The challenge
GoDaddy sells dozens of products in dozens of countries, and every one of them needs advertising that looks like GoDaddy. That is not one design problem repeated — it is a multiplication. Each product against each market, each market in its own language, each language across every ad size the networks will take. The count runs into the thousands before anyone has had a single creative idea.
The traditional answer to that is people: a standing team of designers and developers producing variants by hand, and a review process to catch anything off-brand before it ships. It works, expensively, right up until review becomes the constraint — and past that point adding designers makes things worse rather than better, because every extra pair of hands produces more work for the same reviewers.
Our approach
DAS replaced hand production with a pipeline. Creative was defined once and generated programmatically across every product, market, language and ad size a campaign called for, so brand consistency stopped being something a reviewer looked for at the end and became a property of the thing that made the ad.
Around the generator sat the part that made it survivable day to day: a review system built for the volume it was about to receive. Campaigns arrive as sets rather than as loose files, each row carrying its own variant count. Opening one shows every locale beside every other, every size within each locale, and the rendered ad itself rather than a filename — so a reviewer reads a campaign in one pass instead of opening several hundred files. Sets move through review states together, and every variation keeps its own version history and packaged download.
Finished work went out as Google DCM packages, which is where the pipeline deliberately stopped. Its job was to put correct, on-brand, correctly targeted assets into the ad platform — not to run the media buy.
The outcome
The pipeline sustained around seven thousand ads a year across a multi-year engagement, into more than thirty markets.
Running it took four people where hand production had needed twenty. The saving is real, but it is not the interesting part — quality, consistency and turnaround all improved at the same time, which is the reverse of what normally happens when output goes up.