Skip to content
Original research Published November 18, 2026

How real posts win. Faces win reach.

On TikTok, 57% of the most-viewed videos had a face on the cover, against 34% of the least-viewed (141 videos in each group, 48 small local business accounts). Instagram pointed the same way: 26 of 56 top posts showed a face against 14 of 56 bottom posts (28 accounts).

  • 48 local business accounts
  • 394 post covers coded blind by one AI coder
  • Data pulled November 2, 2026

Observational study of a hand-picked sample. Covers were coded by one AI coder, not double-coded. Read the method and limits below before you quote it.

Key findings

A face on the cover tracked with reach.

Each account's three most-viewed videos were compared with its three least-viewed. Every figure below has its sample size beside it.

57%vs34%

A face on the TikTok cover

Most-viewed against least-viewed videos. 80 of 141 against 48 of 141, 48 accounts.

46%vs25%

A face on the Instagram cover

26 of 56 against 14 of 56, 28 accounts.

64%

Owner or team covers in the top group

50 of 78 TikTok covers built around the owner or team. Half would be no difference.

  1. Faces showed up more on the videos that reached more people.

    80 of 141 most-viewed TikTok covers showed a face (56.7%), against 48 of 141 least-viewed (34.0%). n = 282 covers, 48 accounts

  2. It was not one big account doing the work.

    Of the 48 accounts, 28 showed a face more often in their top videos than their bottom ones. 3 showed it less often and 17 were level. Because each account is compared with itself, account size does not explain the gap. n = 48 accounts

  3. Instagram pointed the same way.

    26 of 56 top posts showed a face (46.4%), against 14 of 56 bottom posts (25.0%). The sample is smaller, so treat it as a second look, not proof. n = 112 posts, 28 accounts

  4. Covers about the owner or the team did best.

    50 of the 78 TikTok covers built around the owner or team were in the top group (64.1%). On Instagram it was 16 of 25 (64.0%). Half would be no difference. n = 78 + 25 covers

  5. Designed flyers and graphics looked weaker: an early signal.

    On TikTok, 3 of 12 designed-graphic covers were in the top group. On Instagram, 5 of 16. Together, 8 of 28. This is an early signal from 28 covers, too few to be sure. The same businesses appear on both platforms, and on TikTok alone the gap is not statistically firm. n = 12 + 16 covers

  6. Someone talking to the camera leaned the same way, but not clearly.

    22 top TikTok covers showed someone talking to the camera, against 12 bottom covers. That is 34 covers in all, too few to call. n = 34 of 282 covers

  7. Text on the cover made no difference.

    87 of 141 top covers carried text (61.7%), against 90 of 141 bottom covers (63.8%). n = 282 covers

Charts

What was on the cover, top against bottom.

Share of covers that showed a face

Most-viewed against least-viewed posts, same accounts in each pair.

Most-viewed posts Least-viewed posts

How often each kind of TikTok cover landed in the top group

Out of every 100 covers of that kind, how many belonged to a most-viewed video. The dashed line is 50: no difference. Sample sizes sit on the right.

Kinds with fewer than 20 covers (storefront, customers, result, flyer, other) are shown for completeness and swing a lot on a few posts.

Method and limits

How we did it, and what it cannot tell you.

Read this before you quote a number.

