From Traditional Buying to Agent-to-Agent Technology

The digital advertising market currently operates primarily through two models: Direct IO and programmatic.

• Direct IO offers transparency and control but is less scalable and requires manual processes.
• Programmatic offers scale and speed but operates through multiple technological layers. As a result, a significant portion of the budget is allocated to infrastructure rather than media.

The agentic model organizes this differently. Instead of an auction-based model, the demand and supply sides communicate directly with each other through structured data exchange. Negotiation and activation take place automatically based on predefined rules.

Working Media and Operational Efficiency

Within this campaign, the share of Working Media increased from approximately 70% to 94%. Concretely, this means that instead of around €0.30 per euro being allocated to technological intermediaries, approximately €0.06 is spent on infrastructure. About €0.24 per euro is therefore redirected toward actual media impact.

Additionally, a campaign can go live within approximately two hours, from briefing to activation. Whereas traditional buying processes involve multiple intermediaries and manual steps, this process is fully AI-driven.

Continuous Optimization

Importantly, optimization fundamentally changes. Because negotiation, activation, and adjustment take place continuously via agents, optimization can occur in real time based on context, performance, and price. This does not happen at fixed evaluation moments but continuously and at scale.

As a result, optimization can take place more deeply, faster, and more frequently than is possible within a traditional human-managed model. The operation shifts from campaign management to agent management.

How Does It Work in Practice?

On the publisher side, Planet Nine Media configures the Sales Agent. This agent makes inventory available, including contextual information, quality parameters, and pricing agreements.

The Buying Agent receives the briefing from BridgeFund, including target audience definition, contextual preferences, and performance criteria. Through AdCP, this is translated into a structured request. The Sales Agent evaluates this request and responds with suitable inventory, including contextual data and pricing agreements.

If necessary, the agents submit counterproposals based on predefined negotiation rules. Once agreement is reached, the media buy is automatically created and activated.

Within this model, no traditional DSPs or SSPs are used, and no open auction takes place.

Kees de Boekhouder is the go-to accounting firm for freelancers. With ambitious growth goals, the brand decided to significantly scale up its digital marketing activities. More channels, more budget—but also more complexity. To enable data-driven decision-making with real business impact, Kees de Boekhouder turned to Draft Digital.

The problem: Growth without insight leads to noise

To accelerate growth, heavy investments were made in a broader media mix: from SEA (PMAX and non-branded) to display and social advertising on TikTok and Meta. While this increased reach and reduced the brand’s dependence on word-of-mouth, lead quality declined at the same time.

Two challenges emerged:

1. No insight into which ads were responsible for the decline in lead quality.
   
2. Push channels like display and social generated few direct leads, yet claimed a large share of conversions—without GA4 being able to measure their real impact.

In short: the true value of each channel remained invisible.

The solution: AI-driven attribution via Billy Grace

Together with Draft Digital, an innovative solution was implemented: an integration between the HubSpot CRM system and the Billy Grace AI platform. This provided full visibility into the customer journey, including:

• Linking marketing data and sales data
  Through dynamic parameters, every touchpoint was captured. When someone schedules a call, this data is linked to HubSpot—and updated again once the lead becomes a customer.
  
• Session stitching & lead journey tracking
  Billy Grace recognizes unique customer journeys and uses data-science models to assign value to each channel based on actual impact.
  
• UMM model with impression attribution
  Channels focused on qualitative reach and fewer direct website clicks (such as display or TikTok) were finally valued through marketing mix modeling—something that GA4 cannot do.

“The beauty of this case is that we used smart technology to remove the clutter from performance marketing. No more assumptions—only action based on what truly works.”
— Aliks Roling, Digital Marketing Consultant at Draft Digital

The result: deep insights, better decisions, higher conversions

The integration delivered surprising insights and direct control over lead quality.

KPI results at a glance

• TikTok generated 400% more leads than GA4 previously attributed.
  
• 34% of conversions previously marked as “Direct” were now correctly assigned to other channels.
  
• Meta generated 40% more leads than initially measured.
  
• Branded search saw a 10% decrease in attributed value, while non-branded search decreased by 20%.
  

Lead quality by channel

• Direct traffic: index 133
  
• Branded channels: index 108
  
• Push channels: average index 71
  
• TikTok: a standout performer with an index 217

One of the most important insights from the connection between marketing and sales data is the impact of branding on conversion rates. Where the focus previously lay mainly on direct conversions from performance channels, the new data shows that the strength of a strong brand directly increases the effectiveness of all marketing efforts.

The more specifically someone searches for “Kees de Boekhouder,” the higher the likelihood that they will actually become a customer. Branded search terms and direct traffic—both indicators of brand awareness and brand preference—consistently generate higher conversion rates than push-oriented channels.

