Start here: follow one customer from enquiry to a useful answer. Separate time spent working from time spent waiting. Choose one delay to fix, then decide whether AI has a useful part to play.
When I was leading acquisition and retention teams at an agency, I had ideas about how we could bring in clients, onboard them and help them stay. The CEO heard me out, but the company was going in a different direction. It was a good place to learn. I also realised I wanted more freedom to try things.
Later, I worked inside a company to understand how decisions affected the product as well as the marketing around it. That connection still matters to me. An advertisement can bring someone in, but what happens next involves the rest of the business.
It is also where I would start with AI business automation: choose a problem you can observe, understand why it happens, and test whether AI helps you solve it. Buying a tool is only one part of that work.
Find the delay before choosing the tool
Imagine a business receiving enquiries that take two days to answer. An AI assistant could draft replies quickly. But suppose the delay happens because nobody knows who should reply, or because every quotation needs the owner’s approval. Faster writing may leave most of the waiting untouched.
This is an illustrative example, but you can check for the same problem in your own business. Follow five recent enquiries from arrival to the next meaningful step. Record who handled each one, what information was missing, and where it waited. Five cases will not establish a reliable benchmark. They can show you where to investigate.
Separate the time spent doing the work from the time spent waiting. A ten-minute task can sit untouched for two days. Those two problems need different changes.
A completed example: where an enquiry waits
Here is a made-up service business, following one enquiry. The times are illustrative. They show how to separate work from waiting; they are not client results or a target for your team.
- Enquiry arrives
09:00 Monday
The customer asks about a service. - It waits for an owner
24 hours waiting
Nobody has picked it up. - A reply is drafted
10 minutes of work
The team uses approved information. - Approval holds it up
4 hours waiting
A routine reply needs the founder. - The customer gets an answer
13:10 Tuesday
The next step is now clear.
I would start by giving the enquiry an owner and agreeing which routine replies they can send. If the team still spends too long writing, then I would test AI drafting. That order gives each change a job.
A small test for this example
- Change: the person on the enquiry rota owns the next step. They can send replies covered by the approved service notes; exceptions go to the founder.
- Measure: time from enquiry to a useful answer, plus corrections and customer follow-up. An automatic acknowledgement does not count as an answer.
- Review: look at the next five comparable enquiries. Did waiting fall? Did mistakes or repeated questions increase? Treat this as a clue for further testing, not proof from a large sample.
For a worked way to review those conversations, use the five-enquiry review.
Your free process note
Copy these five lines into a note. Fill them in for one customer journey before you choose a tool.
- Customer needs: what are they trying to do?
- Work and waiting: what happened, when, and where did it pause?
- Next owner: who can move it forward, and what can they decide?
- One change: what will we try on the next few cases?
- Review: did the customer get a useful answer sooner, with no extra corrections?
Keep customer names and sensitive details out of any AI tool unless your business has approved that use. The process is what you are reviewing.
How we prioritised marketing requests on one Monday.com board
When I was CMO at a fintech company, I came in to address bottlenecks in content creation, landing pages and design. Requests were coming from different people and teams. There was no shared filter for what was relevant, and no clear way to prioritise the work with the highest business impact.
We created one board on Monday.com. Any person or team wanting to put an initiative forward had to submit it there. We then prioritised the requests using estimated revenue and business impact, deciding which initiatives went into the production queue.
We had been getting through around five initiatives a week. With the shared process, that became roughly seven or eight. There was less arguing about which task should come first. The people responsible and the stakeholders had more time to work on the tasks themselves.
These are approximate weekly counts from my experience, not a standardised efficiency measure. They do not account for differences in the size or difficulty of each initiative.
The part I would repeat is asking people to explain the initiative and its expected impact before it enters production. The revenue estimate is an assumption to examine, not revenue the business has already earned.
How we use AI for this now
We have since developed an AI system that client teams are using to assess tasks and recommend priorities before production. It gives the teams a clearer view of how to prioritise the work internally. Its outputs require human approval.
We are still improving the system as the models improve. It is already being used by clients; the people reviewing its outputs remain part of the process.
The weekly initiative counts above describe the earlier Monday.com process change, not a measured result from the AI system.
A request-to-handoff template you can adapt
The fields below are a suggested extension of that approach, not a copy of our original board. Use them in the system your team already has.
- 01 / Describe the request
- What are we trying to change, for whom, and why? Put the relevant information in one place before asking another team to produce something.
- 02 / Explain the expected impact
- What revenue or business outcome might this affect? Record the estimate, its assumptions and what supports it. Mark what you do not know.
- 03 / Decide its priority
- Consider it alongside the work already proposed. Record whether to proceed, wait or ask for more information, and why. The request needs a decision before a production commitment.
- 04 / Make the handoff clear
- For accepted work, record the owner, next action and acceptance rule. Make it clear what must be ready and who checks that the output answers the request.
Copy the request and handoff note
Use one note per initiative. These are proposed fields to adapt, not a required Monday.com setup.
REQUEST: What should change, for whom, and why? EXPECTED IMPACT: What revenue or business outcome might change? What supports the estimate? PRIORITY DECISION: Proceed, wait or clarify? Why, compared with the other work? READY: What information is needed before production can start? OWNER AND ACTION: Who takes the next step, and what will they do? ACCEPTANCE: Who checks the output, and what makes it ready to use?
Review the queue as well as the output
As a small test, follow a few comparable requests from submission to a priority decision and then to an accepted output. Record time spent waiting and anything returned for clarification or correction. Keep those measures separate from whether the finished work later produced revenue.
If you use AI to organise or draft requests, keep the assumptions visible. Someone still needs to judge their relevance, agree the priority and own the work.
Use the customer journey above to find a delay, or review five enquiries to decide what needs attention. If you would like to work through a bottleneck together, book a free consultation.
When does AI actually improve efficiency?
There is evidence that AI assistance can help. A study of 5,172 customer-support agents found an average 15% increase in issues resolved per hour after an AI assistant was introduced. The largest benefits went to less experienced and lower-skilled workers. This was one company’s support operation, so it is evidence of a useful application, not a promise for every business. Read the published study, Generative AI at Work.
A reasonable first candidate is a repeated task with clear inputs, an outcome you can check and mistakes you can catch before they reach a customer. Drafting a reply from approved service information might fit. Making an exception to a contract or promising a delivery date the business cannot meet needs a different level of control.
Some tasks are better served by a simpler change. A shared owner, a shorter form or a standard reply may solve the problem without adding another system to maintain.
Run one small test with a clear owner
Write a short test plan before connecting anything:
- The problem: describe what is happening, with a recent example.
- The proposed change: say exactly what the AI will do and what a person will check.
- The outcome: choose a measure such as time to a useful reply, correction rate or successful customer onboarding.
- The owner: name the person who reviews exceptions and can stop the test.
Use information the business is authorised to process, and begin with the minimum access required. A drafting assistant does not need permission to send messages or change customer records just to prove that its drafts are useful.
Include the time spent checking and correcting the output. If drafting saves twenty minutes but reviewing adds twenty-five, you have learned something worth knowing. Faster generation is not enough to establish a saving.
Check what happens to the next team
In Alchemy Avenue’s SmartHeart case study, the agency reports that a change to the buying journey helped more customers complete their purchases. The increased demand then put pressure on the teams delivering the service. This was a customer-journey improvement, not an AI experiment.
It illustrates why an automation test should include the next step. If more enquiries become customers, can onboarding handle them? If replies become faster, are they accurate enough that customers do not have to contact you again?
I wanted to understand these connections when I moved closer to the product. They are still the questions worth asking when the new tool happens to use AI.
