How AI agencies can assess tech client reliability before onboarding

Key takeaways

• Check a client's financial health before you start any project. A formal AI agency client credit assessment prevents you from working with companies that can't or won't pay.
• Use a simple client evaluation checklist for every new prospect. This should cover company background, payment history, and contract terms to standardise your vetting process.
• Implement risk scoring tools to make objective decisions. Assign points to different risk factors like company age and contract value to get a clear go/no-go signal.
• Structure your payment terms based on the client's risk level. Higher-risk clients should be asked for prepayment agreements or milestone payments to protect your cash flow.
• Bad debt can destroy an AI agency's profitability. One unpaid invoice for a large AI model development project can wipe out the margin from several other successful clients.

Why is client credit assessment critical for AI agencies?

An AI agency client credit assessment is the process of checking if a new client can and will pay you. For AI agencies, this is more than just a background check. It's a fundamental part of protecting your cash flow and profitability.

AI projects often involve high upfront costs. You might need to pay for cloud computing credits, API calls, or specialist data engineers before you invoice the client. If that client doesn't pay, you're left covering those costs yourself. Your gross margin (the money left after paying your team and tech costs) disappears.

Many tech startups and scale-ups, which are common clients for AI agencies, can be financially volatile. They might be pre-revenue, burning venture capital, or have unpredictable cash flow. A proper assessment helps you spot these risks early. It lets you decide if you want to work with them, and if so, on what payment terms.

In our experience working with AI agencies, the ones with a formal vetting process have significantly fewer payment problems. They also spend less time chasing invoices and worrying about bad debt. This gives them more mental space to focus on delivering great AI solutions.

What should be on your client evaluation checklist?

Your client evaluation checklist is a standard set of questions you ask about every new prospect. It turns a gut feeling into a structured review. A good checklist covers three main areas: who the client is, how they pay, and what the project involves.

First, gather basic company information. How long have they been trading? Are they a limited company you can check on Companies House? Who are their directors? For tech clients, check their latest funding announcements or ask about their runway (how many months of cash they have left).

Second, investigate their payment behaviour. Ask for trade references from other suppliers. A simple question to a reference like "Do they pay within your agreed terms?" is very revealing. You can also use online tools to check for county court judgments (CCJs) against the company or its directors.

Third, assess the project specifics. What is the total contract value? Is this a one-off project or a retainer? How complex is the scope? Larger, more complex AI integrations carry more risk if the client relationship sours mid-project. Your checklist should prompt you to consider all these factors before saying yes.

We provide a simple starting framework for agencies. Your checklist doesn't need to be fifty items long. Focus on the ten to fifteen questions that give you the clearest picture of financial risk. Specialist accountants for AI agencies can help you tailor this to your specific service offerings and client base.

How do risk scoring tools work for agency vetting?

Risk scoring tools help you make an objective decision by turning client information into a number. You assign points to different risk factors, add them up, and the total score tells you how to proceed. This removes emotion and inconsistency from your AI agency client credit assessment.

Start by defining your risk categories. Common ones include company stability, payment history, and project risk. For each category, create a simple scoring system. For example, a company older than five years gets 0 points (low risk), one to five years gets 5 points (medium risk), and less than a year gets 10 points (high risk).

Other factors to score include the client's industry (a struggling sector scores higher), the size of the contract relative to your average, and whether they provided satisfactory references. The total score places the client in a band: green for "low risk, standard terms", amber for "medium risk, enhanced terms", or red for "high risk, avoid or prepayment required".

You don't need expensive software. A simple spreadsheet can act as your risk scoring tool. The key is that everyone in your agency uses the same system for every new client. This consistency is what protects you over time. It also helps you justify decisions to your sales team if you turn down a risky but exciting project.

According to commercial credit experts, formalising your assessment process significantly reduces bad debt rates. It forces you to look at facts, not just the appeal of a project.

When should you ask for a prepayment agreement?

You should ask for a prepayment agreement when your client credit assessment shows elevated risk. This means getting some or all of the project fee upfront before any work begins. It's a standard way to protect yourself when a client's financial position is unclear.

Prepayment agreements are common with certain types of AI agency clients. If you're working with a very early-stage startup that has just raised a seed round, asking for 50% upfront is reasonable. They have the cash from investors, and it secures your commitment. For one-off projects with new clients, regardless of size, a milestone payment structure is wise.

Another scenario is when the project requires significant upfront investment on your part. If you need to provision £5,000 in AWS credits to train a model, getting that cost covered before you spend it is simply good business. Frame it as standard practice, not a lack of trust.

The structure of prepayment agreements can vary. It could be a simple 50% deposit to start, with 50% on delivery. For longer projects, you might use a monthly prepayment against a retainer. The goal is to align cash inflow with your cost outflow. This prevents you from becoming an unwilling lender to your client.