This confirms the strategic importance of branding: creating awareness, building trust, and staying top-of-mind ultimately lead to better performance across every channel in the funnel. Branding is therefore not a cost center, but a lever for performance.

A deep dive into the SEA keywords also showed that certain high-intent keywords—despite having a relatively high cost per lead—perform strongly on the KPI “Cost per Client.” This indicates that we need to manage the flow of HubSpot data to the marketing channels more intelligently. 

“With this data-driven approach, we’ve taken the first clear steps toward understanding our growth opportunities. We can now better see what does and doesn’t work—not just per channel, but also in the quality and conversion of leads. That gives us a strong foundation to optimize even more precisely in the coming period. I look forward to taking those next steps together.”
— Daniël Muller, Head of Growth at Kees de Boekhouder

Next steps: turning insights into automated growth

Now that the impact of each channel has been clearly mapped out, the focus shifts to translating these insights into a more efficient, data-driven marketing approach. In the coming period, Kees de Boekhouder will focus on three pillars to drive further growth:

1. Smart budget allocation based on real performance

Thanks to the connection between marketing data and sales data, optimization no longer focuses solely on cost per lead (CPL), but on cost per client — the KPI that truly matters. With the help of automation, budgets are dynamically adjusted:

• Channels with a low CPL and high conversion rate automatically receive more budget.
  
• Channels with lower lead quality or higher costs are scaled down or optimized.
  
• Campaigns are optimized based on cost per client, which allows for a stronger focus on high-intent keywords. 

This performance-driven approach ensures that every marketing euro delivers maximum results and contributes directly to growth ambitions.

2. Larger investments in branding 

The analyses show that brand awareness has a direct impact on conversions. People actively searching for Kees de Boekhouder convert significantly better than cold traffic. Therefore, intentional investments are being made in branding campaigns focused on visibility, trust, and top-of-mind presence within the target audience.

The goal: bring more people into the funnel who are already familiar with the brand, so that performance channels also deliver stronger results.

3. Targeted growth through social content

The remarkably high lead-to-client conversion rate on TikTok (52%) demonstrates that this channel holds substantial untapped potential. In the coming period, the focus will be on a strategic content approach dedicated to TikTok and other social channels.

A new standard for programmatic advertising

The programmatic landscape is becoming increasingly complex. With the rise of AI-generated content and the growing number of Made-for-Advertising sites, maintaining control over advertising environments is more challenging than ever. 

To tackle that challenge, Draft Digital became the first agency in the Netherlands to introduce Brand Standards — an AI-driven solution from Scope3 that evaluates content based on nuance and context, rather than relying on simple keyword blocking.

Instead of using a standard brand safety blocklist, an LLM agent (large language model) determines in real time whether a webpage is suitable to place a brand alongside. It evaluates context, brand values, fraud prevention, CO₂ emissions, and attention metrics.

“We are proud that Draft Digital was the first Dutch partner to implement Brand Standards. It shows how innovation and sustainability can go hand in hand.”
— Kate Moeller, Director  of Partnerships Europe, Scope3

Strategy, implementation and guidance by Draft Digital

Draft Digital oversaw the entire process: from the strategic development of the prompt to the technical implementation and A/B testing to make the impact measurable.
Together with BridgeFund, a Brand Standards prompt was developed that precisely defined which content is ‘aligned’ and ‘unaligned’ with the brand. The following, among others, was excluded:

• content related to violence, fraud, war, or sexually explicit material
  
• economic crises or negative policies affecting vulnerable groups
  
• puzzle, gaming, and radio sites
• In addition, preference was given to high-quality content in specific languages, including Dutch, Turkish, Arabic, French, German, Italian, Spanish, and English.

The prompt was then tested on 183,000 Dutch articles. With feedback from the team, the AI agent was continuously refined— a strong example of the ‘Human in the Loop’ process that Draft Digital prefers in its AI-driven approach.

Measurable impact on brand metrics and performancee

The pilot, carried out by Draft Digital in collaboration with BridgeFund, delivered compelling results:

• +2% higher viewability (from 84% to 86%)
  
• –30% lower CPA
  
• +3% more ad exposure >10 seconds (from 36.87% to 39.61%)
  
• –50% decrease in Cost per Lead in an A/B test between two line items

The insights confirm that brand safety does not have to come at the expense of performance. By using context more intelligently, greater impact was achieved with less advertising waste.

“With this approach, we were able to define much more precisely where our ads did and did not belong. That gave us greater control and noticeably better results.”
— Mark Hoekx, Performance Marketeer, BridgeFund

The next step: moving toward agentic advertising

Following the success of the pilot, Draft Digital is now working on the structural integration of Brand Standards across its full programmatic approach. BridgeFund is one of the first partners to also apply this structurally within its campaigns.
The next step? Testing with Brand Stories — an advanced LLM solution that personalizes ads in real time based on audience segmentation.

