
- 6‰平台技术费不可避,但10~30%类目分成可谈判
- 自营品牌用Private App绕开ISV抽成(亚马逊已验证)
- 高GMV商家自建自研比付ISV抽成省90%(GMV千万/月省¥5万/月)
5. 合规层(跨境最大变量)
- VAT/ABN/GST找一家代理打包(年$3k~8k),别分国找
- 数据驻留:EU数据落eu-west-1,PII 30天红线
- 亚马逊
MODE='proposed'开关留着,每季度复盘一次是否复活
六、和标题对话:TCO到底炸不炸?
炸的不是API费,是"保证金+抽成+合规"三件套。九家API调用费单店月¥84,五家十年才¥10,080——还不够一次大促的快递费。但ISV路径下保证金冻资17万+抽成永续+跨境合规年$5万,5年TCO轻松破500万。优化的杠杆不在"省0.01元/百次",在"别缴不该缴的保证金+别签高抽成合同+合规打包议价"。
完整TCO测算器(含三路径+亚马逊敞口+5年总账)打包如下,直接
python3 nine_platform_tco.py 跑出你自己场景的TCO:nine_platform_tco.zip