How PilotMyX Measures X/Twitter Account Performance

Scope

PilotMyX analyzes the authorized X/Twitter account connected by its owner. The product works from the account history included with the selected plan and from metrics X makes available for those items.

The methodology is designed to answer account questions such as:

• What changed between two periods?
• Which posts or replies contributed most to a result?
• Is the result broad or concentrated in a few outliers?
• Which topic, opening, or timing pattern deserves another test?

It is not a reconstruction of the X recommendation system and does not claim causal attribution.

Content definitions

Post refers to an original top-level post or supported quote published by the connected account.

Reply written refers to content published by the connected account in response to another post.

Replies received refers to responses other accounts leave under a post. This is different from the replies written by the connected account.

Reply count is a metric attached to a post. It records the number of responses to that post and is not a count of replies written by the account during the period.

Reposts are excluded from the imported author-content cohorts because the connected account did not author the reposted text.

Metric definitions

PilotMyX stores and displays the metrics returned for the connected account. Availability can differ by item age, authorization, and current X API behavior.

The official X metrics reference defines the main fields, including:

• Impressions: times a post appeared on a screen. This is not a unique-person count.
• Likes: likes recorded for the item.
• Reposts: reposts without quote commentary.
• Quotes: reposts that include commentary.
• Replies: responses received by the item.
• Bookmarks: saves recorded for the item.
• URL clicks and profile clicks: private metrics available for authorized owned content when X provides them.
• Engagements: the total engagement field returned for authorized owned content when available.

X states that non-public, organic, and promoted metrics have a 30-day availability limit through the API. An older item may therefore have fewer fields than a recent item. PilotMyX does not replace an unavailable value with an estimate.

Date cohorts

A selected period contains items published during that period. Their displayed metrics are the latest values synchronized for those items, not necessarily interactions that happened only inside the selected dates.

This distinction explains why a PilotMyX total may differ from a native X report that groups activity by the day the impression or engagement occurred.

The immediately preceding comparison period has the same length as the selected period. A 30-day current period is compared with the 30 days directly before it.

Posts and replies

Users can analyze original posts, replies written, or both. When both are selected, the account total is the sum of the eligible authored items in the period.

The separate views remain important because the two content types can play different roles and have different sample sizes. A combined total should not be used to imply that a top-level post received the performance of a reply, or the reverse.

Engagement rate

When impressions are available and greater than zero, engagement rate is calculated as:

engagements ÷ impressions × 100

The rate is only as comparable as the underlying cohorts. Changes in topic, content type, item age, promotion, sample size, or metric availability can alter the result.

PilotMyX reports the returned numbers. It does not label a universal “good” engagement rate because the useful baseline is the account’s own comparable history and goal.

Timing analysis

Timing analytics groups eligible authored items by publication weekday and local publication hour.

Items must be at least 24 hours old before they enter timing analysis. This gives very recent content time to collect early activity, but it does not make every item equally mature. A seven-day-old post can still have had more time to accumulate lifetime metrics than a two-day-old post.

Each timing result should be read with its sample size. One high-performing item in a cell is an observation, not a proven best time.

Freshness

Metrics can change after publication. PilotMyX shows the synchronization time so the user can judge whether a close comparison needs fresher data.

Paid connected accounts receive a daily synchronization. A manual refresh can be requested when a newer view is required between scheduled updates, subject to the product’s current refresh rules.

Outliers and concentration

Totals and averages can hide concentration. A useful account analysis checks:

• the share of the period produced by the top one, three, or five items;
• the median alongside the average when the distribution is skewed;
• the result after removing the largest outlier;
• whether a pattern repeats across topics, dates, and content types.

PilotMyX does not automatically declare an outlier irrelevant. A viral result may contain a useful lesson. The concentration check prevents that one result from silently becoming the baseline for every future decision.

What the data can and cannot support

The data can support statements such as:

• “Three posts on this topic produced above-median profile visits in this period.”
• “One reply generated 60% of reply impressions during the month.”
• “The current period had lower reach with a similar engagement rate.”

It cannot by itself support statements such as:

• “This phrase caused the algorithm to boost the post.”
• “Tuesday at 10:00 a.m. is the universal best time to post.”
• “The account is shadowbanned.”
• “A reply will produce the same result if rewritten as a post.”

Use the observation to design the next test, then compare the new result with a suitable baseline.

Reproducibility checklist

When you compare posts and replies on one account, record:

1. the account or account type;
2. the exact date range;
3. the data freshness timestamp;
4. the number of posts and replies included;
5. the metrics and formulas used;
6. exclusions and missing-data rules;
7. the largest outliers;
8. why the result should not be generalized to other accounts.

The posts versus replies guide turns that checklist into a practical account review without presenting one account as a market benchmark.