Who is in the sample
48 public TikTok accounts run by real local businesses, picked by hand from TikTok keyword searches such as “bakery day in the life” and “plumber day in the life”. We left out coaches and food influencers. Mix: Bakeries and cafes 14, Home services 11, Restaurants 10, Beauty and grooming 7, Retail and florists 5, B2B services 1. The same handles were looked up on Instagram, and 28 matched.
What we pulled
The 911 most recent TikToks from those accounts (up to 20 each), with views, likes, comments, shares and saves. For Instagram, the 334 most recent posts (up to 12 each) with likes and comments. Data was pulled on November 2, 2026 through a third-party data service from public pages.
What we compared
Inside each account, the three most-viewed TikToks against the three least-viewed (282 covers). On Instagram, the two highest against the two lowest by likes and comments, measured against that account’s typical post (112 posts). Comparing an account with itself takes account size out of the picture. TikToks posted after 2026-10-30 were left out because their views were still settling. The coded TikToks run from 2025-04-23 to 2026-10-30. Some Instagram accounts post rarely, so their sampled posts go back as far as 2016-03-19.
How covers were coded
One coder, an AI model, looked at each cover image with the outcome hidden and the covers shuffled. It recorded whether a face was visible, whether someone was talking to the camera, whether text sat on the cover, and the main subject. The coding was not repeated by a second coder, so coding error is possible and we did not measure it. A cover is one frame, not the whole video.
Limits
This is observational. It shows what travelled with reach, not what caused it. The sample is 48 accounts chosen by hand, not a random draw, and the search method favours businesses that already film themselves. It is not a measure of sales or buying intent. Views were still growing on recent posts. Results for the smaller cover types rest on a handful of posts.
What is not published
No account names, handles or post images. This page shows aggregates only. We also ran a check on buying-intent comments, but the sample was too small and the tagging cannot be reproduced from what we kept, so we are not reporting it.
Our interest
Pylot makes social media software for small businesses. We ran this study to decide what to build and are publishing it so others can check it.

The data

Every count, in one table.

Counts of covers in each group. Percentages are the share of that group.

Cover features, most-viewed against least-viewed posts
FeatureTikTok top (of 141)TikTok bottom (of 141)Instagram top (of 56)Instagram bottom (of 56)
Who is on the cover
Face visible80 (56.7%)48 (34.0%)26 (46.4%)14 (25.0%)
Hands or body only31 (22.0%)46 (32.6%)10 (17.9%)15 (26.8%)
No person30 (21.3%)47 (33.3%)20 (35.7%)27 (48.2%)
Other features (TikTok only)
Text on the cover87 (61.7%)90 (63.8%)n/an/a
Someone talking to the camera22 (15.6%)12 (8.5%)n/an/a
Main subject of the cover
Owner or team50 (35.5%)28 (19.9%)16 (28.6%)9 (16.1%)
Meme or skit20 (14.2%)16 (11.3%)8 (14.3%)3 (5.4%)
Space or storefront9 (6.4%)6 (4.3%)4 (7.1%)3 (5.4%)
Customers7 (5.0%)6 (4.3%)1 (1.8%)2 (3.6%)
Work in progress18 (12.8%)20 (14.2%)7 (12.5%)6 (10.7%)
Food or product26 (18.4%)32 (22.7%)12 (21.4%)15 (26.8%)
Result or before and after5 (3.5%)12 (8.5%)1 (1.8%)6 (10.7%)
Designed graphic or flyer3 (2.1%)9 (6.4%)5 (8.9%)11 (19.6%)
Other or off topic3 (2.1%)12 (8.5%)2 (3.6%)1 (1.8%)
Accounts compared with themselves (TikTok)
Face more often in top videos28 accounts higher, 3 lower, 17 level (of 48)

Sample: 282 TikTok covers from 48 accounts and 112 Instagram posts from 28 accounts. Main-subject rows add up to 141 per group on TikTok and 56 on Instagram.

Cite this

Quote it, link it.

Published November 18, 2026. You are welcome to quote the figures with a link back to this page.

Suggested citation

Pylot AI. (2026, November 18). How real posts win: a study of 48 local business accounts. https://www.getpylot.com/research/how-real-posts-win

Short form

Pylot AI, "How Real Posts Win", 2026-11-18, https://www.getpylot.com/research/how-real-posts-win

Questions about the data or the method? Write to support@getpylot.com. More from us: the blog.

What to do with this

Show the real thing. Start with a face.

Pylot asks for one small, real moment from your business, you film about 10 seconds on your phone, and Pylot edits it, writes the caption and posts it.

Start my free week

7 days for $0 · card required · then $49.99/month · cancel within 7 days to pay nothing