In your contract, clearly state what happens to the prepayment if the project is cancelled. Typically, it covers work done to date and any non-recoverable costs. Having this clarity upfront avoids difficult conversations later.

What are the red flags in a tech client's financial health?

Red flags are warning signs that a client might struggle to pay you. For AI agencies dealing with tech clients, some flags are specific to the industry. Spotting them early is the whole point of a thorough AI agency client credit assessment.

A major red flag is a company that is constantly "about to close a funding round". While common in tech, it means their cash position is precarious. If they haven't secured the money, they can't guarantee they'll have it to pay your invoice in 30 days. Be wary of clients who want to start work "now" but delay payment until their "round closes".

Check Companies House for filed accounts. Consistently late filings can indicate disorganisation or financial trouble. Look at the balance sheet in their latest accounts. Do they have retained losses piling up? Is their cash balance very low? These are quantitative red flags.

Behavioural red flags are also important. Does the prospect refuse to provide a trade reference or sign a standard contract? Do they push back aggressively on your payment terms, insisting on 90-day payment without negotiation? This often signals how they will behave as a client.

Finally, consider the project itself. Is the scope vague or constantly changing before you've even started? A client who doesn't know what they want is a high risk for scope creep and disputes over invoices later. All these factors should feed into your risk scoring tools and final decision.

How do you balance risk with business growth?

Balancing risk and growth means not saying no to every slightly risky client, but not saying yes to everyone either. Your AI agency client credit assessment should have a middle ground. This is where you accept the client but structure the engagement to protect your agency.

For clients in the "amber" risk band, you might proceed but with modified terms. This could mean a smaller initial project to build trust, rather than a large annual retainer. It could mean stricter payment terms, like payment on receipt of invoice rather than 30-day terms. The contract value might also trigger a different approach.

Another balance tactic is to limit your exposure. If a risky client wants a £100,000 AI integration, could you phase it into four £25,000 milestones? Each milestone is paid for before the next one begins. This caps your maximum loss if things go wrong and improves your cash flow throughout the project.

Growth often requires taking on some risk. The key is to be conscious of the risk and to price for it. A higher-risk project should carry a higher margin to compensate for the increased chance of payment delays or additional admin. Don't compete on price with high-risk clients; you need the extra profit as a buffer.

Using your client evaluation checklist and risk scores consistently helps you see patterns. You might find that certain types of clients, like pre-series A startups in a niche sector, are consistently amber. You can then create a standard "amber risk" package of terms for them, making the process efficient for your team.

What legal and contract safeguards should you use?

Your contract is your last line of defence if a client doesn't pay. It should include clear safeguards that support your AI agency client credit assessment findings. Standard terms for all clients are a good start, but you can add specific clauses for higher-risk situations.

Every contract must have clear payment terms. State the exact number of days for payment (e.g., "Payment is due within 14 days of the invoice date"). Include your right to charge statutory interest on late payments under UK law. This isn't being harsh; it's your legal right and sets a professional tone.

For higher-risk clients identified in your assessment, consider adding a "right to suspend work" clause. This allows you to pause delivery if an invoice becomes overdue. In the fast-moving world of AI development, a client relying on your work will often prioritise payment to avoid project delays.

Ownership of intellectual property (IP) is crucial for AI agencies. Your contract should state that IP (like code, models, or datasets) only transfers to the client upon full and final payment. This gives you leverage if payment fails. It also ensures you don't hand over a valuable AI solution for free.

Finally, define the process for scope changes. AI projects often evolve. Your contract should require a formal change order, signed by both parties, for any work outside the original scope, with associated fees and timelines. This prevents "scope creep" from eroding your margin and leading to billing disputes. For complex engagements, getting a legal review of your master services agreement is a wise investment.

How can your finance team support client assessment?

Your finance team, or your external accountant, should be a key partner in the client assessment process. They bring a numbers-focused perspective that complements the sales team's desire to win work. Involving them early turns credit control from a reactive chore into a strategic advantage.

The finance function can run the formal checks. This includes pulling reports from Companies House, analysing filed accounts, and using commercial credit reference agencies if you subscribe to one. They can calculate key ratios from a prospect's accounts, like their current ratio (assets vs liabilities), which indicates short-term financial health.

They can also own and maintain the risk scoring tools. By tracking which clients from each risk band actually pay on time, they can refine your scoring model over time. This data-driven improvement makes your AI agency client credit assessment more accurate.

Finance can set clear policies for the sales team. For example, "Any new client with a risk score above X requires director sign-off" or "Contracts over £Y require a 30% deposit." This gives commercial teams clear guidelines and speeds up decision-making.

If you don't have an in-house finance team, this is where a specialist accountant adds huge value. A good accountant for AI agencies doesn't just do your tax return. They help you build commercial systems like client vetting that directly protect your profitability and cash flow.