Skip to content
Krevaya
Measurement11 min read

How to do a social media competitor analysis in an afternoon

Review five accounts over 30 days, compare posts against each account's baseline, and turn the findings into two social media experiments.

Spot the pattern. A brass loupe magnifies one of five smoked-glass profile cards; inside the lens, a single electric-lime bar rises far above the card's ivory median line, and an amber tab marks a second card.
Original editorial illustration for Krevaya Fieldnotes, generated with AI assistance.

The short answer

Spend one afternoon reviewing five relevant accounts on one platform across 30 days. Record a consistent sample in one sheet, compare posts with each account's own median for similar formats and ages, and inspect the context behind unusual results. Turn the strongest observations into two original experiments with a metric, owner, and review date.

What to take away

  • Review five relevant accounts on one platform using a dated, consistent sample from the last 30 days.
  • Compare visible metrics with each account's own median for similar formats and ages, marking missing or insufficient data.
  • Turn two observed patterns into separate experiments with original material, a fixed measurement age, an owner, and a review date.

Set a four-hour limit and one question

A social media competitor analysis is a structured review of how relevant accounts publish and how audiences respond. For a small team, the useful output is a decision about what to test next.

Give the review a fixed scope: five accounts on one platform, posts published over 30 days, and one sheet. Start with a question your next month of content could answer, such as which ways of explaining your service invite useful customer questions.

Reserve four hours and divide the work:

  • 30 minutes: choose the question, platform, and five accounts.
  • 90 minutes: collect a small, consistent sample in one sheet.
  • 60 minutes: calculate baselines and inspect the strongest patterns.
  • 60 minutes: choose two experiments, assign owners, and reserve publishing slots.

These are working limits for the review. If an account produces more material than you can inspect, use the sampling rule below. Finish with a publication slot and a review date for each experiment.

Choose five accounts your audience would consider

Choose accounts that help explain a buying decision or a content problem. Use this mix as a starting point:

  • Three direct competitors: businesses serving similar customers with a comparable offer.
  • One adjacent account: a business or creator helping the same audience with a related problem.
  • One format reference: an account serving a relevant audience through a format you could realistically produce.

Keep all five on the same platform for this afternoon. Write one sentence explaining each choice, including the audience overlap and what you want to learn. A large account can be useful when its readers and subject matter fit.

Before opening individual posts, note who the account says it helps, what it sells, and the action its profile asks visitors to take. This makes an educational post or sales pitch easier to judge in context.

Follower totals describe size, but one snapshot cannot establish follower growth. They also do not tell you which post was unusually strong for its author. Replace inaccessible or inactive accounts before collecting data, and keep your own account as the later test destination.

Collect 30 days of evidence in one sheet

Use a free spreadsheet and what is visible on a public profile. Record the start and end dates for the last 30 complete days, plus the date you collect the numbers. Give each sampled post one row.

Log every permanent post if an account published 12 or fewer in that window. Otherwise, select 12 posts evenly spaced in date order, including the oldest and newest. Choose by date before looking at performance, and check publication dates on pinned posts.

This gives you at most 60 rows. Label the selection as a sample, and keep the same rule next month. A sample of standout posts alone cannot tell you what is normal.

What to record in the competitor sheet
Field group Record Decision it supports
Account Profile link and reason for inclusion Whether the comparison fits your audience
Post Post link, publication date, capture date, and age Whether the observation can be compared fairly
Content Reader problem, format, opening, and call to action Which element you could test
Measurement Metric name, visible count, and missing-data note How the post compares with its own account
Context Relevant comments, collaboration, promotion, or timely event What else could explain the response

A 12-post sample cannot establish how often an account published. If cadence matters to your question, count all visible posts in the full window separately and record that count. Otherwise, leave frequency out of your conclusions.

Record unavailable numbers as "not visible." Do not assume access to a competitor's reach, saves, shares, audience details, or sales. Views are not a universal public field either; record them only where shown, with the platform's metric label.

Keep your own analytics separate from this public evidence. For example, YouTube Analytics provides channel and individual video reports in YouTube Studio, including watch time and audience information. Those reports describe your channel, not an equivalent view into a competitor's account.

