The Recurring Problem in Every Campaign: The Ad Works, the Date Is Outdated
An ad that works well is an asset. The trouble starts when the promotion it advertises ends, and you need to replace a date or price without compromising the rest of the design. It used to mean opening a source file, searching for the right font, and hoping nothing broke along the way. With a flat PNG file and no source, the situation is even worse.
This is where Nano-Banana Pro enters our process. Not as a magic tool, but as a solution to a specific, recurring problem: an ad that has already proven itself and just needs a refresh without initiating a brand-new design project.
Every client running an ongoing performance campaign encounters this, and not just in retail. A real estate agency updating an apartment price, a training course updating a cohort start date, a restaurant changing a specials menu: they all need the exact same minor, frequent update on an existing ad.
Until recently, there were two options: maintain organized source files indefinitely—requiring a level of discipline that clients managing their own assets don’t always have—or hire a designer for every small update and pay for it as if it were a full project.
What Nano-Banana Pro (and its Edit) Does Well
The model excels at targeted edits on an existing image: replacing a background, modifying lighting, or moving an element without touching the rest of the frame. This differs from generating an image from scratch, where granular control is much harder to maintain. When you have an existing ad that only needs part of it “refreshed,” the ability to keep everything else exactly as it was is the core advantage—especially when the ad has already cleared a lengthy client approval cycle.
Alongside it is Gemini 3 Pro Image, which we deploy when a brand-new image is required rather than an edit to an existing one. One edits, one generates, and the choice between them depends entirely on what the client already has.
Another use case worth mentioning, though less common: Nano-Banana Pro Edit also handles derivative formats, such as adapting the same ad across different aspect ratios for various platforms without compromising the original composition.
This distinction carries real budget implications. Generating an image from scratch requires multiple iterations to achieve a result aligned with brand identity. Targeted editing of an existing image almost always succeeds on the first or second attempt, because most of the composition is already correct from the outset.
Field Example: Weekly Ad Cycles for a Retail Client
For a client updating promotional offers weekly, we built a workflow where the background and composition remain fixed, while only the price and date change. The background was edited once in Nano-Banana Pro to fit the brand identity, and from then on, only the text layer is updated and overlaid each week.
What once required a full day of designer work now takes under an hour—often even less—because we are merely adjusting a single value rather than rebuilding an ad from scratch. Along the way, the client maintains long-term visual consistency since the background does not shift between variations. Their weekly workflow comes down to a single file: input the week’s price and date, run the overlay process, and receive all required format sizes ready to publish.
There is also a second, less obvious benefit: when an ad retains the same composition across months, it becomes far easier to pinpoint what drove performance shifts week over week. If every element changes in each iteration, it’s impossible to determine whether the price or the creative drove the outcome.
What We Still Don’t Hand Over to the Model
Here, too, our core rule applies: ad text, pricing, and dates are never generated directly by an AI model. Nano-Banana Pro edits the surrounding visuals, while the text layer is rendered separately and overlaid with precision. It’s the same principle guiding our video production, applied to static creatives.
This separation also streamlines future iterations: when the next promotion rolls around a week later, there is no need to run the image model again. Simply replacing the text layer over the existing, edited background is sufficient.
It also minimizes human error. Updating a single price field in a template file is hard to miss, whereas manually editing a complex design file risks accidentally nudging another layer out of place unnoticed.
When This Approach Saves the Most Time
The greatest payoff comes in campaigns with frequent update cycles: seasonal promotions, rotating inventory, or A/B testing variations that require a single minor change to evaluate a new angle. Instead of starting each iteration from zero, edits are built on top of what is already proven to work.
Our recommendation for onboarding new clients: before designing an ad from scratch, review existing creative assets to see if one is close to what’s needed, and simply edit it. In many cases, this also bypasses the full approval workflow, as the client has already reviewed and approved the foundational design.
Ahead of a new campaign, we review the previous quarter’s ads alongside the client to identify candidates for re-editing instead of net-new production. This saves budget, shortens approval timelines, and kicks off the campaign from a baseline that has already demonstrated market traction. Clients adopting this workflow over several months establish a core library of baseline ads, making every new campaign launch as simple as selecting from existing templates.
Do you have ads that need recurring, minor updates? It’s worth evaluating whether they qualify for this workflow. In most cases they do, and the efficiency gains are far greater than they appear at first glance.