AI for marketing agencies: what clients still pay for

AI for marketing agencies is most useful when it improves research, production and delivery around a clear client problem. Faster copy, ads or websites do not decide who to target, what to promise or how an enquiry becomes business. Clients can still pay for those decisions and for someone to put the work into practice.

That distinction runs through Nas Vou’s CEO Insights conversation with Matt Raad, co-founder of eBusiness Institute. Matt discusses the question as someone who coaches digital business owners and buys and improves businesses with his wife, Liz. His view is optimistic, but the practical lesson for a marketing agency is specific: review what the client buys, how you deliver it and where the money is made.

Will AI replace marketing agencies?

AI changes the work inside an agency. That is different from establishing that the whole business will disappear. If a client mainly buys basic pages or undifferentiated posts, cheaper production creates pressure on that offer. Calling the same package “AI powered” does little to explain why the client should keep paying.

Matt acknowledges disruption. His argument is that business owners still need help getting the foundations right and implementing the work. He describes lawyers, accountants and plumbers who are busy doing their own jobs. They may be able to generate a website, but they still need the digital marketing around it to bring in enquiries.

This is Matt’s assessment of the opportunity, rather than proof that every agency is protected. The test for your business is whether the client would still need your judgement, implementation or support if the first draft became cheaper tomorrow.

What clients still pay a marketing agency for

Matt draws a line between building a website and getting the phone ringing. At 11:42 in the conversation, he puts the value here:

For a marketing agency, that advice becomes concrete through a brief: which buyers matter, what they need to understand, which channel fits the situation and what should happen after they respond. Our guide to ICP marketing develops the buyer research behind those decisions.

Consider a B2B service firm asking for ten posts and a landing page. Before quoting production, find out what prompted the request. Does the sales team lack useful conversations? Do visitors misunderstand the offer? Do enquiries arrive but receive no follow-up? These are different problems, even if the deliverables look similar.

This map is our application of Matt’s advisory point to marketing work. Use it to explain the scope in terms the buyer can assess. A proposal might say: “We will clarify the offer, build the campaign and enquiry path, and connect it to a follow-up process your team can use.” Then specify the deliverables, responsibilities and how progress will be measured.

Agree what counts as a relevant enquiry and which parts of the sales process the agency controls. An intended improvement belongs in the brief; a revenue promise without evidence does not.

Useful AI use cases for marketing agencies

At 16:13, Matt says clients are already asking how to implement AI in their businesses. Later, he gives a specific example: a coach in his community, Lucy, brings in specialists to deliver automations for clients. The work he describes includes database improvements and email client reactivation. Hear that example at 46:31.

The useful point is the delivery model. Lucy does not have to perform every technical task herself to assemble and deliver the service. Matt tells this as a community example, not a benchmark for what every agency can charge or achieve.

For a marketing agency, practical starting points include:

1. Research and briefing. Organise discovery notes and customer interviews, then check the themes against the source. The reviewer decides which buyer problem deserves attention.
2. Content and campaign production. Draft page structures, ad variations or email copy from an approved brief. Check facts, claims, tone and relevance before using the output.
3. Delivery and follow-up. Identify repetitive handovers, reporting tasks or reactivation work. Test the workflow with the client team and make ownership clear.

These examples are Do It Digital’s recommendations, rather than tasks Matt claims to have tested in our agency. Start with one workflow you can already judge well. Record the time, revisions and failures before and after the change, so a faster draft does not hide a longer checking process.

For search content, Google’s guidance on generative AI calls for accuracy, quality and relevance, including review of metadata. Producing more pages without adding value can fall foul of its scaled content policy. An agency’s content marketing process therefore needs research and review as well as a writing tool.

Review profit, traffic and conversions

At 25:52, Matt names the three areas he examines when improving a business: profit, traffic and conversions. He applies the same lens to a digital agency.

For the agency itself, traffic means the sales funnel and prospecting that bring potential clients in. Conversions mean turning that interest into won work. Keep those numbers separate from the impressions, clicks and leads in a client’s campaign report.

The measures in these cards are our suggested way to put that review into practice. Look at delivery cost by service, the source of qualified opportunities and the points where sales conversations stall. Use AI to help organise the information or challenge your assumptions, then check its conclusions against the business records.

Matt is explicit about growing sales without sacrificing profitability. Lower costs can give you room to improve delivery, strengthen margin or change pricing. They do not settle which choice is right. Review the scope and the buyer’s needs before changing the fee.

Build recurring value before adding more services

Matt recommends setting the business up for sale from the beginning, even if an exit is not the immediate goal. He points to reliable recurring revenue, clear figures and systems. Hear his reasoning from 24:45. That is an agency design principle, rather than a promised valuation.

A retainer needs a continuing reason to exist. A client may need monthly campaign decisions, content research, testing, reporting or support. The first build becoming easier does not establish the value of those ongoing tasks. Spell out what continues after launch and how the client can assess it.

Focus matters too. Matt describes a couple he coached who kept returning to the temptation of buying other kinds of digital assets. His advice was to stick with the agency that was already working. The lesson for an owner considering another AI service is to ask whether it strengthens the current offer or simply adds another delivery process to manage.

Review one agency offer this week

Take a recent proposal and compare it with the work the client actually needed:

1. Write the client problem in their words. Use the discovery notes, not a generated guess.
2. Mark the decisions the client needs help making. Clarify the audience, message, channel and enquiry path.
3. Pick one production task AI could assist with. Define what acceptable work looks like and who checks it.
4. Review delivery costs and the sales process. Decide which of profit, traffic or conversions needs attention first.
5. Rewrite the offer around that need. Keep deliverables and responsibilities explicit, then test the revised proposal with the next suitable buyer.

This exercise is our editorial application of the interview. It gives you a way to act on Matt’s central point without making the offer depend on a particular tool.

For help connecting the offer, website and distribution, explore Do It Digital’s marketing services or book a strategy call. For Matt’s wider discussion of agencies, digital assets and focus, watch the full episode.