Chronometrica vs HarvestAPI for Public LinkedIn Comment Exports

Compare the collection scope first

An Actor is a hosted collection program. Both products are Actors on Apify. For this comparison, the useful question is whether the documented collection scope matches your research question. The table summarizes the checked product facts. Chronometrica documentation, HarvestAPI documentation

Do not choose by exported row count alone. A row might represent a direct comment, a reply, or a collection outcome. Those units answer different questions. First decide what one observation means. Then compare the outputs that contain that observation.

Why Chronometrica fits a public author sample

Chronometrica accepts 1–100 public LinkedIn post/feed URLs. It removes duplicate inputs that refer to the same supported post identity. It also saves unique public comment IDs. These features help you avoid counting the same source or saved comment twice. Chronometrica documentation

These checks serve different purposes. Removing duplicate post inputs protects the source list. Unique comment IDs protect the comment observations. Neither turns a comment count into a people count. One author can write several comments, so the author list needs its own check.

The optional `includeAuthorDetails` setting requests public details for unique saved authors. It defaults to `false` and can add a separate author-detail billing event. Enable it when those details serve the research question. Chronometrica documentation

The strongest fit is a study that can use an explained sample of public top-level comments. You can retain the observed comments, build a separate author list, and keep the collection limits with the export. That gives a reviewer a clear basis for deciding what the sample supports.

It does not establish that every discussion participant was observed. It also does not make a public profile a qualified sales lead. Those conclusions need further evidence.

When reply needs change the choice

HarvestAPI accepts post and comment URLs. For post inputs, it documents up to five nested replies per root comment, inside that root result. For comment URLs, `scrapeReplies` can return replies as separate results. Its `postedLimit` and author-detail modes can also affect the choice. Check their documented settings against your needs. HarvestAPI documentation

Align the root/reply unit before comparing row totals. A root is a top-level comment. Replies attached to that root add text observations, but they do not necessarily add distinct people.

Chronometrica does not collect full reply threads. If your requirement is every comment and reply in a discussion, it does not meet that requirement. Removing a numeric limit cannot change its collection scope. Chronometrica documentation

Define the people you need to observe

The buyer need behind this comparison concerns finding profiles of people who comment on or like the buyer’s LinkedIn posts. The supporting evidence is a saved public search snippet, rather than a full discussion transcript. It establishes a research need, not a verified collection result. Public request

Before selecting an Actor, write one sentence that defines your required population: the group of people your study must cover. For example, you might need authors of visible top-level comments on three selected posts. A different study might need reply authors, likers, or all three groups.

Keep those groups separate. A comment export cannot establish a complete liker list. A top-level sample cannot establish the reply-author population. If the missing group is required to answer the question, stop and change the collection plan.

Also define the claim you intend to make. “These authors appeared in the collected sample” needs less evidence than “These are all participants.” A smaller claim can still be useful when it matches the available evidence.

For more detail on this distinction, read Can a Public LinkedIn Comment Export Represent the Full Discussion?.

Start with a small, checked input

A Task is a saved input preset for an Actor. Open the public author-details Task, then inspect its actual settings before you run it.

The saved evidence confirms that the Task page was publicly accessible. That page check does not establish its current saved input. Use the preset as a starting point and check these settings yourself:

1. Replace the saved URLs with your intended public post URLs.
2. Set `maxComments` to a small limit for the first check.
3. Set `includeAuthorDetails` to `true` if you need public author details.
4. Review the remaining settings before starting the run.

This example uses fields and values that passed a local check against the saved public JSON Schema. A JSON Schema defines allowed input fields and values. This check confirms the input format; it does not confirm that the source will be available during a run. Chronometrica input documentation

{
  "startUrls": [
    "https://www.linkedin.com/posts/timferriss_how-to-become-dramatically-more-productive-activity-7386884213343428608-wrF3"
  ],
  "maxComments": 10,
  "includeAuthorDetails": true
}

Replace the example URL with your chosen public source. It is a documented example target, not evidence of a new collection result.

The checked schema allows 1–100 entries in `startUrls` and an integer from 1–1,000 for `maxComments`. The latter is a per-post limit. A setting of 10 requests a bounded sample; it does not promise ten saved comments. Chronometrica input documentation

Use the first sample to check whether the saved text and author identifiers support your intended analysis. Increasing the volume before this check can produce a larger export that still answers the wrong question.