“Agentic advertising is the future. With this technology, brands like BridgeFund can advertise at scale, internationally, and with meaningful relevance. We’re excited for what comes next.”
— Aliks Roling, Digital Marketing Consultant, Draft Digital

The Challenge

As a lottery brand, Lot of Happiness aims to reach a broad audience with inspiring campaigns, but in a responsible way. On Meta, large-scale reach campaigns were purchased without insight into the content alongside which the ads were appearing. While Meta allows targeting by age group, a feature already in use, there was still a potential mismatch when it came to the context in which the ads were shown. 

The Solution

To address this challenge, Lot of Happiness partnered with Draft Digital to use Scope3’s innovative Brand Standards solution. This AI-driven system enables brands to create their own brand agent, fully tailored to their values and guidelines. Scope3 has access to all content IDs within Meta and uses Large Language Models to analyze whether the content aligns with the brand profile. Any content that does not match is classified and excluded from the campaign within 3–5 hours of being posted on Meta.

What makes this solution unique:

• Personal brand profiles: The AI is trained with input from the brand, ensuring maximum relevance.
• Sustainable impact: In addition to the Brand Agent, Scope3 also takes CO₂ emissions into account across different times, placements, and creative types. 
• Broad focus: In addition to brand safety, Scope3 also looks at attention KPIs to maximize impact.
• Human-in-the-loop process to validate content through human review, allowing the AI model to be trained even more effectively. 

Results

The initial results are promising:

• 25% of content IDs were blocked during the test period. For Lot of Happiness, this shows the importance of adding this classification method, as there is still a large amount of Meta content that does not align with the brand’s values. 
• No negative impact on performance — reach, CPM, viewability, and exposure time remained stable.
• Better brand protection by preventing impressions alongside unwanted or sensitive content.
• Greater relevance for the end user through a more precise alignment with the brand profile.

“The context in which people see your ads is – especially for us as a lottery – extremely important. With Scope3, we can be sure that our ads won’t appear alongside content that doesn’t align with our values, such as content related to children or addictions.

“On the other hand, we want to be able to adapt the content of our ads to the context. For example, in an article about charities, we like to show images of our annual donations, while in an article about long-distance travel, we prefer to highlight the chance to win big. We’re really looking forward to experimenting with this together with Scope3 and Draft.”
— Leo Nijs, Online Marketer at Lot of Happiness

“As an agency, we believe it’s important to stay ahead of the curve. The Scope3 solution is not only scalable and easy to implement, but it also delivers immediate, visible results.”
— Lars Postmus, Founder of Draft Digital

Next steps

The positive results mark the starting point for a broader deployment of Scope3 within Lot of Happiness’s media mix. In the coming period, Brand Standards will be tested within programmatic campaigns to gain more control over context and brand safety there as well. In addition, the current brand profile will be further refined to improve performance. 

At the same time, we will experiment with the use of Scope3 Brand Stories, in which personas take center stage. This combination of brand safety and relevant storytelling not only protects against unwanted associations but also creates opportunities for improved performance. The ambition is to make AI-driven content matching a permanent part of Lot of Happiness’s always-on media landscape.

Kees de Boekhouder has been a trusted name for creative entrepreneurs for years, helping them manage their administration smartly. Although the brand has a strong presence in search, social advertising had never been structurally used as a conversion‑driven channel. That changed.

The Problem

Social channels such as Meta were previously used mainly on a campaign basis, without a clear distinction between the stages of the funnel. Objectives like reach, engagement, and conversion overlapped, as did the creatives and copy. This led to fragmentation and limited the effectiveness of social as a conversion channel. 

In addition, we found that the best‑performing ads were not always being used, even though the creative accounted for between 47% and 70% of the impact (various sources including Nielsen Catalina, 2017, Think with Google – ABCD of Effective Creative, and Meta & Analytics Partners – The Role of Creative) contributes to the overall sales result. 

The Solution

We restructured the entire social channel on Meta according to the See‑Think‑Do model, with content, copy, and campaign objectives carefully aligned for each phase. The key innovations:

• Conversion‑driven advertising: Do‑phase campaigns were given a clear lead‑generation objective instead of generic engagement or reach goals. In addition, standard tracking settings such as server‑side tracking were implemented so Meta could optimize more effectively.
  
• Smarter use of the algorithm: All placements were combined into a single ad set so that the algorithm could optimally determine where the content performed best.
  
