The Creative Memory Gap: Why Your Ads Keep Starting From Zero

Monday, 9:40 a.m.

Sam runs growth at a large wellness brand selling in India and the UAE. Two agencies, hundreds of creators, two markets, each with its own learnings. On Monday morning she sees it: the creative pattern that carried India last quarter, a dermatologist explaining one ingredient in 20 seconds, is fading. Cost per order is climbing.

She types into the team WhatsApp group: “What made those work? Will the same pattern hold in the UAE?”

Three different answers come back. The strategist who briefed the originals left in August. One agency's deck says “authentic, expert-led”; the other agency never ran the pattern. The creator tracker sits in a separate Excel sheet, tagged differently.

By lunch, both agencies are briefed on “something like that”. Nobody is quite sure what “that” was.

This isn't a talent problem. It's a memory problem.

Sam is a composite, drawn from our conversations with direct-to-consumer (D2C) growth teams.

Is creative really the biggest lever left in performance marketing?

Bidding, targeting and budget pacing are now automated by platform algorithms such as Meta Advantage+ and Google Performance Max. Every brand gets roughly the same machine. What's left is the ad itself.

Nielsen's analysis of around 500 consumer packaged goods (CPG) campaigns found that creative drives 49% of the sales lift from advertising. Yet a 2025 Google and Kantar study found that only 46.2% of marketers have any analysis in place to measure creative effectiveness.

Your best lever is your least-measured one.

Why do D2C brands keep repeating creative mistakes?

Sam's Monday is the norm. Most performance teams run the same loop:

• Ship 20 ads. One or two win. Nobody can say why.
• The winner is fatigued. The next brief says “make something like that one.”
• The agency changes or a strategist leaves, and the learnings leave with them.
• Six months later, the team re-runs a test it already lost.

Motion's 2026 benchmark of $1.29 billion in Meta ad spend found that only about 5% of creatives become winners, taking 55% of spend. Volume without memory is just a faster way to repeat mistakes.

At scale the loop fragments. Creator content now makes up 44% of paid media creative, according to CreatorIQ, and each agency, creator roster and market keeps its own notes, so a pattern that wins in one market is never cleanly tested in the next. And almost every public creative benchmark is built on US or European data, so brands in India, the Middle East and Southeast Asia have to build their own memory, from their own data.

What is creative intelligence?

Creative intelligence is the practice of breaking every ad into its parts, measuring which parts are linked to business results, and feeding those learnings into the next ad.

Most creative analytics tools stop at a dashboard of which ads won. Galvor is built for the next part: carrying the learning into the next brief, and keeping it after people move on.

It works alongside your existing reporting and agencies, not instead of them, and reads every agency's and creator's output against one set of attributes, so comparisons are like for like.

At Galvor, it runs in four steps:

1. Decode. Break every ad, yours and your competitors', into 50+ attributes: hook, presenter, pacing, claim, format, language and call to action (CTA).
2. Prove. Read each attribute against the outcomes that matter: cost per order (CPO), customer acquisition cost (CAC) and return on ad spend (RoAS). Not “this ad won”, but “for this brand, ads with a dermatologist presenter and a problem-first hook have delivered a lower CAC than everyday-user testimonials.”
3. Remember. Store every learning in a Brand Memory: voice, visual identity, what was tested, what won and what lost, built from the brand's own first-party data and never pooled with another brand's.
4. Create. Generate the next briefs and assets from the winning pattern. Your team approves everything before it ships.

A word on method. Platform data shows association, not cause: Advantage+ and Performance Max decide who sees each ad, so a “winning” attribute may partly reflect its audience. That is why Galvor treats each pattern as a hypothesis. The next round of briefs puts it back in market, and the result confirms or overturns it in memory.

Memory is what lets the loop compound: each round starts from what the last one proved. Brands running this loop every 14 days have seen RoAS rise by 91% in skincare and 67% in supplements.

Where creative calls actually get made: WhatsApp, Excel, email

Sam's team doesn't make creative calls in an analytics tool. They make them in a WhatsApp group at 11 p.m., in an Excel tracker shared with the agencies, and in an email where the founder approves a script with “ok, go”. That is where learning lives, and where it gets lost.

So Vira, Galvor's assistant, works inside the tools teams already use:

• WhatsApp: ask what worked last quarter and get the answer, with numbers, in the same thread.
• Excel and Google Sheets: creative and performance trackers stay where they are and gain decoded attributes, with no new tool to learn.
• Email: briefs, approvals and agency feedback sent to Vira are captured as they happen.

Vira reads only the WhatsApp groups, sheets and email threads your team chooses to connect to, and that access can be withdrawn at any time. Galvor has read-only access to ad accounts: it never edits campaigns or moves budget.

Within what you connect, every exchange feeds Brand Memory. That is the difference between a tool your team has to remember to use and a system that remembers for them.

Monday, three months later

Sam's best ad is fading again. Every ad does.

This time she asks Vira in the same WhatsApp group. Within a minute she has the three attributes behind the India winners (an expert presenter, a problem-first hook and a single-ingredient claim), the CAC each delivered in India and in the UAE, and new scripts for both agencies that keep those elements. Tests already lost in either market are flagged so nobody runs them again.

The strategist who left is still gone. The learning isn't.

Three questions to ask your team this fortnight

1. Can we name the top three attributes behind our last five winning ads, with the CAC for each?
2. Where do our creative learnings live (WhatsApp, Excel, someone's inbox), and what happens to them when that person leaves?
3. Do we judge every agency, creator and market against the same attributes?

If the answer to any of these is “not sure”, that's your creative memory gap.

Why did we build Galvor on Google Cloud and Gemini?

We built Galvor on Google Cloud and Gemini because they let us give brands what they need most from creative intelligence.

Three reasons, each visible in what you get.

1. You can buy it the way you already buy Google Cloud. Once our Google Cloud Marketplace listing is live, you can pay through your existing Google Cloud billing, and eligible purchases can count towards your committed spend.

2. Ads are video, and video has to be understood all at once. Gemini takes in picture, voice and on-screen text together, so we decode a hook the way a viewer experiences it.

3. Your brand's data stays your brand's. Your data is encrypted, isolated and never pooled, and ad-account access is read-only.

Next issue: the 2-second hook problem

What 50+ attributes tell us about the hooks that work in India, the Middle East and Southeast Asia, drawn from aggregated, anonymised attribute data, never one brand's results.

Want to know which attributes are behind your own winning ads? Talk to the founders about a read-only creative audit. We decode your recent ads across agencies and show you three things: the attributes behind your winners, the tests you have already lost, and how your agencies and creators compare, side by side.