Marketing agencies are the most important adopters of AI; and it’s not surprising.
The IAB’s State of Data 2025 found twice as many agencies and publishers have fully scaled AI in comparison with brands. And Forrester’s 2024 report found 91% of U.S. promoting agencies are either using or exploring generative AI.
This isn’t surprising because AI is great at:
✅ Automating tedious tasks that include running an agency
✅ Leveling up client strategy and deliverables
✅ Saving time and resources
However, despite the momentum, most agencies still haven’t fully integrated AI into their operations. In fact, the identical IAB report found that only 34% of agencies have adopted AI at scale. The rest are still figuring it out.
Based on our talks with agency leaders, three key issues are holding teams back:
- Too many tools, not enough strategy. With over 30,000 AI tools in the marketplace, most agencies only use a fraction, and even then, not with a transparent strategic roadmap.
- Data chaos. Fragmented sources, security concerns, and inconsistent inputs make it hard to trust the outputs.
- Fear of breaking client trust. Many agencies are still unsure of tips on how to be transparent about AI use without making clients uncomfortable.
To assist you to overcome these challenges and implement AI successfully, we talked to agency leaders who’ve been through the method and compiled their top AI best practices in this text.
1. Be transparent together with your clients
This is the number one rule. If you’re using AI in any a part of your workflow, your clients must know.
“Many agencies don’t tell, advertise to, or are transparent about their use of AI in client-facing work. This causes a bunch of problems,” says Ryan Anderson, President at Markiserv, a creative agency. “We’ve had clients switch to us because their previous agencies used AI without disclosing it.”
These clients often feel misled after they discover that work they assumed was human-crafted had, in fact, been generated or supported by AI. This form of breach of trust is amazingly costly to repair.
“Clients could easily use AI themselves for free,” Ryan says, “but they hire agencies for the added strategy, creativity, and quality that human oversight provides.”
Beyond trust, compliance is one other major concern. “Many organizations have strict rules against using AI-produced work in their marketing campaigns or creative assets,” Ryan points out.
Given AI’s murky copyright history, failing to reveal AI usage can result in legal complications.
The solution is to be upfront. Explain where AI is used, similar to in research or ideation, and emphasize that human expertise still drives the ultimate deliverables. Transparency strengthens trust and keeps clients with you.
(*7*)2. Check your clients’ AI policy
Speaking of compliance, knowing your client’s AI policy upfront is non-negotiable.
As Robin Emiliani, Founder and CEO at Catalyst Marketing explains, “My biggest advice to agencies is to seek out out their AI tools, regulations, and protocols first.”
Different clients can have different mindsets and policies towards AI and never clarifying this in advance can result in awkward situations, churn, and even legal battles.
“Out of the library of 25 or 30 clients, we’ve probably 1 / 4 of them that say, ‘you cannot use AI with us’. And then one other segment that claims, ‘Okay, you need to use AI but only these specific tools, and you have got to VPN into our workspace to have the option to make use of it,’” she shares.
For example, one in all Robin’s clients approves the usage of Jasper, while one other one will approve Gemini. And it’s crucial to seek out out which tools your clients would give the green light before you begin using them.
The best solution to do that is to bake the AI-question into your onboarding process. For instance, you’ll be able to add these inquiries to your client onboarding document:
- Do you permit AI in our workflow?
- If yes, which tools are approved?
- Are there any security restrictions we’d like to follow?
Then, undergo this on an onboarding call to be sure each you and your client are aligned on AI. Robin advises, “Really get a transparent understanding from the clients, what you’re allowed to do and what you’re not allowed to do before going all in.”
3. Evaluate your processes before automating
AI can automate just about any process however it’s necessary to be sure that process is definitely vital and price automating first.
“A nasty process, when automated, just becomes a faster bad process,” says Peter Lewis, Founder & CEO at Strategic Pete, a marketing consultancy agency.
“I’ve seen agencies jump into AI for lead generation with no good system in place to follow up on the leads. The leads began flowing in, however the sales didn’t since the team had no good workflow to handle the load.”
Before automating anything, audit your processes. Ask yourself:
- Is this process worthwhile and really vital?
- What inefficiencies can we eliminate?
- Is there a better solution to achieve the identical goal?
Once you’ve identified that a particular process is vital and worthwhile, you’ll be able to got down to automate it with AI.
Marketing agency leaders are bombarded with AI product launches, latest features, and competing claims from tech giants.
The result’s what Jeff Su, Product Marketing Manager at Google, calls an “AI tools paralysis”—the overwhelming feeling of not knowing which tools are literally value using.
He warns that the majority AI tool discussions concentrate on hype slightly than real-world productivity. “New features and impressive benchmark scores mean nothing in the event that they don’t translate to real-world impact,” he explains.
Rather than chasing every latest AI tool, Jeff recommends constructing a minimum viable toolkit: a small, focused set of tools that address essential business needs.
