Does the Word “Best” Hurt Your ROAS?

AdScale’s creative language analysis keeps flagging the same word in underperforming ads. Here is what the data shows about “best” and ROAS, and what to test instead.

Open any eCommerce feed and count how many ads promise the best. The best leggings. The best moisturizer. The best coffee for your mornings. Superlatives in ad copy feel safe, which is exactly why they are everywhere. But when AdScale’s data science team ran ad copy through a creative language analysis and segmented campaigns by ROAS, “best” surfaced as one of the most over-represented words in the lowest-performing cohort. The pattern held across eCommerce verticals.

So the short answer is yes: the word “best” hurts your ROAS more often than it helps it. It is a reliable red flag in performance ad copy. AdScale’s analysis of creative metadata across thousands of eCommerce campaigns in its production data warehouse shows the same thing again and again. After filtering out brand names and common stop words, “best” appears far more often in low-ROAS creative than in high-ROAS creative. This is a correlation, not proof of causation, and that distinction matters. Still, when a single word keeps showing up at the bottom of the performance table in Beauty, in Fashion, and in Home Goods, it stops being a coincidence. It becomes a pattern worth acting on.

The creative that outperforms it does something structurally different. It drops the adjective and picks up a verb.

Key Takeaways

• AdScale’s creative language analysis flags “best” as a top-tier outlier in low-ROAS ad copy across eCommerce verticals, after normalizing for stop words and brand names.
• The relationship is correlational. “Best” does not mechanically lower ROAS, but it reliably marks creative that underperforms.
• Independent research supports the pattern: Nielsen Norman Group found objective language beat promotional language on measured usability, and a large patent study linked promotional wording to worse evaluation outcomes.
• High-ROAS creative in AdScale’s dataset leans on short action verbs such as “get” and “try.” Treat any specific replacement word as a hypothesis to A/B test, not a guarantee.
• The fix is structural: replace status claims about the brand with outcome statements about the customer, then measure the delta properly.

Why Do Ecommerce Ads Rely on Superlatives?

“Best” is the default word of a brand that has not finished its positioning homework. It feels safe. It compresses an entire value proposition into four letters. Moreover, nobody gets fired for approving it in a creative review meeting.

The problem sits on the other side of the screen. A shopper scrolling Instagram sees “best” a dozen times an hour. Every competitor claims it. None of them can all be right, and shoppers know it. As a result, the word carries zero new information. The brain treats it like white noise: filter and move on. Worse, “best” appears almost exclusively in advertising. It therefore works as a visual marker that says “this is an ad,” which accelerates scroll-past behavior before the product ever gets evaluated.

AdScale has watched this play out across thousands of eCommerce accounts. Brands with genuinely strong products keep restating their category, such as “best skincare for dry skin.” Meanwhile, their specific and testable advantages sit unused in a product spec sheet. The superlative is not just weak copy. It is a symptom of unextracted value.

What Does the Data Say About Superlatives in Ad Copy?

Three separate bodies of evidence point the same direction.

AdScale’s creative language analysis. The data science team ingested ad copy metadata from campaigns across AdScale’s production database. Campaigns were segmented into ROAS quintiles by vertical. Next, the team normalized for brand names and common stop words (the, and, a). Then it looked for words over-represented in the bottom quintiles relative to the top. “Best” surfaced as a top-tier outlier in the low-performance cohort across verticals including Beauty, Fashion, and Home Goods. Two honest limits apply. First, raw function words still have higher absolute counts, so “best” is not literally the single most frequent word in bad ads. Second, without controlled split tests, the finding is a cross-vertical correlation rather than a causal law. Even so, as a directional signal about which creative habits cluster with wasted spend, it is one of the most consistent patterns in the dataset.

Nielsen Norman Group’s usability research. In NN/g’s foundational web writing studies, users tested five versions of the same site content. The objective, non-promotional version measured 27% better on usability than the promotional “marketese” version. Users actively disliked boastful subjective claims. Jakob Nielsen’s explanation was that promotional language imposes a cognitive burden. Readers must filter hyperbole to reach the facts, and credibility drops when the exaggeration is visible. That is a near-perfect description of what “best” does inside an ad.

Promotional language in high-stakes evaluation. A 2026 analysis of 2.7 million US patent applications found that a higher density of promotional words was associated with a lower probability of a patent being granted. The gap in success rate between the lowest and highest promotional-density groups was 5.5 percentage points. Different arena, same human reaction: evaluators discount claims wrapped in hype.

Ad copy, website copy, patent filings. Three contexts, one consistent finding: superlative language costs trust.

