What Is Sales Analysis?什么是销售分析？
Sales analysis is the structured examination of revenue, transactions, pipeline, customers, products, channels, and sales activity to explain performance and support a specific decision. It goes beyond reporting a number: it compares that number with a target or prior period, finds where the change occurred, tests plausible drivers, and assigns a next action.
销售分析是对收入、交易、管道、客户、产品、渠道和销售活动进行结构化检查，以解释业绩并支持一项明确决策。它不只报告数字，还会把数字与目标或历史期间比较，定位变化发生在哪里，检验可能的驱动因素，并确定下一步行动。
A useful analysis begins with a decision question such as “Why did enterprise revenue miss plan?” or “Which channel created the healthiest pipeline?” It does not begin by adding every available KPI to a dashboard. The question determines the population, time window, metrics, comparison, and segments.
有效分析从决策问题开始，例如“企业客户收入为何未达计划？”或“哪个渠道带来了最健康的销售管道？”，而不是先把所有 KPI 塞进仪表板。问题决定分析对象、时间窗口、指标、比较基准和分群维度。
States what happened: revenue, closed deals, pipeline value, and performance against target.
说明发生了什么：收入、成交数、管道金额以及相对目标的表现。
Explains where and why results changed, what evidence supports the explanation, and what should happen next.
解释结果在哪里、为何变化，哪些证据支持解释，以及下一步应采取什么行动。
A Reusable Sales Analysis Template可复用的销售分析模板
Use this Sales Analysis Brief before building charts. One completed brief should answer one decision question. If two audiences need different decisions—for example, a weekly manager review and a quarterly board review—create separate briefs rather than forcing both into one report.
在制作图表前先填写这份“销售分析简报”。一份简报只回答一个决策问题。如果两个受众需要不同决策，例如周度销售经理复盘和季度董事会复盘，应分别建立简报，而不是强行合并成一份报告。
| Template field模板字段 | What to enter填写内容 | Quality check质量检查 |
|---|---|---|
| Decision question决策问题 | One question with an owner and deadline一个具有负责人和截止时间的问题 | Can the answer change a decision?答案能否改变一项决策？ |
| Population and window对象与时间窗 | Market, team, opportunity cohort, currency, and dates市场、团队、商机 Cohort、币种和日期 | Are periods and cohorts comparable?期间和 Cohort 是否可比？ |
| Metric contract指标契约 | Formula, grain, filters, source, and owner公式、粒度、筛选条件、来源和负责人 | Can another analyst reproduce it?其他分析师能否复算？ |
| Comparison and segments比较与分群 | Target, prior period, product, region, channel, customer, or rep目标、上期、产品、区域、渠道、客户或销售人员 | Are small samples labeled?小样本是否已标注？ |
| Finding and hypothesis发现与假设 | Observed fact, possible explanation, and counterevidence观察事实、可能解释和反证 | Is inference separated from fact?推断是否与事实分开？ |
| Action and validation行动与验证 | Owner, action, due date, expected signal, and review date负责人、行动、截止时间、预期信号和复核日期 | Will the result be measured again?结果是否会再次衡量？ |
Template rule: keep observation, interpretation, and recommendation in separate fields. “Win rate fell from A to B” is an observation. “Lead quality declined” is a hypothesis until evidence connects source mix or qualification changes to the outcome.
模板规则：把观察、解释和建议分开填写。“赢率从 A 降至 B”是观察；“线索质量下降”仍是假设，直到渠道结构或资格判定变化与结果之间存在证据连接。
Prepare the Right Sales Data准备正确的销售数据
For B2B analysis, the minimum useful grain is often one row per opportunity plus a stage-history table. For transaction-led businesses, it may be one order line, invoice line, or daily product-store record. Do not mix these grains in one table without defining how they join.
对于 B2B 分析，最低可用粒度通常是每个商机一行，并配有阶段历史表；对于交易驱动业务，则可能是每个订单行、发票行或每日产品—门店记录。不要在未定义连接方式的情况下混用这些粒度。
Opportunity ID, owner, account, stage history, created date, expected close, close date, amount, source, and outcome.
