When a company decides to automate its work, almost every department has a reasonable request. Finance wants fewer spreadsheets. Operations wants fewer mistakes. Sales wants less data entry. Management wants reliable reports.
My starting point is the customer.
I believe CRM should be one of the first directions a company considers for automation. The ability to sell quickly and well can matter more than having an excellent product that you cannot sell. A good product still needs a company that can find customers, understand their needs, follow up, and turn a conversation into an order.
That is why I would begin by examining how the business works with customers. Then I would bring together the right implementation team and decide which parts of that work need structure, which need automation, and where AI can help.
Why Start with CRM?
Customer relationship management, or CRM, is about organising the relationship between a company and its customers. The software supports that work, but the business purpose comes first.
A company may already receive orders. Some customers may buy almost automatically because they know the product and have purchased before. That does not necessarily mean the company has an organised sales process.
Can you explain where those customers came from? Do you know what happened before their first order? Can another salesperson continue the conversation if the original employee is absent? Do you understand why one customer returned and another disappeared?
These questions matter because sales should be something the company can understand and improve. If knowledge remains in personal inboxes and employees' memories, management sees only part of the picture.
CRM helps make that picture visible. It connects customer information, communication, responsibility, and the next action. That gives the company a foundation for both sales process automation and better decisions.
This is my recommendation, not a rule that overrides every other business problem. A critical financial or operational failure may need attention first. But when a company asks where to begin improving routine work, I want to understand how it acquires and serves customers before discussing a long list of automations.
Start with the Implementation Team
If I were beginning a CRM implementation, my first step would be to assemble the team.
The team needs both technical and organisational competence. It also needs people who understand the business. When only technical specialists participate, the project can become focused on making a program work rather than making the company work better.
I would involve people from sales, product, and customer service, together with a business representative who can make decisions. Finance may also have an important role, particularly when sales planning, payment conditions, and the reliability of the forecast affect the project.
Each brings a different part of the customer relationship:
- Sales understands the conversations, objections, and decisions that move a deal forward.
- Product specialists understand what the company can offer and how that offer addresses a customer's need.
- Customer service sees what happens after the sale, including recurring questions and reasons for dissatisfaction.
- Finance can challenge assumptions about expected sales and explain the commercial information the business needs.
- The business owner or sponsor sets priorities and resolves conflicting requirements.
- Implementation specialists translate those requirements into records, workflows, integrations, and usable controls.
With this combination, CRM implementation becomes a business project. The goal is no longer simply to introduce software. It is to help the company plan sales, understand why it loses customers, compare sales performance, and identify practices that deserve to be repeated.
I discuss the importance of evaluating the people behind a project in my guide to choosing an implementation consultant. The same principle applies here: the team's ability to understand the business matters alongside its technical skills.
Make the Customer Journey Visible
Before automating individual actions, agree on what the company needs to know about a customer.
I would begin with the source of the relationship. Was the customer introduced by someone, attracted by a campaign, or already known to the company? What prompted the first enquiry? What was the customer trying to achieve?
Next, introduce a clear way to work with potential customers. A lead represents a potential customer or enquiry that still needs qualification. Qualification establishes whether there is a relevant need and a realistic opportunity to do business together.
Once qualified, the lead can become an opportunity and move through the stages of your sales process. The exact stages should reflect how your customers buy. They should help the team understand progress, not merely provide labels for a colourful pipeline.
For each opportunity, the team should be able to explain:
- What the customer needs and what has already been discussed.
- Who is responsible for the relationship.
- What the current stage means and what evidence supports it.
- What needs to happen next, who will do it, and when.
- When a buying decision is expected.
- Why the opportunity belongs in the current sales forecast.
Follow the relationship through to the first order and subsequent purchases. This helps distinguish attracting a new customer from maintaining an existing relationship. Both matter, but they involve different work.
For a closer look at the records and settings behind this approach, see my practical explanation of CRM leads, opportunities, and activities in Odoo. Here, the important question is what the business needs those records to help people do.
Use CRM to Understand Sales, Not Just Record Them
A useful CRM should help the company build a sales plan and explain what is happening against that plan.
An expected closing date is valuable when it reflects the customer's situation. A forecast category is valuable when the team understands why a deal belongs there. If these fields are filled in only because the system requires them, management receives a report that looks organised without necessarily being trustworthy.
The same applies to performance comparisons. It is useful to know who sells more and who sells less. But I also want to understand the practices behind those results.
Does one salesperson ask better questions? Follow up more consistently? Involve a product specialist at the right moment? Work with a different type of customer?
CRM can make the relevant evidence easier to examine. Management still needs to interpret it in context. A count of calls alone does not explain the quality of the conversations.
Likewise, when you compare conversion rates, compare clearly defined groups over a suitable period. Leads created this week and deals closed this week may belong to different customer journeys.
Which CRM Processes Should You Automate First?
Once the team agrees on the process, look for routine work that prevents employees from responding, following up, or understanding the customer.
I would examine the following areas before building a complicated automation roadmap.
1. Incoming Lead Handling
An enquiry should enter the system with the information needed to begin work. It should have an owner and an appropriate first action.
