Instagram Content Performance by Pillar and Saves

Use the Content Performance Analyzer for Instagram with counts you supply. The Instagram context does not connect your account, inspect posts, or collect insights automatically.

Give each Instagram pillar a reader problem

A useful Instagram pillar names a recurring subject that your audience can recognize, not simply a style of post. “Small space plant care” gives an editor a concrete area to investigate. “Educational” may be too broad if it includes watering, choosing a pot, delivery questions, and room decoration. Start with the decisions you want to make after the review. If two labels would lead to the same next assignment, they may not need to be separate. If one label hides several unrelated customer problems, it may need a clearer boundary. Keep the vocabulary manageable rather than giving every individual post its own category.

This guide uses a fictional balcony plant shop called Mossline Studio. Its editor compares two pillars: Care and Styling. Care answers practical questions about maintaining a plant in a small home. Styling explores where plants sit and how arrangements look. The initial review contains three posts per pillar, all from one Instagram account with organic traffic and a declared observation age of 72 hours. These are invented records for learning the method, not a claim about typical shop results. The editor wants to decide which pillar deserves a small additional production round aimed at making genuinely reusable reference posts.

Keep topic and format separate from the start. A care demonstration can be a reel, while a care checklist can be a carousel; both still belong to Care if that is the agreed subject boundary. Label a post according to its main reader problem before looking at its results. Otherwise an attractive result can tempt you to invent a flattering category afterward. The analyzer accepts your labels but does not read the image, caption, or video to verify them. For a broader plan that connects social subjects with website resources, the Content Hub Planner offers a separate manual workspace.

Collect saves without inventing the reason behind them

A save count records an action reported by the source you use. It does not tell you whether someone returned, followed the instructions, bought a plant, or recommended your shop. A practical reference may be worth reviewing through saves because keeping something for later is relevant to its intended role, but that is your editorial hypothesis. Do not turn it into a verified explanation of audience intent. Similarly, a like count and a share count describe different actions. Choose the metric that matches the immediate question, then leave other counts available for a separate reading rather than treating them as interchangeable evidence of usefulness.

Mossline supplies Care save counts of 18, 24, and 30, with corresponding reach values of 900, 1200, and 1500. Styling has 20, 28, and 36 saves, with reach of 2000, 2800, and 3600. The editor records the native count labels and collection details in a separate working sheet. The tool has no source-verification field and cannot confirm that a number was copied correctly. When the native surface does not supply a count for a selected post, leave it missing. Do not estimate a save count from visible likes, assume every format exposes the same metrics, or borrow an account total for one post.

Use a consistent collection routine so the declared 72 hour age means something in practice. A snapshot taken shortly after publication is not comparable with a mature lifetime count simply because both were entered on the same afternoon. The observation age setting is a declaration, not an adjustment formula or a check against Instagram. Keep paid and organic collections separate when the decision needs that separation; a mixed label only describes a blend. Native metrics and availability can vary, so preserve the meaning of the source you actually used instead of assuming that a familiar metric name guarantees an identical definition everywhere.

Compare save volume with saves per reach

Start by selecting saves as a raw count objective and grouping by topic. In the fictional collection, the Care median is 24 saves and the Styling median is 28. That says a typical supplied Styling post has the higher save count in this small set. It does not show why: Styling also has higher supplied reach values. If the production question is about the count associated with an ordinary post from this account, the raw view is useful. If the question concerns the relationship between recorded saves and recorded reach, switch deliberately to saves per reach rather than calling the raw comparison exposure-adjusted.

The Care posts each have 2 saves per 100 reach: 18 divided by 900, 24 divided by 1200, and 30 divided by 1500, each multiplied by 100. The Styling posts each have 1 save per 100 reach. Their group medians are therefore 2 and 1 for that rate objective. The ordering differs from the raw save ordering because the question has changed. Neither view cancels the other. Styling has more saves per typical supplied post, while Care has more save actions relative to the supplied reach denominator. Both observations can fit the same collection without implying a contradiction or a universal recommendation.