商机 ID、负责人、客户、阶段历史、创建日期、预计成交日、实际成交日、金额、来源和结果。
Actual booked or recognized revenue, quantity, discount, return, currency, product, invoice, and customer identifiers.
实际签约或确认收入、数量、折扣、退货、币种、产品、发票和客户标识。
Quota by rep or territory, headcount dates, working days, territory assignment, and plan version.
按销售人员或区域划分的配额、人员日期、工作日、区域分配和计划版本。
Marketing source, product usage, customer segment, pricing changes, support events, and calendar effects when relevant.
在相关情况下加入营销来源、产品使用、客户分群、价格变化、支持事件和日历效应。
Before calculation, deduplicate business keys, normalize currency, decide whether revenue is booked or recognized, freeze historical stage snapshots, and document how deleted or reopened opportunities are treated. A polished chart cannot repair an unstable metric definition.
计算前应去重业务主键、统一币种、明确使用签约收入还是确认收入、冻结历史阶段快照，并记录已删除或重新打开的商机如何处理。精美图表无法修复不稳定的指标定义。
Core Metrics for Sales Analysis销售分析的核心指标
Choose a compact metric set that covers outcomes, pipeline mechanics, efficiency, and planning. Each metric needs an owner and a documented contract; labels alone are not definitions.
选择一组精简指标，覆盖结果、管道机制、效率和计划。每个指标都需要负责人和已记录的指标契约；单有名称不等于定义。
| Metric指标 | Basic calculation基础计算 | Question answered回答的问题 | Common trap常见陷阱 |
|---|---|---|---|
| Revenue收入 | Sum of eligible revenue符合口径的收入之和 | What value was produced?产生了多少价值？ | Mixing bookings and recognition混用签约与确认收入 |
| Average deal value平均交易金额 | Closed-won value ÷ won deals成交金额 ÷ 成交单数 | Is value per win changing?单笔成交价值是否变化？ | Outliers and currency mix异常值与币种混合 |
| Win rate赢率 | Won ÷ closed eligible opportunities成交 ÷ 已关闭合格商机 | How often do qualified deals win?合格商机多常成交？ | Open deals in the denominator分母误含未关闭商机 |
| Stage conversion阶段转化率 | Cohort reaching next stage ÷ entering stage进入下一阶段数 ÷ 进入本阶段数 | Where does pipeline leak?管道在哪个阶段流失？ | Using current-state counts误用当前状态数量 |
| Sales cycle length销售周期 | Close timestamp − defined start成交时间 − 规定起点 | How long does conversion take?转化需要多久？ | Inconsistent start events起始事件不一致 |
| Pipeline coverage管道覆盖率 | Eligible pipeline ÷ remaining target合格管道金额 ÷ 剩余目标 | Is there enough pipeline?管道是否足够？ | Ignoring probability and timing忽略概率和成交时间 |
| Quota attainment配额达成率 | Credited actual ÷ assigned quota计入业绩 ÷ 分配配额 | Who or what is on plan?谁或哪些区域达到计划？ | Mid-period territory changes期间内区域调整 |
| Forecast variance预测偏差 | Actual − forecast, with signed percentage实际值 − 预测值，并保留方向 | How biased or inaccurate was the forecast?预测偏差或误差多大？ | Overwriting forecast history覆盖历史预测快照 |
Keep acquisition cost, retention, gross margin, and lifetime value when they are needed for the decision, but do not call a revenue-only view “sales health.” Revenue can rise while margin, win rate, or future pipeline deteriorates.
当决策需要时，可加入获客成本、留存、毛利和客户生命周期价值，但不要把只看收入的视图称为“销售健康度”。收入可能增长，同时毛利、赢率或未来管道正在恶化。
How to Perform Sales Analysis Step by Step如何逐步完成销售分析
- Write the decision question.写出决策问题。
Name the audience, decision, deadline, business population, and desired level of confidence.