Lead management automation can connect the incoming channel with CRM, preserve available source information, assign responsibility according to agreed rules, and create a task. The first question is whether enquiries currently wait, disappear, or require repeated manual entry.
2. Qualification and Handoffs
Employees should not repeatedly ask for information the company already has. Nor should a qualified prospect lose context when moving between people.
Automation can help collect required information, flag incomplete records, and hand the work to the right person. The qualification criteria remain a business decision. A workflow cannot resolve disagreement about what makes an enquiry relevant.
3. Follow-Up and Nurturing
A potential customer may have a real need without being ready to buy today. Lead nurturing keeps the relationship organised over time.
That can mean scheduling a relevant follow-up, sharing useful information, or reminding the salesperson to return to a conversation at an agreed time. The communication should reflect the customer's stage and responses.
Automating nurturing should support a relationship. Sending the same sequence regardless of what the customer says undermines that purpose.
4. Opportunity Review and Next Actions
Open opportunities need attention even when nobody has changed their stage. A deal can remain in the pipeline while the customer has stopped responding and the next action is unclear.
Automation can identify missing actions and overdue reviews. AI-assisted analysis can go further by examining the available communication and suggesting what should happen next.
5. Call Review and Preparation
Customer calls contain information that does not always make it into a structured record. What did the customer object to? Which question went unanswered? What did the salesperson promise?
Where call recordings or transcripts are available for the agreed workflow, AI can help analyse that content and prepare recommendations for the next conversation. The useful result is a better-prepared employee who understands what to clarify, explain, or follow up.
Choose the First Workflow by Business Value and Readiness
Starting with CRM still leaves a practical choice: which workflow comes first?
Use a short assessment rather than assuming that the most sophisticated automation will be the most useful.
| Question | What to establish |
|---|---|
| Where does customer work break down? | Identify unanswered enquiries, missed follow-ups, incomplete handoffs, or unreliable opportunity reviews. |
| How often does it happen? | Measure the volume and manual effort instead of relying only on memorable incidents. |
| Is the process clear? | Agree on the trigger, owner, expected action, and exceptions. |
| Is the information available? | Check customer records, communication history, data quality, and the systems that must connect. |
| What will it cost to operate? | Include implementation, integration, training, subscriptions, maintenance, and human review. |
| What would count as improvement? | Choose a measure such as response time, fewer overdue actions, less manual entry, or more useful deal reviews. |
For example, if incoming enquiries have no consistent owner, a simple registration-and-assignment workflow may deserve attention before AI call analysis. If enquiry handling already works well but managers spend hours reviewing stale opportunities, opportunity review may be a better first project.
These are illustrative choices, not claims about a particular ERPixel customer. The point is to connect the investment with the actual constraint.
When estimating ROI, distinguish released capacity from reduced spending. Saving a salesperson time creates an opportunity to use that time better; it does not automatically reduce payroll costs or produce additional sales. Our article on measuring implementation ROI provides a broader discussion of evaluating a system investment.
What AI Should Add to the Sales Process
I see value in AI when it helps the team draw better-supported conclusions from customer communication and decide how to act.
A useful opportunity review can combine the available CRM information, correspondence, and call history. It can identify what is missing, explain why the opportunity needs attention, and recommend a next step.
For example, a recommendation might be to clarify a requirement, confirm a buying date, involve a product specialist, or wait until an agreed follow-up point. These are possible recommendations, not conclusions to apply to every deal.
The result should appear where the employee works. If the analysis lives in a separate tool and someone must manually translate it into a CRM task, part of the operating problem remains.
Our AI sales and CRM automation example explains an ERPixel workflow that connects opportunity context with structured recommendations and follow-up actions.
AI recommendations also need review appropriate to the decision. Customer-provided facts should remain distinguishable from generated interpretations. Better context can support a more useful analysis, but it does not make every AI conclusion correct.
What ERPixel Can Implement
ERPixel is a professional implementation team that can deliver CRM end to end: from understanding the customer process and defining responsibilities to configuring the system, connecting channels, and introducing automation.
We can automate incoming lead handling, qualification handoffs, nurturing, and routine follow-up. We can also implement AI-assisted deal analysis that explains recommended next steps, and call analysis that helps employees prepare for the next customer conversation.
The scope can extend across the systems involved in that work. Our business process automation services cover workflows and handoffs, while our AI automation services address processes that benefit from interpreting communication and business context. Where systems need to exchange information, integration work connects the required records and actions.
The goal is practical: make routine work faster and help people reach more informed conclusions about customers. The right solution depends on the process, available information, systems, and level of human control required.
Begin with a Customer Problem You Can Explain
My recommendation is to begin with customers, bring the right people into the implementation team, and make the sales process understandable before adding complex automation.
Choose one recurring problem. Explain who encounters it, what information is missing, what action should happen, and how you will recognise an improvement.
CRM then becomes the foundation for more organised customer work: a way to plan sales, understand losses, learn from effective practices, and support the next conversation.
If you want to apply this approach to your company, book a conversation with ERPixel. Bring an example of an enquiry, a follow-up, or a deal review that does not work well today. That is a useful starting point for deciding what to automate first.