Tom is made up, though plenty of founders will recognise his spreadsheet. He started an IT support company in Charlotte nine years ago and grew it to about $4 million in revenue, mostly by closing deals himself. Now he has four sales reps. In March they closed eleven new contracts. In April they closed two, and nobody could explain why.
Each rep sold their own way. One spent weeks courting dental practices that never signed. Another sent proposals after a single phone call. Tom himself still jumped into deals at the last minute and rescued some of them, which made everyone else’s numbers impossible to read.
Keeping existing clients mattered just as much, and that was slipping too. Research cited in Harvard Business Review found that increasing customer retention by 5% can raise profits by 25% to 95%. A sales process that only chases new logos misses half the point. It has to set honest expectations, so the people who sign actually stay.
Here’s roughly how Tom rebuilt his, and where AI tools helped along the way.
Start with how your customers actually buy
Tom’s first instinct was to download a template with seven stages. That didn’t fit. His small-office clients usually decided after one site visit, while his larger healthcare clients needed a security review, three meetings and a sign-off from somebody’s board.
So he mapped what really happened. For most deals, the path ran from a referral or inbound inquiry to a quick qualifying call, then a discovery visit, a written proposal, a bit of back-and-forth on terms, and a handover to the onboarding team. Healthcare deals added a compliance step in the middle. Small deals skipped the formal discovery visit.
It also helped to find out what his reps were actually doing day to day, rather than what he assumed they did. Sales consultancies like Mettle and Method focus specifically on measuring that gap between real rep activity and what leadership believes is happening. In Tom’s case, the gap was big. One rep was spending nearly half his week on prospects who’d never had the budget.
Decide who’s worth calling before you call
That rep wasn’t lazy. He’d just never been told who a good customer looked like.
Tom went through his best twenty clients and looked for what they had in common. Most had between fifteen and eighty staff. They were usually in professional services or healthcare and already paying someone for IT that kept letting them down. They also tended to call after a specific trigger, often a ransomware scare, an office move or a failed audit. That became his ideal customer profile.
He then wrote four qualifying questions that every rep asks on the first call. What are you using now? What’s prompted you to look? When would you want a change in place? Who signs off on something like this? None of it is meant as an interrogation. It’s a quick check that the deal is real before anyone spends a week on it.
Some CRMs now use AI lead scoring to rank inquiries by how closely they match past winners. Tom tried it. It was useful as a rough sort, and it was wrong often enough that reps still had to make the call themselves.
Discovery is where deals are actually won
The rep who sent proposals after one call kept losing to competitors who asked more questions. Tom made discovery a proper conversation with a handful of standard prompts:
- What problem are you trying to fix?
- How is it affecting the business day-to-day?
- What have you already tried?
- What would “this worked” look like in six months?
- Who else needs to be in the room?
One law firm asked for “faster laptops.” Twenty minutes of discovery revealed the real issue. Their document system crashed every afternoon when everyone saved files at once. A laptop quote would have missed that completely.
Reps now record discovery calls, with permission, using an AI meeting assistant like Zoom’s AI Companion. It produces a summary and a list of action items. The rep checks it, corrects anything it got wrong, and saves the notes, budget hints, and next steps to the customer relationship management system. If someone else picks up the account later, they don’t have to start from scratch.
Follow up like it matters, because it does
Most of Tom’s lost deals hadn’t been lost to competitors. They’d simply gone quiet. A prospect would ask for a proposal on Monday, the proposal arrived nine days later, and by then the prospect had stopped thinking about it.
The team now follows a few simple rules. A short recap goes out within a day of any meeting. The prospect is told exactly when the proposal will arrive. Every conversation ends with a date for the next one. If a prospect asks for more time, that gets logged in the CRM with a real follow-up date, not “check in sometime.”
AI drafting tools speed this up. A rep can turn meeting notes into a recap email in under a minute. Tom’s rule is that nothing goes out unedited, though. One AI-drafted recap once promised a client “24/7 onsite support,” which Tom’s company doesn’t offer.
Each follow-up should earn its place. It might answer a question left over from the meeting, clarify a line in the proposal, or share a case study from the same industry. Sending a third “just checking in” email teaches prospects to ignore you.
Find out where deals actually stall
Revenue alone told Tom nothing about why April was so bad. So he started tracking a handful of numbers each month:
- How many leads became qualified opportunities.
- How many opportunities turned into signed clients.
- How long deals took from first call to signature.
- The average contract value.
- The total value of the active pipeline.
- How many clients renewed.
The answer to April turned up almost immediately. Lead volume was fine, and so was the qualification rate. Deals were dying after the proposal stage. Prospects liked the discovery meetings, then went silent when they saw the price.
The fix wasn’t more outreach. Reps started talking about budget ranges during discovery, so the number in the proposal didn’t come as a shock. Tom set targets against his own past figures rather than industry averages, because his market wasn’t the average one.
Some CRMs now offer AI pipeline forecasting that flags deals at risk based on things like slow email replies or meetings being pushed back. It’s useful as a nudge, but it only works if reps keep the CRM up to date.
Get sales, marketing and delivery talking
Tom’s marketing contractor had been writing blog posts about cloud migration. Meanwhile, his reps kept fielding questions about cyber insurance requirements. Once they compared notes, the content changed, and inbound leads started arriving better informed. These days some businesses go further and use AI marketing agents to adjust campaigns based on what sales is hearing.
Delivery was the other gap. Reps had occasionally promised two-hour response times that the support team couldn’t meet. Now the onboarding lead tells sales what’s realistic, and every closed deal comes with a written handover covering the client’s goals, the agreed scope and any specific promises made. If clients keep running into a gap between what they were promised and what they get, it shows up in that loop before it costs a renewal.
Review it monthly, change one thing at a time
Every month, Tom’s team spends an hour going through the pipeline. They look at which deals stalled and why, which objections keep coming up, and how close they are to target. The point is to find patterns, not to put individual reps on the spot.
When a few healthcare prospects dropped out after demos, the team pulled the discovery notes. The demo had focused on features those clients didn’t care about and skipped the compliance reporting they’d asked about twice. They changed the demo and left everything else alone. That way, they could tell whether it was the demo that made the difference.
Reps are asked what’s getting in their way too. One pointed out that proposals took four days because Tom had to approve every price personally. He now signs off only on discounts above 10%.
June came in at nine new contracts. July was eight, and Tom still isn’t sure what happened in April. He hasn’t jumped into a deal at the last minute since May, though.
Frequently asked questions
Q. What are the main stages of a sales process?
Most include lead generation, qualification, discovery, a proposal, negotiation and closing, then a handover into onboarding. Your own stages should match how your customers actually decide, so some businesses add steps and others skip them.
Q. How can a business improve its sales process?
Look at where deals stall, using conversion rates and reasons for lost deals, and ask reps what’s slowing them down. Change one thing, measure it against your own past numbers, and keep the CRM up to date so the data stays reliable.
Q. How often should a sales process be reviewed?
Monthly pipeline reviews catch immediate problems, and a broader quarterly review shows longer-term trends. Review it sooner if win rates drop or customer expectations shift.
Related: RPO Services: Why Companies Are Rethinking How They Hire