Here’s tips on how to apply it:
- Identify your recurring challenges. What repetitive tasks in your agency take too long? Example: researching digital promoting trends.
- Find one tool that solves the issue. Jeff tested multiple AI research tools and located that Perplexity outperformed others in speed and accuracy.
- Master it before adding anything latest. He committed to using Perplexity exclusively until it became second nature.
This method ensures you don’t waste time switching between tools that overpromise but underdeliver.
5. Embed AI prompts into workflows
Even for those who’ve identified specific tools that may be great for your agency, you or your team might still not be using them consistently due to something called “death by prompts”.
This means you understand that a particular AI tool may help (e.g. Claude) but you’re not using it consistently since it looks like an excessive amount of effort to craft the best prompt each time. Jeff shares an example:
“I actually have this incredible prompt that transforms my writing into clear and concise copy. But there’s absolutely no way I’d be using this 20 times a day if I needed to type it out each time.”
His solution is to? Reduce friction. Here’s how:
- Use a text expander. Instead of retyping lengthy prompts, install a text expander tool like Raycast (Mac) or Beeftext (Windows). Jeff demonstrates how an easy shortcut (e.g., ::chptconcise) immediately loads his favorite prompt into ChatGPT.
- Embed prompts directly into workflows. Instead of keeping prompts in scattered notes, Jeff advises making a database of prompts in your favorite project management tool, like Notion.
Then, he advises adding a link to the relevant prompt directly where you would like it.
For example, if you have got a task in your Google calendar to “Write a webinar promotion email”, you’ll be able to add the link to the prompt that may assist you to write or edit the e-mail on to the duty.
This reduces the mental overload of remembering where to seek out your prompts and helps you get to results faster.
6. Centralize and clean your data
You can’t construct smart AI workflows on messy data. If your team is plugging AI into siloed spreadsheets, inconsistent naming conventions, or outdated reports, you’re just automating confusion.
“Data readiness and cleanliness are really necessary,” says Artūras. “If you consolidate cross-channel data and ask AI to generate summaries, your mind can be blown by how well AI can detect all the pieces.”
This is where a marketing intelligence platform like Whatagraph comes in. With it, you’ll be able to:
- Centralize your data from 55+ marketing platforms into one space
- Organize the info to make sure data quality. Fix naming inconsistencies, eliminate duplicates, and standardize formats.
- Blend data from different sources together to get a more holistic view of promoting performance
Once your data is clean and centralized, any insights AI gives you about that data is more relevant, accurate, and really actionable.
For instance, on Whatagraph, you’ll be able to ask our AI chatbot any questions on your marketing performance like:
- Which ad campaigns brought in the best conversions?
- How much money did we spend on X campaign last quarter?
- Which ads converted probably the most?
The AI gives you succinct answers based on the precise data source and timeframe you should analyze. And you’ll be able to fully trust the answers because again, it’s based on centralized and cleaned data.
Based on these accurate insights, you’ll be able to take the best actions like:
- Adjusting bids in your ads
- Turning campaigns on/off
- Optimizing creatives to spice up conversions
7. Build a habit of learning AI repeatedly
AI isn’t a one-time implementation—it’s an ongoing capability. Tools evolve quickly, latest use cases emerge consistently, and staying ahead requires greater than just initial adoption.
Agencies that make AI learning a part of their routine are those in a position to adapt, scale, and lead with confidence.
Jeff Su, Product Marketing Manager at Google, refers to this because the Impact Loop Strategy:
“Even for those who can only spare half-hour per week, taking motion and implementing something you’ve learned is infinitely more powerful than passively consuming something.”
Here’s how you’ll be able to embed regular AI learning into your operations:
- Choose one or two trusted newsletters like The Rundown AI, Marketing AI Institute to remain informed without getting overwhelmed.
- Set aside 30–60 minutes per week to check tools, refine prompts, or explore latest workflows, and document what works.
- Encourage internal knowledge sharing by making a central place to log AI use cases, prompts, and lessons learned.
The Bottom Line
AI isn’t any longer a buzzword for marketing agencies. It’s becoming a core a part of how work gets done.
But implementing AI successfully isn’t nearly picking the best tools. You need the best systems, clear communication, and a willingness to learn and adapt.
To recap, listed here are the AI best practices we’ve discussed in this text:
✅ Be transparent with clients about how and where AI is used
✅ Check your clients’ AI policies during onboarding to avoid compliance risks
✅ Evaluate internal processes before automating (only automate what works)
✅ Build a Minimum Viable AI Toolkit and master it first
✅ Embed AI prompts into existing workflows to scale back friction
✅ Use a marketing intelligence platform like Whatagraph to centralize, clean, and prepare your data for AI
✅ Make AI learning a weekly habit—test, share, and document what works
—
Find more practical strategies and AI tool recommendations from agency leaders in this AI playbook. Download our AI Playbook for Agencies in 2025.
Read the complete article here
