Pre-spend creative scoring: test ideas before you pay the auction

Pre-spend creative scoring predicts how likely an ad is to earn clicks before it runs. It compares a creative's message, format and visual approach with patterns from real audience response, so weak concepts can be reworked before launch. It does not replace live testing; it decides which ideas deserve the testing budget.

Why test before you spend?

The usual creative workflow treats the ad auction as a research lab. Five concepts go live, budget spreads across them, and a week later two are switched off. You learned something, but you paid full media rates to learn it, and some of those lessons were predictable. In accounts with modest budgets, the testing tax can eat a large share of monthly spend before anything profitable scales.

The proposition behind FMedia's creative tools comes straight from SOMIN's work on paid-media profitability: score creative clickability before media spend, and invest in what truly converts. It is the practical answer to chasing illusions. Instead of discovering in the auction that an idea was weak, you filter on evidence first.

How does creative clickability scoring work?

FMedia uses SOMIN's creative clickability prediction. SOMIN, an AI audience-research and paid-media intelligence platform, has learned from large volumes of real ads and the audience response they earned. When you submit a creative, it reads elements that tend to drive or suppress clicks:

• Message: the core promise, its clarity and how it maps to tensions the audience actually voices.
• Hook: the first frame or headline and whether it earns attention in a feed.
• Format: static, carousel, short video, and the fit with placement.
• Visual approach: composition, product visibility, people versus product, contrast.
• Audience fit: how the concept aligns with what the target segment responds to.

The output is a score plus the reasons behind it, so a creative team can fix the weak element instead of guessing.

What scoring cannot do

Scores are predictions, not verdicts. They are strongest at separating clearly weak concepts from promising ones, and weaker at ranking two strong ideas against each other. They predict clickability, not post-click behaviour, so a highly clickable creative can still attract buyers who return the product. That is why FMedia reads scores alongside true-profit ROAS once ads go live. The pre-spend score decides what to test; profit decides what to scale.

A five-step pre-spend workflow

1. Brief from evidence. Start with audience tensions and recent winners, not a blank page.
2. Produce more rough concepts than you need. Eight rough ideas beat three polished ones at this stage.
3. Score every concept. Read the reasons, not just the number.
4. Rework or drop low scorers. Often a new hook or clearer product shot is enough.
5. Launch the shortlist with equal starting budgets. Let live data, judged on profit, decide which to scale.

Does scoring make all creative look the same?

It can, if you treat the score as a target rather than a filter. The fix is to score for clickability while protecting your brand's distinct voice. Score variations within your own style, not imitations of competitors. Our sister brand GPT3 Marketing writes thoughtfully about how generative AI pushes brands towards sameness and how grounding content in what real audiences respond to is the antidote. The same principle applies here: evidence should sharpen your ideas, not average them.

For an example of how audience insight shapes creative direction on the brand side, the Pepsi India case study on somin.ai is worth a look.

Definitions

• Clickability: the predicted likelihood that a creative earns a click from its target audience in a feed or results page.
• Testing tax: media spend used to learn that a creative underperforms.
• Shortlist: the subset of scored concepts that go live for paid testing.

An illustrative before and after

Imagine a skincare brand preparing a refresh with eight concepts. Scoring flags two as weak: one buries the product behind a lifestyle scene, the other leads with a claim the audience does not care about. A third scores in the middle because its hook arrives too late in the video. The team drops the first two, re-cuts the third so the product appears in the opening second, and launches six instead of eight. The testing budget per concept rises, each test reaches a readable result sooner, and the live data is spent comparing good ideas. None of that required a new tool for the creative team, only a step between briefing and launch.

Let the auction choose between good ideas. Do not pay it to reject bad ones.

Where scoring fits with fatigue and budget

Pre-spend scoring is most valuable at the moment a creative tires. When FMedia flags a creative moving into "rest" (see our creative fatigue guide), the replacement brief starts from current audience signals and every candidate is scored before launch. The refresh is faster, and the first week of the new creative wastes less.

It also helps budget reallocation. When spend moves to a campaign with profitable headroom, that campaign needs creative capable of absorbing it. Scoring tells you whether you have it before you commit the money.

How FMedia runs it

In FMedia, you upload concepts as images, video or copy, choose the target audience and placement, and get a clickability score with reasons from SOMIN within the same workspace where your profit numbers live. Scores are stored, so after launch you can compare predictions with results and see where your own judgment, and the model, were right or wrong. Over time that loop makes the creative team sharper, which is the point.

Frequently asked questions

Is pre-spend creative scoring accurate?

It is best at separating clearly weak creatives from promising ones. Treat it as a filter for what to test, then let live results judged on profit decide what to scale.

Does scoring replace A/B testing?

No. It makes A/B testing cheaper by removing predictable losers first, so testing budget goes to ideas with a real chance.

What does FMedia use for creative scoring?

FMedia uses SOMIN's creative clickability prediction, part of SOMIN's paid-media intelligence, which compares creatives with patterns from real audience response.