Automation is the easiest thing in marketing to sell and the easiest thing to buy badly. The demo always works, because the demo is a happy path with clean data and no angry customer in it. Growth Bully, a Malta performance marketing agency, is regularly asked to repair lead flow that a tool was bought to improve.
The failure is rarely the software. It is that nobody agreed what the system was for, who owns it when it misfires, and what happens to the data and the workflows when the contract ends.
What should automation actually do first?
Replace one repetitive process that is already defined and already measured. Automation applied to a working process makes it faster. Applied to a process nobody has written down, it industrialises the confusion. Pick the step that happens most often, costs the most when it is missed, and has an obvious right answer.
In most small businesses that step is the first response to an enquiry, because it is high volume, time critical and consistently dropped when the day gets busy. That is the argument for fixing speed to lead before anything more ambitious. Automation that reliably answers every enquiry inside minutes beats a clever system that does nine things adequately.
Which questions separate a real vendor from a demo?
Questions about failure, not features. Anyone can show a workflow succeeding. Ask what the system does when the input is messy, when an integration is down, when a customer replies something unexpected, and when the output is wrong. The answers tell you whether the thing has been run in production or only on a stage.
- Which exact process does this replace, and what is it costing us now? If neither side can answer in a sentence, the project has no success condition and will be judged on feeling.
- Show me the workflow as a diagram, not a screen. Every trigger, every branch, every point where a human can intervene. Vagueness here is the single best predictor of trouble later.
- What happens when it is unsure? A good system escalates to a person. A bad one guesses confidently and sends the guess to a customer.
- Who else is involved? Most automation is a stack of other companies. Ask which subprocessors touch your data and where they sit.
- What does the handover look like if we stop? Ask it during the sale, when you still have leverage, rather than during the exit.
Write the answers down and attach them to the proposal. A vendor who answers all five plainly is usually worth more than one with a longer feature list, because you are buying their judgement about edge cases as much as their software.
Who is accountable when automation gets it wrong?
One named person inside your business, always. The supplier is accountable for the system behaving as specified. You remain accountable for everything it says to a customer in your name. That split has to be written down before launch, because it decides who checks the outputs and how often.
In practice that means someone reads a sample of what the system sent, every week at first, then monthly once it is stable. Automated messages drift quietly as data changes around them. The most common failure we see is not a dramatic error, it is a sequence that kept sending for months to people who had already bought, because nobody was reading it.
What should you agree about your data before you start?
Where it is stored, who can see it, how long it is kept, whether it is used to improve anyone else's product, and how you get it back. Agree all five in writing. Customer contact data is one of the few genuine assets a small business owns, and it is routinely signed away in an onboarding form nobody read.
Keep the system of record under your control, then let tools connect to it. That ordering matters more than any single feature choice, and it is the main reason a properly built CRM is worth doing before the automation layer on top of it. If the vendor holds the only copy of your contact history, every future decision is made with their pricing in mind.
How do you test automation before it touches real leads?
Run it narrow, run it supervised, and compare it against what you do now. A pilot is not a trial period with everything switched on. It is one process, one segment, a fixed window, and a number you agreed to beat before you started.
- Pick one process and one audience. New enquiries from a single source is ideal, because the volume is predictable and the baseline is known.
- Keep a human in the loop for the first two weeks. Every outbound message gets read before it sends. It is tedious and it catches almost everything.
- Agree the measure in advance. Response time, qualified conversations, or meetings booked. Not time saved, which nobody can audit.
- Keep a control. Leave a comparable slice running the old way so you can tell improvement from seasonality.
- Review on a fixed date. Pilots without an end date become production by accident, usually without anyone owning them.
If it clears the number, widen it one process at a time. If it does not, you have lost a few weeks rather than rebuilt your whole operation around a tool that never worked. The same staged logic applies when you automate the follow up itself, which is covered in lead follow up automation.
What do you keep if you change provider?
Your contact data in a usable export, your message history, and a written description of the workflows. Logic built inside a platform rarely transfers, so expect to rebuild it. What you must never lose is the data and the record of what was sent to whom, because that is what makes the next system work on day one.
Ask for a sample export during the sale rather than taking the word portable on trust. A supplier who can produce one in a day is telling you something real about how they are built. This is the same ownership question that decides whether your lead generation is an asset you own or a service you rent, and it is worth settling before the contract is signed. If you are weighing up a system now and want the questions pressure tested first, send us the proposal and we will tell you what is missing from it.