• Creative iteration:Finally, it was crucial to always show the right creative to the right person. For this reason, the creative iteration program was introduced. In this program, different types of content were tested in a dedicated test campaign, and the best‑performing ads (winners) were moved to the main campaigns. These creative formats were then further integrated to prevent ad fatigue. 

This approach was innovative because, for the first time, creative effectiveness was tested broadly and in an algorithm‑driven way. The focus was not on micro‑segmentation, but on finding winning formats that could be scaled across the entire funnel.

Example: Creative Testing Program 

Results

The new approach had an immediate impact:

• Leads increased by +464
• Costs per lead decreased by -74%
• Reach grew by +185%
• CPM decreased by -36%

“The new structure and focus on creative effectiveness have made Meta a serious conversion channel for us. The difference from before is huge — we are now getting real results from social.”
Daniël Muller — Head of Growth Kees de Boekhouder

Next steps

The next phase will focus on further developing creative testing so that even more scalable formats can be deployed. Currently, a relatively large amount of “user‑generated content” is already being used, but the ambition is to scale this up even further, potentially using AI.

In addition, a link has been made between HubSpot and our attribution platform BillyGrace, so that in the future we can also optimize for actual customers. This will help us avoid irrelevant hooks that generate many leads but no sales. 

Artificial Intelligence (AI) has made significant advances across industries, and digital audio is no exception. One of the biggest breakthroughs is AI’s ability to clone voices and generate text‑to‑speech audio. This innovation is reshaping how content is created, offering unprecedented efficiency and precision. A great example is Aldi, which replaced a voice actor with AI in its commercials: https://nos.nl/artikel/2516494-aldi-vervangt-stemacteur-door-ai-in-reclames-zullen-meer-bedrijven-volgen

Voice Cloning and Text‑to‑Speech

Traditionally, producing audio for commercials required studio sessions with voice actors, sound engineers, and post‑production, a process that was both time‑consuming and costly, with very little room for flexibility. Today, AI‑driven text‑to‑speech technologies have completely changed the game. Advertisers can now create audio instantly, without the need for a studio. AI can even clone voices, allowing for fully personalized audio content. Tools like ElevenLabs make it easy to upload your own voice, something I (Lars Postmus) have tried myself.

A Note of Caution

Ondanks dat de voordelen van het klonen van stemmen gigantisch zijn, is het wel goed om uit te kijken naar mogelijke fraude. Doordat stemmen dusdanig goed nagemaakt worden kan daar ook misbruik van gemaakt worden. In dit artikel van NU.nl While the benefits of voice cloning are enormous, it’s worth being vigilant about its potential for fraud. As AI voice generation becomes more sophisticated, it can also be misused. A recent example was reported by NU.nl, where an employee of a Hong Kong company was deceived into transferring €24 million to criminals. The employee thought he was on a video call with the company’s financial director, but it turned out to be a virtual person with a cloned voice.

Speed and Relevance: Making the Most of the Moment

One of the biggest advantages of AI in digital audio advertising is its ability to respond quickly and stay relevant. For example, during major events like the UEFA European Championship (EURO), advertisers can now produce and launch audio ads in real‑time, capturing the excitement and energy of the moment. This level of responsiveness was virtually impossible with traditional methods, where new content often required long lead times. AI allows advertisers to remain timely and capture their audience’s attention precisely when it matters. This approach is ideal for brands like HEMA, KLM, and Bol.com, which have long leveraged social media for timely messaging. Now, text‑to‑speech allows this to extend seamlessly across radio, video, and audio platforms.

From One‑to‑Many to One‑to‑One

Another major shift AI brings to digital audio is the move from one‑to‑many communication towards highly personalized, one‑to‑one messaging. Whereas traditional audio ads broadcast to a broad audience regardless of individual interests, AI makes it possible to analyze data and tailor ads to specific demographics, interests, and even individual listeners. This precision ensures that the right message reaches the right person at the right moment, maximizing campaign effectiveness and return on investment.

Location‑Based Dynamic Audio

Location‑Based Promotions: Vascobello Coffee can use dynamic audio to highlight daily specials at specific locations. For example, an ad playing in Amsterdam could say: “Enjoy a freshly brewed latte today at our Museumplein location!”

Time‑Based Dynamic Audio

Morning and Evening Messages: A fitness app can use dynamic audio that changes based on the time of day, for example, “Good morning! Start your day with a refreshing yoga session,” in the morning, and “Good evening! Unwind with a relaxing meditation,” in the evening.

Weather‑Based Dynamic Audio

Seasonal Products: A fashion brand can adapt its audio ads based on the current weather, for example: “It’s cold outside! Stay warm with our new winter collection,” on a chilly day.