明确受众、决策、截止时间、业务对象和所需置信程度。
- Freeze definitions and comparison.冻结定义和比较基准。
Record dates, time zone, currency, cohort, revenue treatment, opportunity eligibility, and target version.
记录日期、时区、币种、Cohort、收入处理、商机资格和目标版本。
- Join sources at the right grain.按正确粒度连接数据源。
Create explicit keys between opportunity, account, order, invoice, product, rep, and calendar tables. Test one-to-many joins before aggregation.
在商机、客户、订单、发票、产品、销售人员和日历表之间建立明确键，并在聚合前检验一对多连接。
- Establish the baseline.建立基线。
Calculate outcomes and a small driver set for the current period, target, and a comparable prior period.
计算当前期间、目标和可比历史期间的结果指标与少量驱动指标。
- Segment the change.对变化进行分群。
Break the delta down by product, market, source, customer type, rep, or cohort. Rank contributions rather than displaying every slice.
按产品、市场、来源、客户类型、销售人员或 Cohort 拆分变化，并按贡献排序，而不是展示所有切片。
- Test explanations.检验解释。
Compare timing, mix, conversion, price, volume, and data-quality alternatives. Treat correlation as a lead for investigation, not proof of cause.
比较时间、结构、转化、价格、数量和数据质量等替代解释。把相关性视为调查线索，而不是因果证明。
- Assign action and measurement.分配行动并定义衡量方式。
State the owner, change, due date, expected signal, guardrail metric, and next review date. Archive the query or calculation so the result can be rerun.
写明负责人、变更、截止时间、预期信号、护栏指标和下次复核日期，并保存查询或计算，使结果可重跑。
Sales Analysis Example: Growth With Hidden Risk销售分析示例：增长背后的隐藏风险
Illustrative example only: the figures below are hypothetical and are not an InfiniSynapse customer result or benchmark.
仅为说明示例：以下数字为假设数据，并非 InfiniSynapse 客户结果或行业基准。
Suppose quarterly revenue rose from an illustrative $1.20 million to $1.32 million. Reporting alone would call this growth. Analysis shows that average deal value increased while win rate fell and the sales cycle lengthened. A product split then shows most growth came from one enterprise package, while the mid-market segment lost conversion.
假设季度收入从 120 万美元增长至 132 万美元。单纯报告会把它称为增长；进一步分析发现，平均交易金额提高，但赢率下降、销售周期变长。按产品拆分后，大部分增长来自一个企业套餐，而中端市场的转化率下降。
Defensible conclusion: revenue improved, but the growth was concentrated and the broader pipeline became less efficient. The data supports reviewing enterprise concentration, mid-market qualification, and stage delay. It does not yet prove that a particular campaign or rep behavior caused the decline.
可验证结论：收入有所提高，但增长较集中，整体管道效率下降。现有数据支持检查企业客户集中度、中端市场资格判定和阶段延迟，但尚不能证明某个营销活动或销售行为造成了下降。
Segment Sales Performance Without Creating Noise在不制造噪声的前提下分群销售表现
Segment only when the dimension could lead to a different decision. Product and region may change resource allocation; source may change qualification or spend; customer segment may change packaging; rep may change coaching. A decorative breakdown that cannot alter action is not analysis.
只有当某个维度可能改变决策时才进行分群。产品和区域可能影响资源分配，来源可能影响资格判定或预算，客户分群可能影响套餐设计，销售人员维度可能影响辅导。无法改变行动的装饰性拆分不属于有效分析。
Which segments explain the largest share of the total increase or decline?
哪些分群解释了总体增长或下降的最大部分？
Did volume change, or did the probability of moving between stages change?
是数量发生变化，还是阶段间转移概率发生变化？
Did the share of high-value products, regions, sources, or customer types shift?
高价值产品、区域、来源或客户类型的占比是否变化？
Did close timing, seasonality, working days, or deal slippage move value between periods?
