Hold rate measures whether the body of your ad keeps the attention the hook won. A great hook with a weak hold means a strong opening and a forgettable message.
Example: 9,000 completions ÷ 45,000 three-second views × 100 = 20% hold rate. Read it with hook rate to know whether to fix the opening or the middle.
How well the body of the video holds attention.
Example — Hold Rate in practice
Imagine Anghami runs a 15-second Reels ad for a Ramadan playlist launch. Of the 200,000 viewers who stopped scrolling in the first 3 seconds (the hook), only 60,000 watch all the way to the end. Hold rate = 60,000 / 200,000 = 30%, telling Anghami's team the hook works but the middle of the ad loses two-thirds of its audience before the call-to-action.
لنتخيل أن أنغامي تُطلق إعلان Reels مدته 15 ثانية للترويج لقائمة تشغيل رمضان. من بين 200,000 مشاهد توقفوا عن التمرير خلال أول 3 ثوانٍ (الخطّاف)، يُكمل 60,000 فقط المشاهدة حتى النهاية. معدل الاستمرارية = 60,000 ÷ 200,000 = 30%، وهو ما يخبر فريق أنغامي أن الخطّاف ناجح، لكن منتصف الإعلان يفقد ثلثي الجمهور قبل الوصول إلى الدعوة لاتخاذ إجراء.
Hold Rate, properly understood
Hold rate measures whether the body of your video keeps the attention the hook already won. The formula is (Video completions ÷ 3-second views) × 100 — note the denominator is 3-second views, not impressions, which is what separates it from hook rate. "Completion" is defined differently by platform: Meta's ThruPlay counts a 15-second watch or a full view of shorter videos, TikTok reports watch-through at 25/50/75/100% quartiles, Snapchat has its own view-time buckets. All of this data lives inside the ad platform's video-metrics tab, not in analytics or a CRM, because it depends on video-level view-through tracking the platform alone can measure.
Vertical, sound-off-by-default video on Snapchat, TikTok, and Reels dominates Gulf mobile feeds, so hold rate is heavily shaped by whether captions and pacing work without audio. Ramadan and Eid creative often runs longer-form (family scenes, behind-the-scenes) and tests hold very differently than a quick product demo, so don't benchmark seasonal-format ads against always-on performance-format ads. Bilingual brands should segment hold rate by language variant — an Arabic voiceover cut and an English-subtitled cut of the same video frequently show different retention curves for the same audience.
Comparing hold rate across videos of different lengths is a classic misread: a 6-second video will structurally post a higher hold rate than a 30-second one simply because there's less time to lose people, so only benchmark within similar-length cohorts. When hold rate collapses at one specific timestamp rather than declining gradually, that usually flags a specific frame — a jump cut, a price reveal, a logo card — rather than general fatigue; pull the full drop-off curve, not just the single endpoint number, to find exactly where viewers leave.
Read hold rate together with hook rate (getting the stop) and with downstream CTR and CVR (whether the attention converts). A high hook, high hold, low CTR ad usually means the story is compelling but the ask or CTA is weak or buried; a high hook, low hold ad means the opening promise isn't paying off fast enough in the body of the video.
It's worth separating hold rate measured on organic or owned content from hold rate on paid placements, since the two audiences arrive with very different intent — someone who followed a brand and is watching a feed post has already opted in, while someone served a paid ad mid-scroll has not, and blending the two into a single "video performance" number obscures which context is actually driving the retention number. When testing new creative formats, hold rate is also one of the faster signals to read, since it doesn't need a full campaign's worth of spend to reach a stable sample size the way downstream conversion metrics do — it's often the first useful readout available, well before CPA or ROAS have settled. Reporting hold rate separately for organic and paid placements, rather than blending both into one video-performance number, keeps this distinction visible in the data itself rather than buried in an average.
Put it to work
- Pull completions and 3-second views from the platform's video metrics tab, not GA or CRM.
- Compare hold rate only against videos of similar length and format.
- Review the full retention/drop-off curve, not just the completion percentage.
- Segment hold rate by language variant for bilingual Arabic/English creative.
- Treat a sharp single-frame drop as a creative fix, not general audience fatigue.
- Read hold rate together with hook rate and downstream CTR to diagnose story vs. CTA problems.
Turn the theory into real pipeline.
Get a free 60-second growth audit of your site, or talk to a strategist about your funnel.