Enhancing DCO Across Channels

In addition to creating digital audio ads, text‑to‑speech technology can significantly enhance dynamic visual ads on social media platforms. Traditionally, dynamic ads have focused mainly on visuals, adapting images and videos to capture attention. However, by integrating text‑to‑speech, advertisers can also add dynamic audio to their content, creating a more immersive and captivating experience. This is especially valuable for platforms like Reels and TikTok, where sound plays a pivotal role in grabbing attention. With text‑to‑speech, brands can quickly produce relevant audio that complements their visuals, allowing for seamless, real‑time communication with their audience.

Summary

At Draft Digital, we believe in using technological innovation responsibly and with integrity. Advances in AI‑driven digital audio align with our core values and offer growth opportunities for both advertisers and listeners. By embracing these innovations, we aim to deliver higher performance in every campaign and provide high‑quality, engaging, and relevant audio content tailored to listeners’ preferences. Currently, Draft Digital is fully immersed in the pilot phase, working to create broadcast‑quality ads. Want to learn more about audio and AI‑generated advertisements? Feel free to get in touch via this (link).

Phases & Metrics

There are various metrics and phases you can build a custom algorithm around, from sales metrics such as conversions or Q‑Visits (a quality click) to other metrics tailored to specific time windows, post‑click, or view‑through attribution. By assigning a value to each conversion action within the custom algorithm, you can guide it to optimize exactly as you intend.

It’s also possible to evaluate clicks and assign a value to them. However, we generally advise against this, as a click with no further value doesn’t tell you much.

Finally, you have a range of awareness metrics you can optimize for, one of the biggest improvements over the old custom models. In the next section, we’ll take a closer look at the different metrics and highlight the most important ones.

Viewable Impressions

A simple example of value assignment is the viewability metric. By focusing on viewable impressions, you can guide the algorithm. This approach is fairly similar to the standard viewable algorithm, and it only becomes truly valuable when combined with other metrics.

Time in View

The next variant is far more interesting. By assigning a value to the number of seconds a banner is in view, you can optimize for longer time‑in‑view, a feature not available in the standard algorithms. At Draft Digital, we recommend setting a maximum value, as some sites naturally have long view times that aren’t necessarily high‑quality. A good example is Nederland.FM, an online radio station that often plays in the background. While DV360 will register the banner as in‑view for a long time, this doesn’t always mean it delivers valuable attention.

Completed Views

It’s also possible to incorporate completed views, although this is already available within DV360’s standard algorithms. However, combining this metric can be especially valuable when it comes to video campaigns.

Audibility

This metric is often overlooked in video campaigns, yet it’s incredibly important. Without sound, a video can’t be truly heard, and most advertisers create their videos with audio in mind. This issue is especially common with outstream placements, where sound is typically turned off. Since outstream ads tend to be cheaper than instream placements, standard algorithms often allocate a large portion of the budget to these slots. By adding audibility as a metric and assigning a value to it within your algorithm, you can deliver more impactful results.

Screen Size

This metric is also a strong addition to video campaigns. In programmatic advertising, there’s a lot of inventory that doesn’t deliver quality results, and it’s common to see “instream” video positions sold where the consumer’s attention doesn’t actually land. For example, placements that are technically in‑view, but positioned where the viewer is unlikely to look.

Fold Position

In programmatic advertising, performance can vary greatly depending on whether a banner is placed above or below the fold. Until now, this was only available as a reporting metric, allowing you to include or exclude certain positions in campaigns. By adding this to a custom bidding algorithm, you can gain more control and optimize specifically for above‑the‑fold placements.

Device

Especially with video campaigns, we see that the device used can have a significant impact. Many advertisers now adopt a Connected TV strategy. By adding device targeting to the custom bidding algorithm, you can optimize more precisely for the devices that matter most.

Scoring vs. Multiplier

The advantage of the custom algorithm tool is that it gives you different ways to assign value to impressions. For the most common metrics, you can only apply a score, but for fold positions, you can also add a multiplier. This is especially important for advertisers who want to secure top‑of‑page positions. By applying a multiplier, you can exert more precise control over these high‑value placements.

Usability

The DV360 Custom Bidding model is a highly practical way to optimize for awareness KPIs. Through experiments in DV360, it’s possible to test this approach effectively against standard models. However, you’re still limited to the metrics and KPIs available within DV360 itself. To truly optimize for attention value, third‑party tools such as DoubleVerify, MOAT or Lumen are required.

Summary

The use of custom algorithms within DV360 provides a valuable and practical way to improve the effectiveness of display, video, and native campaigns, especially now that traditional tracking methods are under pressure. By integrating metrics such as viewability, time in view, audibility, and fold position, advertisers can optimize more precisely for the quality of their impressions. While this approach delivers significant benefits, it’s important to acknowledge the limitations of DV360 and, when needed, complement it with third‑party tools such as DoubleVerify, MOAT, or Lumen to enable more in‑depth attention measurement. At Draft Digital, we believe in the power of tailored solutions and continuous optimization, helping us grow together and deliver outstanding results. If you’d like to learn more, feel free to get in touch with us via this (link).