成交时间、季节性、工作日或延期是否把价值移动到其他期间？
Always show segment sample size and compare like with like. A 50% win rate from two closed opportunities is not more reliable than a 28% rate from hundreds. Use minimum-volume rules or confidence intervals when decisions carry material risk.
始终展示分群样本量，并进行同类比较。两个已关闭商机产生的 50% 赢率，并不比数百个样本产生的 28% 更可靠。当决策风险较大时，应设置最低样本规则或使用置信区间。
Choose a Sales Analysis Tool by Data Complexity根据数据复杂度选择销售分析工具
| Approach方式 | Best fit适合场景 | Strength优势 | Watch for注意事项 |
|---|---|---|---|
| Spreadsheet电子表格 | One-off analysis and small exports一次性分析和小型导出 | Flexible and transparent灵活且易检查 | Manual refresh and version drift手动刷新与版本漂移 |
| CRM reportsCRM 报告 | Operational pipeline review运营型管道复盘 | Close to sales workflow贴近销售执行流程 | Limited cross-source context跨源背景有限 |
| Warehouse and BI数据仓库与 BI | Governed recurring reporting受治理的周期性报告 | Reusable models and dashboards可复用模型与仪表板 | Modeling and maintenance effort建模与维护成本 |
| AI analysis layerAI 分析层 | Cross-source, iterative questions跨源与迭代式提问 | Fast exploration with natural language用自然语言快速探索 | Definitions and outputs still require validation定义与输出仍需验证 |
Use a weekly cadence for pipeline exceptions, a monthly cadence for performance and conversion, and a quarterly cadence for territory, pricing, capacity, or strategy. For dashboard design principles and limitations, see the InfiniSynapse guide to a decision-focused data dashboard.
可按周检查管道异常，按月复盘业绩和转化，按季度复盘区域、定价、产能或策略。关于仪表板的设计原则与局限，可参阅 InfiniSynapse 的面向决策的数据仪表板指南。
Common Sales Analysis Mistakes常见销售分析错误
A stage, eligible opportunity, or revenue definition changes between periods, making the trend artificial.
阶段、合格商机或收入定义在期间之间改变，导致趋势失真。
Only won deals are analyzed, so losses, deletions, and stalled opportunities disappear from the explanation.
只分析成交商机，使失败、删除和停滞商机从解释中消失。
A current CRM snapshot is used to reconstruct history even though stages and amounts were overwritten.
使用当前 CRM 快照重建历史，但阶段和金额已经被覆盖。
A segment moves with revenue, and the analysis claims it caused the change without a design that supports causality.
某分群与收入同步变化，分析却在没有因果设计的情况下宣称它造成了变化。
Large deals or a few reps dominate the mean. Add distributions, medians, percentiles, and segment sizes.
大额交易或少数销售人员主导均值，应补充分布、中位数、分位数和样本量。
The team optimizes conversion but damages price, margin, retention, or pipeline quality.
团队优化转化率，却损害价格、毛利、留存或管道质量。
Run Cross-Source Sales Analysis With InfiniSynapse使用 InfiniSynapse 进行跨源销售分析
When sales evidence is split across CRM tables, billing databases, warehouse models, and files, InfiniSynapse can act as an analysis layer over connected sources. Prepare read access, table owners, business keys, metric definitions, time zone, currency rules, and the decision question before beginning.
当销售证据分散在 CRM 表、账单数据库、数据仓库模型和文件中时，InfiniSynapse 可以作为连接数据源之上的分析层。开始前应准备只读访问、表负责人、业务主键、指标定义、时区、币种规则和决策问题。
- Connect the governed sources.连接受治理的数据源。
Use the minimum permissions needed and identify authoritative sources for opportunity state, actual revenue, and targets.
使用最低必要权限，并明确商机状态、实际收入和目标的权威来源。
- Ask a bounded question.提出有边界的问题。
Include dates, population, comparison, and desired segments rather than asking for a generic summary.
