AI-optimised B2B marketing

Do AI-based campaign models make B2B as easy as B2C?

In just over 13 years in the field of social media marketing, I have had the opportunity to help with a wide variety of client campaigns from B2C and B2B and have always noticed major differences – all the more reason a statement made by two speakers at the All Social Marketing Conference in Munich in mid-March 2023 got me thinking. The conferences held there outlined what effective social advertising on Meta and other social media platforms should currently look like.

Both presentations share a common theme. Creatives are essential for good advertising on social media channels. At the same time, targeting is becoming massively less important. The main driver for this development is ultimately the progress that has been made in regard to running campaigns, which can now be done more and more using AI models. With the possibilities of machine learning and the gigantic data sets that Meta, LinkedIn and the like have at their disposal, the old algorithms are gradually being replaced. Campaigns now optimise themselves as they are being run via continuously learning AI.

How exactly does good advertising in social media work on the basis of AI models?

Openness, trial and error and having a wealth of variants are the key considerations here. The targeting criteria are completely open. Meta’s Advantage+ shopping campaigns, for example, only include the country in which they are being run as the target group – there is no age or interest-based targeting and also no fixed target group for retargeting, such as previous buyers or website visitors. The AI utilises data that is already available via pixels or server-to-server tracking. This data is then also important in the further course of the campaign. The AI ‘understands’ which user group is interested in the products based on the users’ reactions to the advertising media and the behaviour on the target pages. This information can be applied accordingly to the existing group of people on the social networks in order to find exactly the right users.

The creatives have a very important role to play here. They work as a campaign’s testing mechanism. Each advertising spot consists of three components: the hook, the core and the CTA (call to action). An attention-grabbing introduction leads into the core message, which ultimately ends with a clear call to action.

In a campaign, as many creatives are used as possible, all of which contain different hooks, cores and CTAs. After an initial test phase, successful components are carried forward and nonperforming components are removed or reworked. This ensures that the creatives are gradually adapted to the customers’ demands in order to work as efficiently as possible.

So far, so good. I can say from my own experience that forgoing targeting works. Even if it feels uncomfortable at first to relinquish control of the targeting efforts and put up with ad waste until things have been optimised, the results of AI optimisation in Advantage+ shopping campaigns speak for themselves.

For one of our e-commerce clients, we spent a long time using a campaign that utilised a proven interest targeting mechanism. In addition to generating reach and attracting new website visitors, this campaign also largely paid for itself. The switch to an Advantage+ shopping campaign led to the reach figures remaining relatively identical, but the number of website visitors and the ROAS in particular showed significant improvements.

But what about in B2B marketing?

At the moment, we are dealing with a B2C e-commerce campaign with a large potential target group, simple products and short product cycles. But how well does this ‘new’ approach work when we venture into B2B marketing? I of course posed this question to the speakers at the ASMC as well. The answer was identical in both cases: ‘it can be used for B2B too!’

But unfortunately, it is not as simple as it sounds. Essentially, the structure outlined above can be applied to B2B campaigns as well, but when we do so, we start running into obstacles rather quickly.

1. The actual target group potential

The AI needs usable information more than anything else, which in our case equates to a certain number of clicks, leads or purchases. In B2C, there is always a large number of potential interested parties, even for very specific products. In B2B, we are usually very specific. Things here have to be done at the decision-maker level by people who have sufficient experience and corresponding jobs or knowledge. The luxury of having relevant target groups consisting of over 50,000 people or more is often not a given. This in turn means that the AI will receive less information to optimise the campaign. And even if the optimisation is successful, the AI will take longer to do it and will have to eliminate far more ad waste. This has a direct impact on costs, and the lead itself becomes significantly more expensive.

There is a compromise to be had here. The target group is not left completely open, but rather is at least roughly defined in advance; for example, only the target audience’s seniorities and their industries constitute well suited criteria – provided this information is available as targeting criteria. With LinkedIn, this is no problem. With Meta and other social media platforms, there are no targeting options for this. It would still be possible to work with completely open targeting criteria on those channels. However, experience shows that only a few customers are willing to bear the higher costs of eliminating ad waste.

This also makes sense when it comes to significantly longer product cycles. If a product only comes back into demand every few years, the already small potential target group becomes even smaller. Lead campaigns then prove futile not because there is a lack of interest in general, but rather because there is a lack of interest at the moment. These phases are ideal for branding, but they are also the most difficult for advertisers, as the hard KPIs are missing for a longer period of time. This inevitably leads to either no branding being done or work having to be done as efficiently as possible, leaving no room for experimentation or AI optimisation.

2. Complexity/specificity of the products

The very narrowly defined target groups are an indication of another trait of B2B campaigns. In many cases, the products, solutions or services are precisely tailored to a specific use case and at the same time, are highly complex and difficult for laypersons to understand. This means that we are more limited when it comes to making creatives than is the case with B2C campaigns. The possibilities regarding conducting extensive testing using creatives as the basis are minimised from the outset. This does not necessarily mean that taking new approaches or using new methods is out of the question here. Unboxing an industrial machine that weighs tons can certainly have its own charm and attract attention. In general, however, we will not start with a wide range of creatives but rather with a few core messages that are only slightly nuanced.

3. Being pragmatic versus being dramatic

For B2C campaigns, generating emotions, and thus branding, is an essential part of standing out from the competition. Then the focus is no longer on the product but rather on the emotions associated with it. These emotions are then often embedded in cinematic short films or more authentic snapshots of users or influencers.

In B2B, branding is done as well, but the values conveyed are much more pragmatic and thus, to some extent, also less exciting than the emotional B2C counterparts. This can be both an advantage and a disadvantage. If a B2B brand knows the essential pragmatic reasons for its customers’ purchasing decisions, then the hook, core and CTA stem from them in logical fashion and the customer is saved the expense of producing variants. At the same time, it is precisely this freedom to use emotions or an appealing brand essence to stand out from the competition that is missing.

Conclusion

So it is all bad and we should carry on like we have been? Yes and no. The approach is in essence promising if the obstacles to it are accounted for. Sooner or later, AI models will also be more successful in B2B than manually created and optimised campaigns. But it will be important to feed the AI well and prepare it for its task. In B2C, the high number of cases make it so that not much preparation is needed. With this type of marketing, the AI is able to be optimised based on the reactions and testing can be done at the creative level. In B2B, the framework conditions have to be specified more clearly, much like in current AI applications, the outputs of which depend massively on the initial prompts – that is, the specific work orders sent to the AI.

If you want to delve a little deeper into this topic and are asking yourself how exactly you need to feed the AI in order to implement successful campaigns, we would be happy to advise you.

Get in touch now

You will find more exciting topics from the adesso world in our latest blog posts.