Data has become a valuable asset for businesses. Collecting and analyzing data can provide valuable insights that lead to improved performance and decision-making. This is especially true in digital marketing and programmatic advertising, where data has been a crucial component of success. Historically, third-party cookies were used to track users across the internet, allowing advertisers to serve the right content at the right time. However, third-party cookies are set to disappear permanently in the near future, and their reliability is already far from optimal. Many browsers have already begun to largely block them.

For this reason, companies are seeking solutions to continue personalized advertising. In many cases, server-side tagging or customer match are used. However, these solutions are typically accessible to large tech companies like Google and Meta, making collaboration between Dutch publishers such as DPG Media or Marktplaats more challenging. To address this, various parties have explored ways to work together with these publishers, and the solution appears to be Clean Rooms.

What Are Clean Rooms?

Data bunkers are an emerging technology that enables secure data sharing and aggregation without exchanging the underlying data itself. Through Clean Rooms, companies can gain valuable insights and make data-driven decisions without compromising user privacy. Here are some ways data bunkers operate. 

Data Aggregation in Clean Rooms

Data bunkers enable the integration of data from multiple sources without sharing the underlying data itself. Instead, aggregated and anonymized statistics and patterns are exchanged, allowing companies to gain valuable insights into trends and behaviors without exposing individual information. Moreover, only Personally Identifiable Information (PII) is used, eliminating dependence on third‑ or first‑party cookies. 

Cross-Reference Analysis via a Clean Room

With data bunkers, companies can compare and analyze data across different datasets without sharing the underlying data itself. This enables businesses to gain insights into customer behavior, market trends, and other valuable information without exposing sensitive data to risk. For example, you could compare your own customer base with that of a partner, allowing you to strategically target potential new customers. 

Data Security in Clean Rooms

Clean Rooms use advanced security protocols to ensure the confidentiality and integrity of data. This includes encryption, strict access controls, and monitoring of data access to guarantee that only authorized users can access the information.

Data bunkers like Infosum (www.infosum.com) & EDM (www.edm.nl) play an important role in finding a balance between privacy and performance. Data bunkers provide a secure environment to store and process data, while enabling companies to gain valuable insights without having to share the underlying data itself. By using these technologies, businesses can safeguard privacy while still benefiting from data analysis and insights. 

So, there is a way to securely share data between different parties, but how does this work in practice, and how could it be used? 

Example of How to Use a Clean Room

In this example, we’re working with the company Draft, which wants to purchase ads on NU.nl. 

• Both Draft and NU.nl upload personal data into their own clean rooms.
  • Draft only uploads the data they want to target into the clean room. For example, people who have previously made a purchase.
  • NU.nl uploads all their data into the clean room to increase the chances of finding matches. Additionally, NU.nl must provide an ID that is used within their advertising platforms.
• It’s important to regularly coordinate how the data is structured. For example, if matching is based on postal codes, it’s crucial that the data is formatted consistently.
• Once the data is in the data bunkers, it is matched and sent back, including IDs, to NU.nl’s advertising platform.
• Draft can now purchase inventory based on these IDs through NU.nl and re-engage the target audience.

We can conclude that data bunkers offer a solution to use your own data on external platforms, but what exactly can you do with them?

How Can You Use Clean Rooms to Improve Digital Performance?

Targeting

In addition to being able to find users again, it can also be used to exclude certain customers. Look-alike audiences can also be leveraged. Since the external platform can see what these users do within their own environment, it can search for people who resemble the initial group. This allows you to deliver more relevant ads. 

Data Enrichment

By matching data with external platforms, it’s also possible to learn from this data. If you share data with Marktplaats, for example, you can discover which categories your target audience frequently browses or uses. If your audience is interested in interior design, this insight can serve as a valuable hook for other marketing efforts.

Channels

In principle, data bunkers can be utilized across all channels where CRM data is matched. Currently, this is mostly applied in display and video advertising in collaboration with partners such as DPG Media, Mediahuis, and Marktplaats. However, it should also be feasible on more traditional channels like TV. TV platforms use set-top boxes that store personal data. Although addressable advertising on TV is not yet common practice in the Netherlands, it is expected to become possible in the near future: https://www.adformatie.nl/media/adverteerders-kunnen-tv-commercials-bij-talpa-en-rtl-richten

From then on, more and more opportunities for personalization on TV will emerge. In Belgium, this is already possible. 