Read the rate as actions per 100 units of supplied reach, not as the probability that a unique person saved a post. The tool does not identify people, deduplicate actions, or verify who had an opportunity to save. It calculates each eligible post rate and then describes the distribution of those rates. It does not pool all saves and reach into a single group rate. Do not add reach across the six posts to claim a unique Instagram audience, either. Repeated exposure across posts makes that total unsuitable for a unique audience statement, even if each individual native reach figure has its own distinct-account definition.

Notice uneven coverage before reading the ranking

Coverage explains which posts actually contribute to the selected comparison. A supplied save count of zero is eligible for the raw saves objective, while an empty save field is not. For saves per reach, a known numerator and a positive known reach are both necessary. A post with zero saves and reach of 800 contributes a rate of zero. A post with 15 saves but no reach contributes no rate, although its save count remains useful for a raw count review. Check eligible and missing counts again whenever you change objectives, because the comparison may now represent a different subset of the inventory.

As a separate coverage exercise, imagine Mossline prepares four additional Styling records. Two contain both saves and positive reach, one contains saves without reach, and one has positive reach but no save count. A saves per reach review of that four-record set has two eligible posts and two missing values. It is not a four-post rate comparison merely because four records appear in the list. If Care has complete data while Styling does not, the missing posts may matter to interpretation. The tool cannot know whether those records would resemble the observed ones, so describe the comparison as based on available eligible values.

Choose the minimum post threshold before treating a category as ready for discussion. Mossline uses four as a working reminder, so both three-post pillars in the original six-post example remain flagged. The groups still appear; the flag does not erase them or calculate uncertainty for them. Lowering the setting to three changes the reminder, not the evidence. There is no significance test or quality certificate behind the threshold. A sensible response is to collect another comparable round, not to search for a setting that makes every warning disappear. Keep unusually thin categories visible in the conversation when they represent a necessary audience need.

Separate a reusable pattern from one exceptional post

Instagram reviews often begin with a memorable post that dominates the team’s recollection. A median helps stop that one result from speaking for an entire pillar, while the mean and range reveal that the collection may be uneven. Read all three rather than choosing whichever summary supports your preferred subject. The original Care save values of 18, 24, and 30 are fairly close together in this invented example. Their median and mean are both 24, with endpoints of 18 and 30. That compact description is more informative than calling the pillar consistent without showing what was actually observed or how few posts were supplied.

Now consider a separate extension in which a fourth Care post records 180 saves. The four values become 18, 24, 30, and 180. Their median is 27, their mean is 63, and the range runs from 18 to 180. The much larger mean is not an error; it reflects the unusually high fourth value. Inspect that post outside the tool before building a new plan around it. Perhaps it was attached to an event, mentioned by another account, or collected under a different condition. Those possibilities are questions for your review, not explanations the analyzer has detected from the counts.

Do not delete an exceptional result just because it makes the chart untidy. First establish whether the record belongs to the declared scope and whether its numbers were copied correctly. If it belongs, keep it and discuss how much the practical decision depends on it. If it does not belong, prepare a properly scoped collection and record why it changed in your own notes. The index is also relative to the current filtered dataset median, so it can move when the collection changes. It is not an enduring score attached to Care, and an unavailable index at a zero baseline is not a negative judgment of the content.

Check whether the pillar is really a format difference

A subject comparison becomes harder to interpret when presentation is distributed unevenly. Suppose all three Care posts are carousels and all three Styling posts are reels. The difference in saves per reach then accompanies both topic and format. The analyzer cannot decide which part matters, and it does not adjust for duration, cover design, audience composition, or delivery. This is not a reason to abandon the review. It tells you what is missing from the evidence. Before announcing that Care is the stronger pillar, ask whether the same subject has been presented in more than one format under reasonably comparable collection conditions.

