# Read TikTok shares and saves for a clear content decision

Use shares and saves as separate actions for a stated content decision to guide your creative review. An Actor is a hosted data collection program. A Task is a saved input preset. The Actor collects source rows, and you do the analysis.

## Collection setup

Run the [TikTok Metrics Scraper](https://apify.com/chronometrica/tiktok-metrics-scraper) on build 0.1.36 using mode knownUrls. Replace the whole five-video sample list with one intended known public video in startUrls, set maxItems to 1, and set includeProfileMetrics to true to collect best-effort current public profile facts via the [TikTok Engagement Metrics Tracker Task](https://apify.com/chronometrica/tiktok-metrics-scraper/examples/tiktok-engagement-metrics-tracker).

The native fields are inputUrl, postUrl, postId, profileId, profileHandle, profileUrl, publishedAt, videoDurationSeconds, views, likes, comments, shares, saves, reposts, profileFollowersCount, postMetricStatus, profileMetricStatus, status, statusReason, and scrapedAt. Missing metrics return null rather than zero. Per-metric availability and precision are tracked inside postMetricStatus.

Run summary statistics are stored in the run summary, separate from individual dataset rows. Its exact summary fields are providedUrls, contentUrls, inputUrls, resolvedUrls, metricRows, noPublicMetricRows, blockedRows, failedRows, and statusCounts. Use both the saved native dataset rows and this run summary to reconcile your intended source target. The single-video control test verifies public counters and profile IDs only, where saves and reposts returned exact and several other counters returned as likely rounded. This control does not prove derived rates, private outcomes, or fake-follower filtering.

## Eight steps to compare shares and saves

1. State one decision. Choose whether a practical tutorial receives more save actions than a quick discussion clip. Pick known videos with a similar topic, age, and observation time. This comparison is descriptive and does not prove content causality.
2. Use the current saved engagement Task with one intended known video and maxItems set to one for a pilot run. Inspect matched IDs, status, exact shares, saves, views, and postMetricStatus. Preserve null versus zero values. The one-video known-URL control succeeded with matched post and profile IDs, public views, likes, comments, shares, saves, reposts, profile follower count, and field precision flags. Several counts were likely rounded, while saves and reposts were exact. This proves current public counter support, but it does not prove retention, unique viewers, or private reach.
3. Keep raw counts and precision labels. A likely-rounded views denominator makes a derived rate approximate. Do not print extra decimal places. Keep missing counts outside computed ratios and note their availability reason.
4. Define each descriptive ratio plainly. Share actions per 1,000 public views equals shares divided by views times 1,000. Save actions per 1,000 public views equals saves divided by views times 1,000. Use only positive available views. Zero or unknown views mean the ratio is undefined, not zero. These are actions per platform count, not percentages of unique viewers.
5. Keep the two actions separate. Do not add shares and saves together and call the sum unique engaged viewers because one person can do both actions. Reposts are a distinct returned counter. Do not silently equate them with shares. Source changes can make a count revision look like a negative change.
6. Build a small comparison table with video age, topic, shares, saves, views, precision, ratios, and collection time. If observation times differ, state that before ranking. Public lifetime totals are not a fixed campaign-period report unless repeat observations are properly scoped.
7. Inspect the actual content and write a cautious next experiment. Save-heavy tutorials might justify testing a clearer reusable checklist. A proposed experiment is not proof of viewer intent, an algorithm rule, or a forecast. Owned clicks and sales data are needed for conversion questions.
8. After the test, collect comparable observations and separate your hypothesis from the result. Do not declare an action universally most important. Tie it to your stated decision and data quality.

## Original fictional exact example

The following examples are original fiction and do not represent real product results.

- Fictional video A has 10,000 views, 80 shares, and 300 saves. It has 8 share actions and 30 save actions per 1,000 views.
- Fictional video B has 10,000 views, 200 shares, and 40 saves. It has 20 share actions and 4 save actions per 1,000 views.
- Fictional video C has 5,000 views, shares as null, and 0 saves. Its share ratio is unavailable and its save ratio is 0 actions per 1,000 views.
- Fictional video D has 0 views, 0 shares, and 0 saves. Both ratios are undefined because the denominator is zero.

| Video | Age | Topic | Views | Shares | Saves | Precision | Share Ratio | Save Ratio | Check Time |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Video A | 7 days | Tutorial | 10,000 | 80 | 300 | exact | 8 actions / 1,000 views | 30 actions / 1,000 views | 2026-10-13 |
| Video B | 7 days | Discussion | 10,000 | 200 | 40 | exact | 20 actions / 1,000 views | 4 actions / 1,000 views | 2026-10-13 |
| Video C | 3 days | Tutorial | 5,000 | null | 0 | exact | unavailable | 0 actions / 1,000 views | 2026-10-13 |
| Video D | 1 day | Update | 0 | 0 | 0 | exact | undefined | undefined | 2026-10-13 |

The exact precision label applies only to available fictional counts. Video C has null shares, meaning its precision and state are unavailable. You must check the public postMetricStatus separately for each individual metric.

Video A is more save-heavy and Video B is more share-heavy under this descriptive table. This does not prove Video A viewers intend to buy or Video B will go viral. A combined count of 380 actions does not equal 380 unique people. If Video A used a rounded views count like 10K, the ratio would be approximate.

## Next action

To test your own content metrics, run the [TikTok Metrics Scraper](https://apify.com/chronometrica/tiktok-metrics-scraper) with the [TikTok Engagement Metrics Tracker Task](https://apify.com/chronometrica/tiktok-metrics-scraper/examples/tiktok-engagement-metrics-tracker) using a known public video URL and maxItems set to 1.

## Sources

- [Reddit TikTok Lounge Discussion](https://www.reddit.com/r/TikTokLounge/comments/1oqmbsg/what_is_the_most_important_metric_inside_tiktok/)
- [TikTok Metrics Scraper Actor](https://apify.com/chronometrica/tiktok-metrics-scraper)
- [TikTok Engagement Metrics Tracker Task](https://apify.com/chronometrica/tiktok-metrics-scraper/examples/tiktok-engagement-metrics-tracker)