Retail Media and Clean Rooms

Naast reguliere advertising zien we ook een trend op het gebied van retail media. Grote partijen zoals Albert Heijn, Jumbo, MediaMarkt en Amazon zijn hiermee aan het experimenteren in Nederland. Doordat deze retailers over zeer veel first party data beschikken kan een databunker ook de connectie zijn tussen de data van retailpartner en adverteerder. Hierdoor zijn ze niet afhankelijk van reguliere customer match oplossing.

Conclusion

Alongside traditional advertising, we’re also seeing a growing trend in retail media. Major players like Albert Heijn, Jumbo, MediaMarkt, and Amazon are experimenting with this in the Netherlands. Since these retailers possess extensive first-party data, a data bunker can serve as the bridge between the retail partner’s and the advertiser’s data. This means they aren’t reliant on standard customer match solutions.

Thanks to data bunkers, companies can collaborate with external partners, such as publishers, to deliver personalized ads to the right audience without directly sharing first-party data. This opens new opportunities for advertising across various channels, including digital platforms and, in the future, even traditional channels like television.

Using data bunkers requires close coordination between the involved parties, such as the company sharing the data and the platform receiving it. Ensuring proper data structuring and agreement on the identifiers used is crucial to guarantee successful matching.

Contact

All in all, data bunkers offer a promising solution to balance privacy with personalized advertising. Through secure data sharing and advanced analytics, companies can gain valuable insights and deliver targeted ads while protecting individual user privacy. It’s a positive step forward for the future of data-driven marketing. If you’d like to receive more information about data bunkers, please contact us at get in touch with us .

Marketing Agency: Is Bigger Always Better?

A recent trend promoted by both Meta and Google is the use of broad buying strategies. This approach means that advertisers no longer focus on precise audience segmentation, but instead target a wider audience with their ads. The idea behind this is that a larger reach allows for better optimization, making the campaign appear, or seem, highly successful.

However, the key question is whether these conversions are truly incremental. In other words, would they have happened anyway, even without the ad exposure? It’s essential for marketing agencies to thoroughly assess whether broad buying strategies genuinely contribute to increasing the number of conversions that wouldn’t have been achieved otherwise.

Examples

Examples of these ad types are Advantage Shopping Campaigns (ASC) or Performance Max campaigns. In such campaigns, it’s often difficult, or sometimes not possible at all, for marketing agencies to make adjustments. Typically, it comes down to uploading creatives, with very limited or no control over audience‑level targeting.

As a result, the agency has less control over performance and fewer levers to adjust. At the same time, it allows platforms like Google and Meta to focus on low‑hanging fruit, people who might have converted even without that final touchpoint. It’s vital to scrutinize the results of these campaigns closely. For example, analyze the share of branded search traffic within your campaigns or separate it from the rest. Branded searches naturally carry higher intent compared to non‑branded searches, so compare their impact with results from the earlier, more manual approach.

One way to do this is by running a geo‑experiment. Since measurement often takes place within the platforms themselves, it can be worthwhile to A/B test outside of the platforms. You can do this by excluding certain regions, creating a test group that sees your strategy or ads, and a control group that doesn’t. By comparing these groups, you can measure the true incremental impact of your channel, ads, or strategy, and this is where a marketing agency can really add value.

The Limitations of Metrics Like ROAS and CTR

Google and Meta excel at optimizing for digital metrics like Return on Ad Spend (ROAS) and Click‑Through Rate (CTR). These metrics are valuable indicators for understanding how ads perform at a surface level, providing insights into their effectiveness in driving clicks and conversions.

However, this is where the limitation lies: these metrics don’t tell the whole story. They don’t accurately measure the true impact of a campaign on overall business results. Evaluating qualitative outcomes, like brand awareness, brand loyalty, and long‑term business growth, is far more complex and often beyond the scope of the measurement tools available. This isn’t something Meta or Google typically account for by default.

In addition, there’s often a significant discrepancy between what your analytics platform measures and the data reported by Google or Meta. As a result, your results can appear better than they really are, making it unclear whether the actual impact has truly been achieved.

Examples

For YouTube campaigns, it can be worth conducting an ad recall or search survey. This goes beyond just looking at completed views or impressions, it asks questions like: “Which brand have you recently seen an ad for?” This way, you can measure whether the viewer truly remembers the ad. By focusing on actual recall, you gain deeper insight into its impact, rather than relying solely on the ‘video viewed’ metric.

Attribution

We also often see platforms attributing too optimistically, making it appear as if more revenue is being generated than is truly the case. In reality, results tend to be less impressive. Meta and Google use their own attribution models, relying on tagging and cookies. These methods enable them to track user behavior and attribute a conversion to an ad that was clicked or viewed. In many cases, the default attribution windows are 30‑day click and 14‑day view for Google, and 7‑day click and 1‑day view for Meta.

