# Facebook Video Ads That Lasted: What the Longest Runners Open With > In one search of the Marxx library, most long-running Facebook video ads did not lead with an offer. - **Author**: Marxx - **Published**: 2026-10-31 - **Category**: Creative strategy - **URL**: https://marxx.ai/posts/facebook-video-ads-that-lasted --- **Which Facebook video ads last, and how do they open?** In one search of the Marxx library, most long-running Facebook video ads did not lead with an offer. The longest runners opened more often on a brand, a mascot or a question, and an offer was the least common opening among the 13 ads that ran 200 days or more. # Facebook Video Ads That Lasted: What the Longest Runners Open With Most Facebook video ad advice starts from the offer. Lead with the discount, the bundle, the deadline. That advice is reasonable, and it is also untested against what actually keeps running. So we looked at what keeps running. We pulled 500 active Facebook video ads that have run 90 days or more from the Marxx library, sorted by run length, and read the top 50. For each one, we coded the opening by hand from its hook: does it lead with an offer or price, a question or pain, a story or humour, a product or claim, or a brand, mascot or recommendation. The counts below are approximate, and they come from one search. What the 50 open on: offer first 10, question or pain 5, story or humour 12, product or claim 9, brand or recommendation 14 ## Most of the sample does not open on the offer Offer-led openings are common in the sample, about 10 of 50. Brand, mascot or recommendation openings are the largest single group at about 14, and story or humour sits close behind at about 12. Question or pain openings and product-or-claim openings make up the rest. Taken alone, that is only a count of what runs. It does not say which opening causes a run to last. An advertiser with a strong brand may open on the brand because the brand is the asset, not because the opening is special. ## Offer-first ads run for a median of about 107 days Among the offer-first ads in the sample, the median run is about 107 days, and most end near 100. Those are real runs. They are not short tests. Run lengths by opening type: offer-first median about 107 days, question-or-pain includes the longest run at 1,636 days, brand or recommendation holds 6 of the 13 ads running 200 days or more The longest run in the whole sample, 1,636 days, is a question-led ad. That is one ad, so it tells us a question can hold attention for years. It does not tell us questions beat offers on average. ## Who ran 200 days or more Thirteen ads in the sample have run 200 days or more. Their openings are more useful to look at than the median, because long runs are the ads an advertiser has chosen to keep paying for. Who ran 200 days or more: 4 open on a question or pain, 6 on a brand, mascot or recommendation, 3 on an offer or discount Four of the 13 open on a question or pain. Six open on a brand, mascot or recommendation. Three open on an offer or discount. So the offer is present among the long runners, but it is the least common opening in that group, not the most common. The learning is not that offers fail. Among the long runners, the openings that show up most are a known brand, a mascot or a question about the viewer's own situation. We can't tell from this sample whether those openings cause the longer run, but they are the ones to test first against an offer. ## Three openings for one offer Here is one hypothetical offer, opened three ways. The offer is the same in each. Only the first line changes. Three openings for one offer: offer-first, question-first and brand-first The offer-first line states the arithmetic. The question-first line names a cost the viewer already feels. The brand-first line makes the offer feel like a reward from a shop they know. All three are hypothetical, and none has been tested on a live account. They show the choice, not a result. ## The four-step opening test Don't copy a long-running ad's opening straight into your own ad. Test it on your own offer first. The opening test, four steps: write three openings, run each as its own ad, read day 14 and day 30 holds, keep the opening that still runs 1. **Write three openings for one offer,** one offer-led, one question-led and one brand-led, so the test compares openings rather than offers. 2. **Run each as its own ad,** with the same audience and the same budget. 3. **Read day 14 and day 30, not day one.** Early clicks can favour a loud opening that doesn't hold. 4. **Keep the opening that still runs,** then test the offer itself. Day one clicks measure curiosity. Day 30 measures whether the opening keeps working once the novelty has worn off, which is the quality the long runners in the sample have shown by staying live. ## Five questions before you copy a winner Before you copy an opening from a long runner, check a few things about the ad you are copying. Five questions before you copy a winner - **Is it still running, or did it just start?** A new ad hasn't yet shown whether it lasts. - **Does the run survive a second search, or only the top 50?** One search is a single slice of the library. - **Is the brand also running other openings?** A brand may be testing several at once, and the one you see may not be the one doing the work. - **Is the offer the reason, or the wrapper?** If the offer is the wrapper, the opening is what you are copying. - **Would the same opening work for a brand with no offer?** If it only works with a discount attached, it is an offer, not an opening. ## What this sample cannot tell you The sample is one search, the top 50 of 500 results, and only ads still active when we looked. That last point is survivorship: ads that ran for 90 days and then stopped are not in the sample, so the counts describe what survived, not what everyone tried. The hook coding is ours, done by hand, and the counts are approximate. Use the pattern as a starting point for your own test, not as proof that one opening wins. ## Where Marxx fits Marxx lets you see what long-running Facebook video ads open with, across the library, before you brief your own. The generation canvas can then take one brief and write openings to test against each other, so the test in this post runs on your offer rather than on a sample. **[Trial the canvas: turn one brief into a finished ad ->](https://www.marxx.ai)** *Related: [Best AI ad generators in 2026: which stage of the brief each tool covers](/blog/best-ai-ad-generators-2026) - [Automatic reel makers](/blog/automatic-reel-makers) - [Best AI UGC tools](/blog/best-ai-ugc-tools)* ## FAQ **Do offers work in Facebook video ads?** Yes, they run. In this sample, offer-first ads ran for a median of about 107 days. But they are not the most common opening among the longest runners, so test the offer against a stronger opening rather than assuming it. **Which opening should I start with?** Write three openings for one offer and run them against each other for at least 30 days. Keep the one that still runs. **Is one search enough to copy a long-running ad?** No. Use it to spot patterns, then confirm on your own account before you scale. --- - [More Creative strategy articles](https://marxx.ai/posts/category/creative-strategy) - [All articles](https://marxx.ai/posts)