Library

Automation plays

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.

Play 01
When the lead arrives

“Every lead looks the same until somebody calls it.”

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.

Form filled
Answer checked
against your rule
Task on the rep,
same minute
Nurture sequence

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 upgrade level

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.

Form filled
Submission and company
read together
Scored, with the reason
written on the lead
Nurture sequence

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.

Check yours

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.

RevKlar Tool Stack
Meta AdsLinkedIn AdsZapierCloseBrevoClaudeGPT
Play 02
Getting the meeting to happen

“They book a call and then they don’t turn up.”

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.

Call booked
Opportunity opened
at the right stage
One hour before,
the booking is checked again
Reminder to the prospect
Their own answers
sent to the rep

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 detail that matters

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.

Check yours

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.

RevKlar Tool Stack
ZapierCalendlyCloseWhatsAppOutlook
Play 03
Keeping the deal moving

“We have no idea what our sales team actually does.”

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.

Rep fills one template
after the call
The outcome field
is read
Lead status, pipeline stage,
owner and date all set
Won job handed
to the ERP

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.

The upgrade level

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.

Call recorded
and transcribed
Answers pulled
from the transcript
Template filled,
rep confirms

The confirm step stays. A CRM full of confidently wrong summaries is worse than an empty one.

Check yours

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.

RevKlar Tool Stack
CloseZapierMake.comHero SoftwareClaudeGPT
Play 04
After the sale

“Our customers are happy. Our reviews don’t show it.”

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.

Job invoiced
Invitation scheduled,
14 days after the job
Review lands
A voucher code is pulled
and marked used
Code emailed and logged,
customer flagged

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.

The same engine, for more than a rating

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.

Survey submitted
Answers land as
structured CRM fields
Larger voucher
pulled and marked used
Emailed from the CRM

Twelve months of that is a list of the reasons you win and lose, written by customers rather than by your own team.

What makes it survive contact with reality

The details matter, to make a play like this work fully automatically.

  • A field on the customer marks that they already had a voucher, so nobody is paid twice.
  • The codes live in a list, and each one is marked used the second it goes out.
  • When fewer than six are left, the office gets a warning. When they run out, the request does not fail silently: two named people are told, with the address to send to by hand.
  • If the customer is not in the CRM, the same thing happens. A person is asked rather than the automation guessing.
Reading the answers back

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.

A year of
structured answers
Read together
Ranked win and
loss reasons
A drafted reply per review,
for approval

Nothing is published without somebody pressing send. The point is that the reading and the drafting stop being the reason it never happens.

Check yours

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.

RevKlar Tool Stack
ZapierHero SoftwareGoogle SheetsTrustpilotHeyFlowCloseClaudeGPT
Play 05
Feeding the machine

“I want the ads to bring customers, not random leads.”

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.

Deal won
Value and currency
attached
Conversion sent back
to the platform

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.

Check yours

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.

RevKlar Tool Stack
CloseZapierGoogle AdsMeta AdsLinkedIn Ads

We analyse where your
funnel loses revenue.

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