Creative Engine: How to make good advertising scalable

A creative engine connects idea development, production, quality assurance and analysis. For Meta Ads, it creates a repeatable process that delivers new advertising assets and builds on existing knowledge.

Imagine a typical production round. The media team needs new ads. The creative team is waiting for the previous round's results. Production could start, but the product claims have not yet been approved. By the end of the week, there are plenty of drafts and very little material ready to use.

Doubling the number of drafts in this situation also means handling more selection and approval work. This is where building a creative engine starts: at the handovers between insight, idea and published ad.

For brands, the task is concrete. How can you regularly develop fresh material that fits the brand, tells a distinctive story and answers a useful test question?

The right volume starts with the learning question

“500 ads per week” sounds like a sign of capacity. To interpret that number, you first need to know what is being counted.

For planning, we distinguish three levels:

A simple calculation: three concepts, each with two openings and three format adaptations, produce 18 files. Behind them are three ideas and six variants with different openings. The file count alone says little about how much new ground a brand is exploring.

Meta also describes creative diversification as variation in messages, visuals and formats. Its guidance includes reusing existing assets, working with creators and AI-enabled production. Source: Meta Blueprint

Our recommendation is to plan the questions a production round should answer first. Then determine scope and sequence based on budget, expected results and capacity for analysis. A wider selection at the initial review stage may be useful; deciding which variants actually enter a test is a separate step.

A production process with clear handovers

A creative engine needs a process everyone involved can understand. We suggest five steps as a starting point. Each ends with a specific output.

1. Turn observations into a brief. Customer questions, sales feedback and campaign data provide starting points. The team selects a relevant question: Which objection remains unanswered? Which product benefit is still being explained too abstractly? The output is a brief defining the reason to buy, message, evidence and test question.

2. Review concepts early. Before full production begins, assess ideas as sketches, storyboards or short copy drafts. The story, its connection to the brand and the reason to believe it must be clear. At this stage, alternative approaches can be discussed in a focused way. The output is a selection of concepts with a documented rationale.

3. Produce material for the planned variants. A shoot can cover different openings, product details and endings. Design can use reusable elements. The selected ideas determine which components are needed. The output is material from which the intended variants can actually be made.

4. Review and approve ads. Copy, imagery, sound, product representation and the landing page are considered as connected communication. Approval includes an identifiable version; later changes to the content must remain traceable. The output is an assigned, approved version in the required formats with documented usage rights.

5. Turn results into the next assignment. Concepts, variants and delivery remain identifiable in the analysis. The team records the observation, its limitations and the resulting decision. The output might be a more precise brief for the next demonstration or a decision to pause an approach.

For this process to work, demand, available material, review capacity and the media plan need to fit together. Anyone commissioning new production needs to know who will review it and when it can be put to meaningful use.

Give static images and videos different jobs

The format mix should reflect how an idea can be communicated clearly.

A static ad can communicate a benefit, a product view or a concise comparison in one image. Video allows room for a process, movement, use and a story unfolding over several moments. A sequence of images can explain a decision step by step.

Consider a product demonstration. If operation is the point, a short film can show the actual steps. If the message concerns available sizes, a clearly designed overview may be more suitable. Both tasks can belong to the same campaign.

The communication task guides the choice of format. Production effort, existing material, placement and previous observations also inform the decision. A universal ratio between images and videos is not a sufficient basis for planning.

Reporting also needs to define what “share” refers to. The share of produced files, delivered impressions and allocated budget describe different things. A large number of static ads does not reveal their share of spend.

For content production, it is therefore worth considering formats during concept development. A vertical crop, for example, still needs to communicate the product, message and subtitles clearly.

Make quality concrete before publication

At greater volume, quality criteria need to be specific enough for everyday use. “Looks good” is rarely enough for a reliable approval process.

Four areas help guide review:

• Claim and evidence: Are the product details correct? Does the demonstration show what it claims? Are any quoted experiences documented?
• Brand and design: Do tone, imagery and product representation fit the brand? Does it remain recognisable across different ideas?
• Clarity and use: Is the copy legible, are subtitles accurate and are important elements visible in the intended format? Are the required usage rights documented?
• Offer and destination: Do price, variant, availability and call to action match the ad's destination?

AI can prepare parts of this work: structure drafts, develop alternative openings, sort material, suggest subtitles or flag defined formal deviations. Suitable tasks depend on the tools, source material and review process.

With generated product images in particular, the team needs to check that properties, proportions and functions are represented correctly. A plausibly worded product benefit also needs a real basis.

We take responsibility for concept and quality. We agree specialist approvals and responsibilities with your team. These criteria remain relevant if a platform subsequently offers automatic adjustments to ad creative; such features should be selected deliberately and their visible results checked.

One potential quality advantage of AI is making different approaches visible early and incorporating criticism before costly production. Whether this succeeds in a particular workflow can be assessed through rework, errors and the quality of approved versions.

Use analysis to shape the next production brief

Declining performance is first a reason to investigate. Repeated exposure to the same advertising may be one explanation. Changes in delivery, the offer, landing page, competition and measurement should also be examined. Rising CAC alone does not establish ad fatigue.

We therefore assess connected observations: How are frequency and responses changing? Are people reaching the landing page? Do relevant purchases or enquiries follow? A fixed replacement schedule for every asset would overlook these differences.

Meta's Performance 5 recommendations consider creative diversification alongside account structure, automation, data quality and results validation. This supports an approach that investigates several possible causes. Source: Meta Blueprint

The comparison method also needs to fit the question. An ongoing campaign report shows results under its particular delivery conditions. Appropriate A/B or lift tests can be considered for targeted comparisons or questions about incremental impact; Meta also describes these in its creative measurement guidance. Source: Meta Blueprint

One possible working rhythm combines regular planning and approval with analysis once the agreed evidence is available. A weekly meeting does not automatically mean every test can be decided after one week.

For the creative engine itself, track three things: time from a complete brief to approval, effort including rework, and the share of published variants that inform a reasoned next decision. Together with brand consistency and errors, these show where the process needs to improve.

Creative Engineering makes collaboration productive

At Davies Meyer, we connect brand strategy, creative work, technology and paid social. For a creative engine, this means product knowledge, idea development, production and analysis have shared briefs, traceable handovers and clear owners.

Your in-house team contributes customer understanding and specialist experience. We add creative perspectives, execution and the connection to the technical process. Together, we establish which components can be reused, where a new idea is needed and how results inform the next round.

For the strategic foundation, read Meta Ads: Why targeting starts with the idea, covering reasons to buy, evidence and useful creative tests.

Where does your ad production get stuck today?

Bring a current campaign and the path it takes from brief to approval. Together, we can examine which handover causes the most work and what an appropriate production and testing process could look like.

Discuss your creative production