Five examples of processes we set up. This is just a taste. Book the audit to see which ones can help you most.
A funnel does not leak in one place. It leaks in the handovers.
These are ordered by where in the funnel they act, because that order is the useful part. Read them straight through and you have the list of places a B2B funnel loses money, from the arriving lead to the won deal.
The form already asks the question that decides whether a lead is worth a call: company size, revenue, whether they are even B2B. That answer arrives in the CRM as a line of text, and then a person reads it, one lead at a time, usually the next morning.
What changes. The rep’s task list only contains leads that passed the filter, and the task lands in the same minute the form was submitted. Everybody else gets an email sequence instead of a call that was never going to happen.
The rule is the weak part. A revenue band on a form works until somebody adds a new option, or the honest answer is “it depends”, or the real signal was in the free-text field nobody reads. A model reads the whole submission together with the company behind it, the website and the size, and scores it against the profile you actually win.
The rule stays as the floor, because a model should not be the only thing standing between a bad lead and your rep’s morning. And the reason is written on the lead, so the rep can disagree with it.
Take last month’s leads. Count how many got a first contact within an hour, and how many of those were above your own minimum size. If you cannot answer that from your CRM in two minutes, that is the finding.
Between the booking and the call there are usually three days of silence. Nothing confirms the meeting, nothing reminds them, and the rep opens the call without having read what the person wrote when they booked it.
What changes. The opportunity exists from the moment the call is booked, so the pipeline is real rather than remembered. An hour before, the prospect gets a message on the channel they actually read, and the rep gets back what the prospect wrote when they booked.
The reminder does not fire blind. An hour before, the automation looks the meeting up again and sends only if it is still there at that time. Without that step you remind people about calls they already moved, which costs you more trust than the no-show would have.
Count the meetings booked last month against the meetings that happened. If nobody in the company can produce that number, the no-show rate is not being managed, it is being tolerated.
This gets read as a reporting problem, so people typically build more dashboards. It is a data entry problem. After every call a rep is asked to update the status, move the stage, set the next task and write down what happened. Four pieces of admin that sell nothing, so they get skipped, and the pipeline stops matching reality within weeks. Another dashboard on top of that only shows you the gap faster.
So the rep fills in one thing: a template with the questions they wanted to ask on the call anyway. Budget, timing, who decides, what happens next. One field on it records the outcome, and that field is the only input the system needs.
What changes. The pipeline is accurate because nobody maintains it by hand. Every deal carries a stage that means something, an owner and a date. The same single answer is what hands a won job to the ERP, so the order is entered once instead of twice.
Even that one field does not have to be typed. The call is recorded and transcribed, a language model pulls the answers out of the transcript into the template, and the rep confirms instead of writing. Everything below the template fires exactly as before.
The confirm step stays. A CRM full of confidently wrong summaries is worse than an empty one.
Open your pipeline and sort by last activity. Every deal that has not moved in two weeks is either dead or being neglected. If the stages are kept by hand, nobody in your company can tell you which.
Asking for a review is somebody’s job that never becomes the urgent thing. When it does happen it happens weeks late, by email, to a customer who has moved on. The people who would leave the best review are the ones you finish with and then never speak to again.
And the moment you offer something in return, it stops being an email and becomes a small logistics problem: who has already had one, how many are left, and what happens when they run out.
What changes. The ask goes out fourteen days after the job is invoiced, not whenever somebody remembers, and it goes out from the CRM so it sits on the customer’s timeline. Every review is traceable to the job that earned it. If you pay for your leads, your review score is part of what those leads cost you.
A public review is one thing worth paying for. Why they chose you is worth more. The same mechanism runs a short survey with a larger voucher, and the answers land on the customer record as structured fields. What decided it, who else they compared, how they rated the price and the trust, and whether timing or budget was the real constraint.
Twelve months of that is a list of the reasons you win and lose, written by customers rather than by your own team.
The details matter, to make a play like this work fully automatically.
A year of survey answers is a few hundred rows nobody opens. A model reads all of them and returns the ranked reasons you win and lose, per product and per segment, and drafts the reply to each public review in your own voice.
Nothing is published without somebody pressing send. The point is that the reading and the drafting stop being the reason it never happens.
Count the jobs you finished last month and the reviews you received. Then ask who is responsible for the gap. If the answer is a person rather than a process, the gap will be the same size again next month.
The ad platform optimises for whatever you report back to it. If the last thing you report is a form fill, it will get you more form fills, and cheap form fills are the easiest thing in the world to buy.
What changes. The platform learns which campaigns produce revenue instead of which produce forms, and the cost per customer becomes a number you can steer. It only works if the CRM knows which deals were won, which is what play 03 is for.
Open your ad account and look at what your conversion event is. If it is a lead form submission, your budget is being pushed towards whoever fills in forms most cheaply.
30 minutes on your CRM, your leads, and what happens between them. You leave with a written list of what to fix first.
Book the audit