Use the format filter when you have the necessary combinations. Grouping by topic while filtering to Carousel compares only the supplied carousel posts. Grouping by format while filtering to Care compares presentations within that pillar. Both filters select records before the baseline is calculated, so their results answer narrower questions than the original review. If only one topic survives the filter, you have a summary of that topic rather than a comparison against an absent alternative. Do not interpret an empty group as zero performance, and do not expect the tool to fill in a Care reel that your team has never supplied.

Mossline therefore plans a future collection with both Care and Styling represented in the same chosen format. Its editor keeps the pillar definitions fixed and chooses a narrow practical question for each assignment. A care carousel might explain how to check a dry pot; a styling carousel might compare two positions beside a window. They need not be visually identical, but the editorial team should understand the remaining differences. The guide to comparing content formats develops that separate review question. This analyzer still processes labels and counts only; it does not judge whether the finished creative executions were truly comparable.

Choose the next pillar test without rewriting the past

Turn the review into a small production decision with an explicit reason. Mossline’s observation is that Care had the higher median saves per reach, while Styling had the higher median raw saves in the original six-post collection. The editor chooses to test three new Care reference posts because reusable instructions are the immediate editorial priority. That is a decision made by the team, not a recommendation generated by the tool. The original Styling results remain part of the record. A follow-up can explore one promising question without declaring that every other pillar should disappear or that the selected subject has been proven superior.

Write the assignment so another person can recognize success in the production process before results arrive. For example: explain one care decision, show the relevant condition clearly, and keep the same planned observation routine. Choose the primary metric in advance and keep the supporting raw count view available. Do not add a new label such as “high intent” after a post receives many saves. The data cannot establish that intent, and retrospective labeling weakens the usefulness of the comparison. Keep production time in a separate note too. The analyzer has no effort field and cannot tell you whether an elaborate reference post justified its additional editing hours.

Reserve space for an essential but less saved topic when your audience still needs it. Delivery information, temporary availability, or a correction may serve a role that is not well described by saves. A pillar review is not a mandate to publish only content that encourages one action. Use comments as a separate objective when the question is about recorded discussion, and review the actual conversation outside the tool when meaning matters. The numbers do not distinguish a helpful question from confusion. SMTasker supports broader account routines separately, so keep the reasoning that shapes your approved Instagram plan visible to the people responsible for it.

Keep a recoverable record of each Instagram review

After entering or importing your collection, check the account label, traffic, observation age, grouping, objective, and filters before analyzing. Quick entry suits a handful of carefully checked posts; the template CSV suits a prepared collection. A native export may need manual preparation because the analyzer does not promise automatic column mapping. CSV replacement retains your current context, so importing a new collection does not establish that its counts match the old declaration. Use the downloaded template rather than renaming unrelated totals until they fit. Unique IDs let you return to the right original record without putting full captions or personal information into the dataset.

Save the full JSON project when you need to continue editing later. The review CSV is not that project and is not the input template, so keep those files distinct in your handoff. There is no autosave or native Instagram sync. If you change a count, label, or setting, run the analysis again before sharing the resulting review. Keep the collection date and the source meaning in your own working record because the supplied counts remain declarations. The tool does not refresh old snapshots, recognize a corrected native total, or know that a post has changed after you entered its numbers.

Read the complete export scope before sending a filtered review to someone else. Copy, CSV, and PDF preserve raw records outside the filters as part of the review evidence; the PDF also attaches the project JSON and full review CSV. A Care-only chart does not mean Styling records have been removed from the file. Check labels and counts across the entire project, then share only with an appropriate recipient. Return to the Instagram comparison workspace for the next collected round. Finish with one precise observation, one chosen follow-up, and enough saved context for another editor to understand how the decision was reached.

Questions about Instagram Content Performance

Continue with another performance guide

Apply the comparison in the Content Performance & Topic Analyzer, or choose another guide below. Discover more resources in all free tools.