Read coverage before you describe the result

Chronometrica separates comment rows from input and coverage outcomes in `OUTPUT`. Read the outcome for each input before you summarize its comments. The coverage record and stop reason help explain why collection ended. Chronometrica output documentation

Follow this small acceptance check:

1. Confirm which input post each result belongs to.
2. Identify saved comment rows through their public `commentId` values.
3. Match authors through `authorUrl` or `authorId` where those values are public.
4. Keep records with missing identifiers unresolved.
5. Read the input’s coverage outcome and stop reason.
6. Decide which claims the observed sample can support.

Do not treat an input status record as a comment. Likewise, do not compare a displayed discussion total directly with saved top-level rows. They can refer to different units and different visible portions of the discussion.

The `collectAllAvailable` setting removes the numeric comment limit. Collection can still stop at the public source, a charge limit, a time limit, or a failure. It does not prove that the public page exposed every comment. Chronometrica collection limits

If your original requirement demands every comment, do not rename an available public sample as a complete export. Chronometrica does not meet an all-comments requirement when that requirement needs content beyond the public top-level source.

Public author fields, dates, and reaction counts can be unavailable. Preserve those unknown values. A missing reaction count is not evidence of zero reactions, and a missing date does not establish when the comment appeared. Chronometrica output documentation

A fictional decision with checked arithmetic

The following example is fictional. It is not an Actor test, a live dataset, a buyer quotation, or a measured cost result.

A researcher collects a preview from one public post. The preview contains 10 top-level comment rows from 7 distinct public authors. Added author details resolve for 5 of those authors. The other 2 remain unresolved.

The researcher reports three separate findings:

• The sample contains 10 observed top-level comments.
• Those comments come from 7 distinct observed authors.
• Added details are available for 5 authors; 2 remain unresolved.

The arithmetic is simple: 7 observed authors minus 5 authors with added details leaves 2 unresolved authors. The 10 comment rows do not represent 10 people. Nor do the 2 unresolved detail outcomes mean that those authors did not exist in the comment sample.

Suppose the research question also depends on 8 replies below those comments. The top-level sample cannot answer it. The researcher assesses HarvestAPI’s reply modes instead. HarvestAPI documentation

Those possible 8 reply rows are a separate unit. They cannot be counted as 8 new people without checking their authors. Some could come from people already in the top-level sample, and several could come from the same person.

If the original question concerns only the 7 observed top-level authors, the bounded sample may be enough. If it concerns everyone who took part, the same sample is insufficient. The decision follows the required population, not the size of the file.

Export four separate records

Keep four tables or clearly separated record groups. Each has a different purpose. This structure helps another person inspect your work without guessing what each row means.

The first table contains saved comment observations. Keep the source post and `commentId` with each comment. Retain the public author identifier when available. Preserve missing values so later analysis can distinguish unknown information from an observed zero.

The second table contains unique observed authors. Build it from the saved comments using public identifiers. Count repeated comments from the same identified author once in this author table. Keep missing identities unresolved; do not assume that blank values all belong to one person or to different people.

The third table contains author-detail outcomes. Keep the link between the observed author and any added details. An author with no added details can still be an observed commenter. Treating detail success as the author count would discard people already present in the evidence.

The fourth table contains per-post public coverage. Retain each input’s outcome, stop reason, and collection context. Keep the post source and collection time with the export. These records explain how to read the sample; they are not comment measurements.

This separation also helps with later interpretation. For example, you can inspect public author information without silently changing the number of observed comments. If your next step concerns author roles, use the comment-author roles guide and retain unresolved cases.

Decide whether to continue after the sample

Before a larger run, check whether the first export answers the stated question. Confirm that you can trace each saved comment to its post, distinguish repeat authors, and explain missing details. Confirm that you understand each input’s coverage record.

Then decide whether more observations would improve the study. A larger top-level sample can add evidence about observed top-level authors. It cannot repair a mismatch between that scope and a question about reply authors or likers.

Keep the acceptance rule specific. For example, a study might accept unresolved author details while still requiring a clear source and coverage record for each post. Another study might require an author identifier for every observation it includes in a people count. These are research choices. State them before summarizing the result.

Do not use collection success as proof of sales value. A public author link establishes a route to a public profile. Lead qualification needs separate criteria and supporting evidence.