We wrote this because we would rather lose a sale at survey stage than deliver a project that disappoints. Almost every difficult automation deployment we have seen was difficult for reasons that were visible before purchase and simply not checked. Work through the 24 points below and you will negotiate from a much stronger position — with any vendor.
A. Your dispensers (the part that decides everything)
- List every dispenser by make, model and approximate age. Not the brand on the canopy — the plate on the unit. Mixed fleets are normal and fine, but they must be known.
- Establish whether each unit has a working pulser output. This is the single most important technical fact about your forecourt. No pulser and no supported protocol means no automation on that pump, whatever the software brochure says.
- Check whether a previous contractor already tapped the pulser. Old abandoned wiring is common and can complicate a clean installation.
- Note which nozzle dispenses which product. Sounds trivial. Gets configured wrong surprisingly often, and produces a permanent phantom variance when it is.
- Find out when your meters were last calibrated. Automation will expose meter drift immediately, and you would rather know it was drift than spend three weeks suspecting your staff.
- Ask whether any pump is scheduled for replacement. Automating a dispenser you plan to replace next year is money spent twice.
B. Your tanks
- Count tanks and record the product in each. Include any tank currently out of service, and say so.
- Do you have the strapping chart for each tank? Converting a level reading into a volume is impossible to do accurately without it. If the charts are lost, that needs solving before commissioning, not after.
- Are there existing ATG probes, and do they work? A dead probe left in place from a previous system is worth knowing about.
- When was each tank last tested for tightness? If automation reveals a leak in month one, you want to know whether that leak predates the software.
- Do you currently measure water level? If not, expect a surprise. It is better as a planned finding than a customer complaint.
- Is tank access safe and unobstructed? Probe installation needs physical access to the tank top.
C. Power and connectivity
- Document your power reality honestly. Hours of mains, generator switchover behaviour, how long the gap is. A vendor who does not ask about this has not deployed in your market.
- Is there UPS protection for the control room? Controllers and servers should not be hard-cycled by every switchover.
- What is your actual internet connection, and how often does it drop? Not the advertised speed — the lived experience.
- Insist the system works offline. If dispensing, POS or card authorisation depends on a live internet connection, you will stop trading during outages. This is non-negotiable, not a nice-to-have.
D. People and process
- Decide who owns this project internally. Deployments that belong to nobody stall. It needs one named person with the authority to make decisions.
- Be realistic about staff literacy and languages. Training and interface language must match the people who will actually use it, or they will work around it.
- Expect resistance, and plan for it. Attendant accountability is the point of the exercise. Some staff will not welcome it. Better to anticipate that than be surprised by it in week two.
- Identify who reads the variance report every day. A daily reconciliation nobody reads is an expensive log file.
E. Your existing records
- Establish what needs migrating. Customer accounts, outstanding credit balances, supplier records, stock. Migration is usually the least glamorous and most underestimated part of the project.
- Reconcile your credit balances before migration, not after. Importing a receivables figure you do not believe means you will never trust the new system either.
- Get a baseline. Reconcile at least one product manually for thirty days before go-live. Without a baseline you cannot demonstrate what automation changed — and you will want that evidence.
- Write down what “success” means, in numbers. “Better control” is not measurable. “Daily variance below 0.15% within three months” is.
The questions that expose a weak vendor
Once you have the answers above, these questions become genuinely diagnostic. A vendor who has done this work will answer them comfortably. One who has not will get vague.
- “Will you survey my forecourt before quoting, and put compatibility in writing?” A firm quote issued without a physical survey is a guess you will pay for later.
- “What happens to dispensing and card authorisation when the internet is down?” Listen for a specific architectural answer, not reassurance.
- “Which of my specific dispensers cannot be integrated?” An honest vendor names some, or names the ones they need to inspect. “All brands supported” is a marketing claim, not an engineering one.
- “Who owns my data, and can I export it?” The answer should be immediate and unambiguous.
- “Can I speak to a station like mine that you have deployed?” This one question is worth more than an entire proposal document.
- “What is the total first-year cost, including hardware, cabling, training and support?” Licence cost alone is not the number you will actually pay.
- “Who fixes it, how fast, and are they in this country?” A forecourt controller failing on a Friday is not a ticket-queue problem.
- “What does the calibration and metrology position look like?” A competent vendor reads the dispenser’s existing metering output and does not touch the certified measurement path — and will say so unprompted.
Two things worth deciding in advance
Do not automate everything at once. If you have several sites, pick one — ideally your most awkward, not your easiest — and get it genuinely stable before touching the second. The second deployment benefits enormously from what the first one teaches, and a network rolled out simultaneously multiplies every mistake by the number of sites.
Insist on a parallel run. Keep your existing manual process alongside the new system until the two agree. It is tedious and it is the only thing that gives you real confidence in the numbers before you start making decisions on them.
And one honest caveat
Automation is not always the right answer. At very low throughput, the hardware and installation cost may not be recoverable within a reasonable period, and a disciplined manual routine with daily dips may genuinely serve you better for now. If your volumes are marginal, ask a vendor to show you the payback arithmetic rather than the feature list — and be suspicious of anyone unwilling to do that sum with you.
Related: Fuel automation platform · Wetstock reconciliation guide · Independent consulting
General guidance, not engineering or regulatory advice. Tank and metering work must be carried out by appropriately qualified contractors under the requirements applicable in your jurisdiction.