What Is the Superlative Tax in Advertising?

Here is the frame AdScale uses internally. Every empty superlative in an ad charges a tax, and the tax gets paid twice.

The first payment is attention. The shopper’s brain must process the word, recognize it as unverifiable puffery, and discard it. That filtering happens in milliseconds. However, in a feed environment where the entire purchase consideration may last two seconds, spending any of them on a throwaway word is expensive.

The second payment is credibility. Once a shopper flags one claim as inflated, every following claim inherits the discount. The “best moisturizer” headline does not just fail on its own. It also makes the shopper trust the “clinically tested” line below it a little less.

Specific, verifiable language pays no tax. “Get smoother skin in 14 days” can be evaluated, remembered, and acted on. “Best skin cream” can only be believed or ignored, and the data suggests most shoppers choose ignored. That is the Superlative Tax: vague claims do not merely add nothing, they subtract from everything around them.

What Should Replace “Best” in Ad Copy?

The honest answer: there is no single magic replacement word. Any vendor claiming otherwise is selling the same puffery this article warns against. What AdScale’s data does show is a directional pattern. Creative in the top ROAS quintiles leans noticeably harder on short, high-velocity action verbs, with “get,” “try,” and “now” among the recurring elements. A separate AdScale analysis of 30 high-performing words in apparel ad copy found the same tilt toward concrete, action-oriented language over status adjectives.

The strongest working hypothesis is the “Get” pivot: replacing a static status adjective with a possession verb plus a specific outcome. Three reasons this structure keeps appearing in winning creative:

1. Direct possession. “Get” moves the sentence’s subject from the brand to the customer. “Best coffee” is about the company. “Get more from your mornings” is about the person paying.
2. Low cognitive load. A three-letter command initiates a mental transaction instead of demanding a judgment call about an unverifiable claim.
3. Specific utility. “Get smoother skin” hands the shopper a roadmap. “Best skin cream” hands them a boast and asks for faith.

Treat this as a testable creative hypothesis, not a benchmark. The correct way to know whether “get” beats “best” for a specific store is a controlled split test, covered in the next section. The wrong way is copying a universal claim from any blog post, including this one.

How Do You Remove Superlatives From Ad Copy?

1. Count the “best” instances. Export active headlines and primary text from Meta and Google ad sets. Search for “best” and other empty superlatives (top, greatest, ultimate, #1). Any primary headline built on one is a refresh candidate.
2. Extract the buried specifics. For each flagged ad, pull the product spec, review data, or guarantee the superlative was standing in for. “Best leggings” usually hides something like “squat-proof, tested through 100 washes.” The specific claim was there all along.
3. Write verb-first variants. Rebuild each flagged headline around an action verb plus a concrete outcome. “The best coffee for mornings” becomes “Get more from your mornings.” Keep everything else identical so the test isolates the copy.
4. Run a real split test. A trustworthy result needs a controlled A/B split with a statistically meaningful sample, ideally 1,000 or more conversions per variant. That washes out confounders like audience overlap and seasonality. Anything smaller is a directional read, not a verdict.
5. Judge on ROAS and CPA, not clicks. Verb-led copy can inflate CTR by attracting curiosity clicks. The metrics that settle the question are cost per acquisition and return on spend over the full test window.
6. Systematize the rotation. Creative fatigue guarantees today’s winner decays. Therefore, build the superlative audit into every refresh cycle, or let an automated system such as AdScale’s Agentic Ad Creatives generate and rotate verb-led variants continuously across Google and Meta.

Frequently Asked Questions

Stop Claiming. Start Promising.

Go back to that feed full of “best” ads. Every one of them asks the shopper to do the brand a favor: take our word for it. The data, both internal and independent, says shoppers decline that favor millions of times a day. The bill shows up as quiet, compounding wasted spend.

The alternative costs nothing but honesty. Find the specific thing the product actually does, put a verb in front of it, and hand the customer an outcome instead of a ranking. Then test it properly, because real confidence comes from split tests, not slogans.

Stop telling customers you are the best. Start telling them what they get.

Keep Learning

• Apparel Ad Copy That Converts: What 30 High-Performing Words Reveal – the word-level companion analysis of what top creative actually says.
• Understanding Poor Ad Performance: Why Your ROAS Is Low and How to Fix It – the diagnostic framework for underperforming campaigns, beyond copy.
• Agentic Ad Creatives: The Ad Creative AI Generator Built Into Your AdScale Platform – how automated creative generation keeps verb-led variants in rotation.
• The Full Guide to Google and Meta Automation for eCommerce – unifying budget and creative decisions across both platforms.