our blog


We let the cat out of the bag,
revealing what’s going on around the agency.

let’s talk

A is for AI in Advertising

TLDR: We cracked how to use AI to make your paid media the most competitive online that it can be, so you deploy measurable ad spend that drives pipeline and growth. Read on for more info – there’s also a video – or save yourself some time and just scroll to the bottom and book a meeting with us, so you can start crushing 2021.


Let’s work smarter, not harder this year.


In 2020, we saw a large inflow of marketing budgets into digital marketing platforms – more than any year before. And as we know, more competition for clicks means higher click prices. If you take LinkedIn alone, for example, keywords that were $4 a click early the year eventually peaked at around $26 in 2020, due to the increased bid competition.

This means that we now need to be very focussed on how and where we spend clients’ budgets online to get the best return. A slight increase at the top of the funnel can have a huge impact on the cost per SQL lower down the funnel, and ultimately the ROI on ad spend.

This is where AI comes into play.

We started using AI in ad serving early 2020, and stepped it up during the year to help us answer three questions:

  • Where do we serve the ads, and to whom, to get best engagement?
  • What’s the best combination of creative + copy + call to action to get the best quality conversions?
  • How and what do we optimise against? CPM, CPC, CPA or something else? Something better? (Read on for something better.)

So let’s look at those in turn.

Where do we  serve  our ads, and to whom?

This first thing we can do is help build your perfect customer profile. To do this we can do some modelling,  but the strongest and most successful way has been to integrate your historical CRM data into the data management platform (DMP), where we can normalize and enrich your opportunity data for you, and then build a true representation of what your ideal customer profile looks like.

We can also extend your universe, and use our DMP to discover your ideal customer profile, using firmographic, technographic and demographic attributes. This pulls from over 1 billion global records to identify the accounts that are most likely to generate revenue for you!

We can help you identify the best-fit profile for accounts that generate the largest amount of revenue for your company in the shortest amount of time, as well as visualising the worst-fit profile – \the accounts you should typically avoid. Your sales team can then focus their efforts on where they have the best chance of winning deals.

We then use a personal-to-corporate identity graph process that’s based on email addresses, NOT IP (this is critical, as not many people are working in their corporate office at the moment). This will enable you to create custom audiences across Facebook, LinkedIn and programmatic display and video at a much higher match rate than cookie and IP-based technologies.

And once we have created the perfect audience, from your historical data and the global market, we can help you can advertise to that target audience with confidence, across multiple ad platforms – digitally surrounding your prospect audience.

Sounds good? There’s more.

We also can connected external data intent sources for specifically targeting those prospects who are showing buying signals across search and social channels. So we can pull live audience data from the likes of Bombora, G2 Crowd, The Aberdeen Group and HG Insights, as well as our own proprietary data that looks for social intent across LinkedIn and Twitter.

Now we have our audience, what’s the best combination of copy + creative + conversion?

This is where AI and machine learning can really help. Typically, you would build an ad, write a couple of lines of copy and send your audience to a landing page to convert. You’d run the campaign for period, and then review and look at how you could make improvements.

When you want to break down the audiences, and messages, you can end up with a large amount of very similar ads to build across multiple ad networks. This configuring, launching and then optimising can be laborious and time-consuming. It’s difficult for even the largest companies to run enough tests to learn meaningful information, and the data always lags. In a fast moving market, it’s hard to keep up.

This is where our AI comes in. We use it to automate a campaign workflow and generate hundreds of test experiments on the fly. So, for example:

  • Manual campaign: one piece of creative plus three lines of copy and one landing page. We set up three ads manually, and compare daily.
  • AI-driven campaign: we produce three different pieces of creative, with five lines of copy and two landing pages, plus LinkedIn lead gen forms, and our tech will build every combination on the fly, and help us optimise for you. This means 45 live campaigns in market for one asset, for less effort than our three manual ads (above).

More importantly, our platform automatically optimises these campaign experiments using the budget and campaign objectives we set, pinpointing the best performing combinations of audiences, ad creative and outcomes that engage your target accounts. We then use the platform to efficiently allocate your ad spend, based on which campaigns drive qualified pipeline and revenue (which we’ll come onto next time).

Finally, how and what do we optimise against? CPM, CPC, CPA? 

Well, at Creation Agency, we’re all about the best, not the easiest!

Measuring against CPM and impressions is like measuring the success of a small shop based on how many people walk past their front door. CPC is better – it’s a least a measure of people walking into the store – but it doesn’t mean they bought anything. So, for modern digital marketing with a demand objective, people typically optimise against CPA – the cost of getting a name (form fill). Which is better. But this has one huge flaw.

Measures tend to breed behaviour, and where you optimise against CPA, your objective is to get the lowest cost engagement from a prospect in your target audience. So you focus on content that your audience likes to read, usually about them and their market. And this content is much more top of funnel, such as industry-related eBooks and Gartner guides on key trends (“Read my eBook on the future of your industry,” “Read this Gartner guide on the key trends and what to look out for”). This type of content is attractive, so has a lower CPA – BUT – these are low-value opportunities, as you haven’t qualified a need, a project, or a budget. All you know is that your readers are interested in the future – and that’s a long way from them putting their hands in their pockets and buying your product or service.

The result is a load of engagement and marketing leads that never progress to sales opportunities – middle funnel bloat. It’s not the end of the world, as you can deploy various tactics to nurture these people, including marketing automation, personalised outreach and social selling (and we can help with all of these). But wouldn’t it be better to actually optimise to build your pipeline, rather than get you the cheapest names? This is where AI can really help.

Through connecting our ad management platform to your CRM system and marketing automation platform, we can optimise ads against pipeline generation on the fly. Our platform provides the visibility needed to understand which campaigns are driving the most qualified pipeline and revenue, and AI optimises your campaigns and reallocates your budget with these business outcomes in mind, so you don’t have to.

You’ll see an example in this video, which I actually did to update our global team on how we are deploying our AI platform. So if you’d like to learn more but you’re not quite ready for a meeting yet, this might help solidify some of your thinking.

Video Thumbnail

Fancy a chat? Keen to learn more?

Well let’s get some time in your diary. Drop me a note on the chatbot below (yes, that’s me on there), or book a meeting in my diary.