Judge TSS automation software on six things: whether it can submit programmatically, what it validates before submitting, how it behaves at your volume, what happens when a submission fails, how it connects to the systems that already hold your data, and whether it leaves an audit trail. Everything else is detail.
This is a buyer’s checklist rather than a recommendation. Choosing TSS software is mostly a matter of asking specific questions and refusing vague answers, so it is written to be handed to a vendor, including us.
Worth asking honestly, because the answer is sometimes no.
TSS is free. GOV.UK confirms there is no subscription and no charge per declaration. If you make a handful of movements a month with a stable product range, the portal is adequate and buying software will not pay back.
The economics change when one or more of these is true:
If none of those apply, save your budget. If three or more do, keep reading.
Do not evaluate software before you fix the data The commonest reason these projects disappoint is that the business bought a tool to solve what was actually a data problem. If your authoritative commodity code lives in three spreadsheets and one person’s memory, no vendor can fix that for you. Sort the product master first, then buy. |
Treat these as your declaration software requirements document. Score every vendor against the same six rather than comparing feature lists, which are written to be incomparable.
| Area | The question it answers | Why it matters |
|---|---|---|
| Submission | Can it submit programmatically, and what? | Determines whether people still type into a portal |
| Validation | What does it check before submitting? | This is what turns faster into better |
| Scalability | Does it hold up at your volume and peaks? | Month-end is when it will be tested |
| Exception handling | What happens when something fails? | The most underestimated area by a distance |
| Integration | Can it reach the data you already hold? | Decides how much manual input remains |
| Auditability | Can you prove what was submitted and why? | You still answer for accuracy, not the vendor |
TSS provides an API, so any serious tool should be able to submit programmatically rather than driving a browser.
Ask which declaration types it handles. The API covers entry summary declarations, simplified frontier declarations, supplementary declarations and full frontier declarations. A tool that only does one of those leaves you with two processes.
Questions worth asking:
For the underlying technical picture, including the test environment and what the interface actually accepts, see what the TSS API can and cannot do.
The single most valuable area, and the one most vendors describe vaguely.
Automating submission without validation just produces wrong declarations faster. Push hard here, and make the questions specific enough that a generic answer is obvious:
A vendor who answers all six concretely is worth shortlisting. One who says the platform validates your data is not answering.
If you want the list of what actually goes wrong, so you can test against it, reducing manual entry mistakes sets out the errors that cause most rejections.
Not just total volume. Shape matters more than size.
Customs work is not evenly distributed. Supplementary declarations for a whole month are due by the tenth calendar day of the following month, which produces a spike whatever your average looks like. Ask about the peak, not the mean.
That last question separates the thoughtful vendors. Item-level effort scales with product variety, not with the number of movements, so a business shipping one pallet of forty different products has more work than one shipping a full load of a single item.
The area buyers ask about least and regret most.
Every submission returns an outcome. Accepted ones are easy. What you are buying is how the tool behaves on the others.
Ask to see this in the demo. Vendors will happily show you a clean submission. Ask them to show you a failed one, then ask what happens next.
The data mostly exists already. The question is whether the tool can reach it.
That last one is a quality question dressed as a technical one. A tool that quietly guesses is worse than one that stops and asks.
Most of what makes an integration succeed sits on your side rather than the vendor’s. getting your product data in order covers the groundwork worth doing before anybody demos anything.
The obligation to declare accurately stays with you. Software does not take it on, and no contract moves it.
Ask the exit question early. The answer tells you a lot about the relationship you are entering.
The fourth question is the most revealing. A vendor with a clear answer has thought about the boundary. A vendor who claims to do everything has not.
List price is rarely the number that matters. Build the comparison from:
Set against that, be equally disciplined about the benefit. Measure your current position first: how many declarations, how many people, how many rejections, how much time between documents arriving and submission. Without a baseline, any saving claimed later is unprovable, including your own.
A note on percentages Be sceptical of accuracy, time-saving or return-on-investment figures presented without a method. Ask how the number was calculated, over what period, for what kind of customer. A vendor who can explain the method is being straight with you. One who cannot is quoting marketing at you. |
| What you hear | Why it should worry you |
|---|---|
| Guaranteed one hundred per cent accuracy | Nobody can guarantee this. The claim itself tells you how the vendor talks about risk. |
| We replace TSS | They do not. TSS is the government route. Anything sitting around it is a layer, not a replacement. |
| Fully automated, no human involvement | The at-risk decision and accountability for accuracy stay with people. Always. |
| A demo using only their own sample data | Ask them to run one of your real invoices, including a messy one. |
| Vague answers on validation | The most valuable area is the one they should be most specific about. |
| No named reference in your sector | Northern Ireland movements have specifics. General customs experience is not the same thing. |
A customs software evaluation works best when it is short, evidence-based and run against your own data rather than a curated demo.
Customs and operations first, IT second, finance third. The hard decisions here are customs decisions. An evaluation run purely by IT tends to produce something technically sound and operationally wrong.
If you are still working out whether the workflow suits you at all, the document to declaration workflow walks through what automation actually changes, and choosing between API and portal submission compares the two operating models directly.
Intermediaries handling many clients have a different set of requirements again, which we cover in TSS software for freight forwarders.
And if you want to see how the pieces fit together in practice, our overview of Northern Ireland customs automation sets out the layer that sits in front of the government service.
Not necessarily. TSS is free to use and adequate for low, stable volumes. Software earns its place when the same data is retyped repeatedly, when documents have to be read by a person, or when declaration volume limits how much work you can take on.
Six areas: programmatic submission, validation before submission, behaviour at your peak volume, exception handling, integration with systems that already hold your data, and a complete audit trail.
Validation, by a distance. Automating submission without checking the data simply produces wrong declarations faster.
Baseline your current numbers, shortlist on the six areas, demo with your own data including a messy document, insist on seeing failure handling, take references, then pilot on one lane and run parallel.
Vendors price differently, by user, by declaration or by volume band. Compare the total: licence, implementation and data mapping, your own IT time, training, support, and the cost of running parallel during transition.
Show me a rejected declaration end to end. Which validations are specific to Northern Ireland movements? Give me a reference with a similar product range. What do you not do? What is the realistic timeline given our data as it is today?
No. The trader answers for the accuracy of what is declared regardless of how it was submitted.
It depends far more on the state of your product and party data than on the software. Businesses with a clean product master move quickly.
Only if you have development capability and the appetite to maintain it as guidance changes. Maintenance is the commitment, not the build.
Buying a tool to solve a data problem. Fix the product master first.
Capture & Upload Data in Seconds with AI & Machine Learning
iCustoms is an all-in-one solution helping businesses automate customs processes more efficiently. With AI-powered and machine-learning capabilities, iCustoms is designed to streamline your all customs procedures in a few minutes, cut additional costs and save time.
Capture & Upload Data in Seconds with AI & Machine Learning