Since every platform uses its own attribution model, and results aren’t deduplicated, you often end up with more conversions reported than were actually achieved. It also raises the question of whether a conversion can still be credited to a channel when the user saw an ad two weeks ago. That’s why it’s important to experiment extensively. Geo Experiments or Conversion Lift Studies can be very effective for this. With a Conversion Lift Study, you can also test different attribution tactics. For example, which approach has a greater impact on actual sales: an attribution model based on both views and clicks, or one that focuses exclusively on clicks?

The Role of a Marketing Agency: Critical Thinking and Maintaining Balance

We’re seeing an increasing trend where agencies simply follow the recommendations of giants like Google and Meta. At Draft Digital, we take a more critical approach. Blindly doing what these platforms suggest can reduce the quality of media buying and limit opportunities for learning. The broader the targeting, the harder it becomes to run meaningful A/B tests. Moreover, this approach tends to focus on platform metrics rather than actual business objectives. At Draft Digital, we excel at translating business goals into clear, actionable marketing KPIs.

That doesn’t mean that Meta and Google have no value, on the contrary, they can be excellent platforms for reaching the right audiences. It’s all about finding the right balance between allowing the algorithm to do its work and making smart, strategic buying decisions. A marketing agency needs to take the lead here, guiding the client to achieve real growth. This means setting up a solid testing roadmap, creating a clear KPI framework, and providing the client with the training and insights they need to make the most of their marketing efforts.

Conclusion

In a world where technology and data play an ever‑increasing role, it’s tempting to rely solely on the recommendations of powerful platforms like Meta and Google. However, marketing agencies must remember that truly effective marketing is a balance between data and insight, between measurable results and intangible impact. It’s the responsibility of marketing professionals to find this balance, ensuring they don’t just serve the objectives of the platforms, but, above all, the goals of their clients.

By thinking critically and developing strategies that address both the quantitative and qualitative sides of marketing, agencies can become invaluable partners in driving sustainable growth and delivering impactful results for their clients. At Draft, we strongly believe in maintaining an experienced, critical eye with our partners — it’s what keeps both sides sharp and leads to better outcomes. Interested in exploring this further? get in touch with us .

Smartly, an advanced technology platform, offers advertisers the capabilities they need for more complex campaign setups within Meta. This way, you can get more out of your social media campaigns and take performance to the next level.

What is Smartly?

Before we dive into why Smartly is so valuable, it’s important to understand what it is. Smartly is an advanced advertising platform that allows advertisers to manage and optimize digital ad campaigns more effectively and efficiently. It uses AI and machine learning to personalize ads and optimize them for maximum impact. Smartly easily connects with Meta Business Manager, making it highly accessible, although it still requires a certain level of expertise. Smartly is more complex than Meta itself, so it’s crucial to have the right people in place to get the best results.

What Are the Benefits for Your Performance?

1. Advanced Targeting and Personalization: Within the Meta advertising ecosystem, reaching the right audience is critical, even when broader targeting is used. Smartly offers advanced targeting capabilities based on demographics, behavior, interests, and even purchase intent.
2. Dynamic Advertisements: While Meta only allows static ad formats, Smartly enables advertisers to create dynamic ads where the content automatically adapts to the user’s characteristics and interests. This is especially valuable for (retail) advertisers, as it allows them to showcase products that are highly relevant to each individual consumer, for example, through retargeting based on product views or using countdowns within ads. Do keep in mind that you can have a maximum of 1,000 dynamic ads live in your account.
3. Optimization at Scale: In digital marketing, automation is a key growth driver, and Smartly is no exception. The platform uses automated optimization techniques to constantly monitor and adjust campaigns based on performance data. This allows advertisers to spend their budgets more efficiently and drive better results. A great example of this is the use of budget pools, which enable smarter allocation of spend across campaigns.
4. A/B Testing and Insights The Smartly platform offers extensive A/B testing capabilities, allowing advertisers to test different elements of an ad to determine which version delivers the best results. This empowers them to continuously refine and optimize their creative strategies based on concrete, data‑driven insights.

Why Wouldn’t You Use Smartly for Performance?

Looking at the benefits of Smartly, it’s easy to see why it can be an attractive choice. However, it’s worth noting that Smartly is a relatively costly tool. Once you link your Meta ad account to Smartly, you’ll pay a fee on all ad spend, regardless of whether you actively use its features. (Tip: consider using one account for Smartly campaigns and another for regular campaigns.) That’s why it’s important to clearly assess the added value Smartly delivers. In most cases, this comes down to performance improvements and time savings. If these benefits justify the cost, Smartly can be a valuable addition to your regular social platforms. Need help making the right choice? We’re happy to support you, contact us for an audit.