How do you conduct a sales audit?

Scoping determines the quality of everything that follows. It fits on one page and sets four things:

• The main objective: confirming a hunch, deciding on an investment, preparing for growth or securing a takeover. These aims do not call for the same depth or the same priorities, and an audit that pursues everything at once produces an encyclopaedic report that leads nowhere
• The scope, and above all what it excludes: auditing the targeting strategy without touching variable pay is a legitimate choice. Writing it down prevents the audit from later being blamed for missing what it was never asked to examine
• The sponsor and the mandate: who commissions the audit, who will decide, and to whom the findings will be presented. An audit commissioned by the sales director but presented to the executive committee is not run like an audit commissioned by the CEO on the sales function
• The framework given to the teams: reps need to know that the audit covers the sales set-up, who will have access to what, and what will be anonymised. If this framework is poorly set, interviews turn defensive and the collection loses its substance

This step starts before the interviews. The figures help prepare the right questions, and interviews held without them waste time.

What to extract, in order of priority:

• 24 months of deal history: won, lost, loss reason, amount, time between creation and close, assigned rep
• the current pipeline, with the age of each opportunity at its current stage. Deals that have not moved for more than two average cycles are the first sign of an inflated pipeline
• invoicing data, to rebuild average deal size, recurrence and portfolio concentration
• upstream activity: number of calls, meetings and proposals sent, if the data exists

Two difficulties almost always appear. The CRM is incomplete, which is itself an audit finding to document. Loss reasons are either empty or set to "price" by default, a reason that listening to the recordings often contradicts.

If the database is too degraded to use as it stands, rebuild a sample of thirty to fifty recent deals, drawn at random. The sample gives a trend, with no statistical value, that is enough to guide the next steps.

This step stays separate from the previous one, because the two kinds of material are not collected the same way and do not say the same thing. The figures locate the blockers and the field explains their causes.

• Individual interviews: management, managers, every rep, and sales administration, often forgotten although it sees every process malfunction. Each interview follows the same outline, one to one, for forty-five minutes to an hour. Group interviews produce surface consensus and are only useful when presenting findings
• Call listening and proposal review: around ten recordings and ten recent proposals, taken from several reps, show the gap between the pitch described in interviews and the pitch actually delivered
• Field shadowing: two or three meetings observed, time-consuming, which make it possible to describe the team's actual practices
• Customer interviews, when the scope allows: three to five recently won customers and, if possible, two lost prospects. Lost prospects are the most instructive and the least often contacted

The analysis brings the two sets of material together, then benchmarks the results.

Cross-checking produces the most solid findings, because it rests on two independent sources. If the conversion rate drops between proposal and signature and the interviews show there is no structured follow-up after sending, the finding is established and quantified. When the two sources contradict each other, for example when reps describe a process that the data does not show, that contradiction becomes a finding in itself.

Comparison puts performance in context. There are three possible references, in decreasing order of reliability: the company against itself over time, reps against each other at comparable profiles, and finally sector benchmarks. Handle benchmarks with care, as published conversion rates rest on stage definitions too varied to compare seriously.

Finally, rebuild the loss rate stage by stage. This rate is the central analytical deliverable. It replaces a vague finding such as "we are selling less" with a precise one, for example "we lose 60% of deals between the first meeting and the proposal".

An audit easily produces several dozen observations, which stall decisions if they are delivered in bulk. Ranking makes the report usable.

Classify each finding on two axes, expected impact on revenue and implementation effort, then sort them into four groups:

• High impact, low effort: launch immediately (follow-up template, CRM discipline, pitch correction)
• High impact, high effort: plan with dedicated resources (process redesign, variable pay review, tool change)
• Low impact, low effort: handle as you go
• Low impact, high effort: set aside explicitly, and say so

The last group matters most, because an audit that gives up nothing does not help anyone decide. It is better to set a finding aside by naming it than to bury it in a long list of recommendations.

Then limit yourself to five to seven priorities. Beyond that, priorities get diluted and few are completed.

The presentation serves to make decisions. It rests on three principles:

• Separate the finding from the recommendation: the finding rests on data and is not open to debate. The recommendation is a choice, and it belongs to the sponsor. Mixing the two invites people to dispute the facts in order to reject the solution
• Present to those who decide: an audit that touches the organisation, presented only to the sales director, will only lead to what they can decide alone
• Plan feedback to the teams: reps gave two hours of their time. Without feedback, they will be less involved in the next audit, and support for the action plan weakens

The written report accompanies the presentation. It includes an executive summary, the methodology and the scope actually covered, findings backed by evidence for each area, the ranking matrix and costed recommendations.

This step separates an audit from a paper exercise.

Before closing the assignment, freeze the baseline metrics: conversion rate by stage, cycle length, average deal size, portfolio coverage rate. They are what will show, six months later, whether the plan has produced anything. Without this baseline, the question stays open indefinitely.

Then set three measurement checkpoints:

• 1 month: are the low-effort actions under way? Tracking covers implementation, results come later
• 3 months: first effects on activity metrics and on the upper stages of the pipeline
• 6 to 12 months: effects on signed revenue, as the sales cycle requires this delay before any conclusion

Tracking takes a few hours per quarter, and it is what allows the audit to produce a return.