在问题中包含日期、对象、比较基准和所需分群，而不是要求泛化总结。
- Review the evidence.复核证据。
Check selected fields, joins, filters, formulas, excluded records, and outputs against known totals before acting.
采取行动前，检查所选字段、连接、筛选、公式、排除记录和输出，并与已知总数核对。
Use InfiniSynapse to explore connected databases, warehouses, and files in natural language, then validate the generated analysis against your metric contract. InfiniSynapse is not a CRM and does not execute lead assignment, outreach, or opportunity stages.
使用 InfiniSynapse 通过自然语言探索已连接的数据库、数据仓库和文件，再依据指标契约验证分析结果。InfiniSynapse 不是 CRM，也不执行线索分配、销售触达或商机阶段管理。
Try cross-source data analysis体验跨源数据分析Validate the Analysis Before Sharing It分享前验证分析结果
- Reconcile totals: compare revenue, deal counts, and closed dates with the authoritative financial or CRM report.核对总数：将收入、交易数和成交日期与权威财务或 CRM 报告对比。
- Test joins: check whether row counts or amounts multiply after account, order, invoice, or activity tables are joined.检验连接：检查连接客户、订单、发票或活动表后，行数或金额是否被放大。
- Inspect exclusions: quantify deleted, reopened, zero-value, duplicate, missing-owner, and missing-stage records.检查排除项：量化删除、重开、零金额、重复、缺负责人或缺阶段记录。
- Stress-test definitions: rerun with reasonable alternative start dates, stage rules, or outlier treatment and note material changes.压力测试定义：使用合理的替代起始日期、阶段规则或异常值处理重跑，并记录重大变化。
- Separate certainty: label facts, estimates, hypotheses, and recommendations so readers know what the data proves.区分确定性：标记事实、估计、假设和建议，让读者知道数据实际证明了什么。
- Make it rerunnable: retain the query, model version, parameters, source timestamp, and next review date.确保可重跑：保留查询、模型版本、参数、数据源时间戳和下次复核日期。
Sales Analysis FAQ销售分析常见问题
Sales analysis is the structured examination of revenue, transactions, pipeline, customers, products, channels, and sales activity to explain performance and support a specific decision.
销售分析是对收入、交易、管道、客户、产品、渠道和销售活动进行结构化检查，以解释业绩并支持一项明确决策。
A useful sales analysis includes a decision question, metric definitions, a comparison period, relevant segments, data quality checks, findings, limitations, and named follow-up actions.
有效的销售分析应包括决策问题、指标定义、比较期间、相关分群、数据质量检查、发现、局限和明确的后续行动。
Start with revenue, units or orders, average deal value, win rate, stage conversion, sales cycle length, pipeline value, quota attainment, and forecast variance, then keep only metrics that answer the chosen decision question.
可从收入、数量或订单数、平均交易金额、赢率、阶段转化、销售周期、管道金额、配额达成率和预测偏差开始，再只保留能回答目标决策问题的指标。
Operational pipeline checks can run weekly, performance reviews monthly, and strategy or territory reviews quarterly. Use one stable definition set so results remain comparable across cycles.
运营型管道检查可按周进行，业绩复盘可按月进行，策略或区域复盘可按季度进行。应使用稳定的定义集，使各周期结果保持可比。
Sales reporting summarizes what happened. Sales analysis compares, segments, investigates drivers, tests alternative explanations, and connects evidence to a decision or action.
销售报告总结发生了什么；销售分析则进行比较和分群、调查驱动因素、检验替代解释，并把证据连接到决策或行动。
No. InfiniSynapse is an analysis layer that can work with connected databases, warehouses, and files. It does not replace CRM workflow functions such as lead assignment, outreach, or opportunity-stage execution.
不能。InfiniSynapse 是可连接数据库、数据仓库和文件的分析层，不替代 CRM 中的线索分配、销售触达或商机阶段执行等工作流功能。