Compare posts with each account's own median

Choose one metric that is consistently visible within a comparison group. Where both counts are available, likes plus comments gives a simple interaction total. If only one suitable count is visible, use that metric throughout the group and label it clearly.

Group posts by account and format. For this first review, exclude posts under 48 hours old from calculations, then separate posts aged two to seven days from those aged eight to 30 days. These age bands are practical limits for this worksheet, not platform benchmarks.

Calculate the median for each group, then divide a post's count by that median. A spreadsheet's MEDIAN function finds the middle value, averaging the middle pair when needed. The NIST explanation of measures of location explains why extreme values affect the mean more than the median.

Write the calculation as: post's visible count ÷ median for comparable posts from that account = baseline multiple. Keep the metric, sample size, format, and age band beside it. Multiples help prioritize posts to inspect; they do not rank the quality of whole accounts.

Use at least five comparable observations as a practical floor, not a claim of statistical reliability. If there are fewer, write "too little data" and inspect the content without calculating a multiple. If the median is zero, keep the raw counts and skip division.

A snapshot has an age problem
Counts collected today reflect different amounts of time since publication. Age bands reduce that mismatch but cannot reconstruct each post's first 48 hours. Treat these multiples as clues, then start recording future comparisons at a fixed age.

Do not compare a likes-only multiple with a likes-plus-comments multiple as if they measure the same response. Keep definitions consistent whenever you compare patterns across accounts.

Read the pattern behind a strong post

Open a promising post beside an ordinary post from the same account and format. Read both before deciding which element deserves a test.

  • Reader need: what specific problem does the opening name?
  • Delivery: how does the post explain, demonstrate, or organize the answer?
  • Response: what questions, objections, or examples appear in visible comments?
  • Context: is there an announcement, collaboration, giveaway, or other reason attention might differ?

Look for a pattern that appears more than once. If an ordinary post uses the same format, the format alone is a weak explanation. If nobody covers a topic, that is an unanswered question about demand. Check that gap against questions from your own customers before planning a post.

Record uncertainty about distribution. YouTube's explanation of engagement counts says paid advertising can contribute to public view counts, so a visible total alone does not establish organic performance. Do not label an unexplained spike as proof of a successful content tactic.

Read comments for the language of the problem, too. A practical question from a possible customer may be more relevant to your experiment than a string of compliments. Record the question without assuming you know the commenter's buying intent.

A worked example: the smaller total deserves a closer look

Imagine a two-person bookkeeping business reviewing five Instagram accounts that serve independent design studios. The numbers are invented to show the method. They are not Krevaya customer data.

The team finds seven comparable carousel posts from one account, all in the eight-to-30-day age band. Likes and comments are visible on each, and their combined totals are 24, 30, 36, 40, 44, 58, and 120. The median is 40, so the strongest post is 120 ÷ 40 = 3 times its sample baseline.

A larger account has comparable totals of 320, 360, 390, 400, 440, 480, and 600. Its median is 400, making the strongest post 600 ÷ 400 = 1.5 times its baseline. The team inspects both, with particular interest in what made the first post unusual for its author.

The first post opens with an overdue-invoice problem and gives three practical steps. Its visible comments include questions about wording a reminder; the account's ordinary posts open with broad monthly tips. That suggests a testable opening and structure, while differences in age and distribution remain possible explanations.

The team queues two experiments: name a specific client problem in the opening, then test a three-step layout. It changes one element at a time, using its own examples and expertise. Neither multiple establishes that those changes caused the response or will work for this business.

Turn findings into tests, then review monthly

Choose experiments you can produce within your normal workload. Give each one a single change, a primary metric, an owner, and a decision date. Queue the second experiment until the first has been reviewed.

Start the first experiment with three test posts alternating with three usual posts on similar subjects. Keep the offer, format, and publishing conditions as consistent as the chosen change allows. This is a small learning exercise, so treat its outcome as a reason to repeat, revise, or stop.

Measure every post at the same age, such as 48 hours. Compare with your own recent median for that format only when those earlier posts were measured at the same age. If you lack those records, build a baseline with the usual posts in this batch; the posting-time guide explains fixed-age comparisons.

Choose a metric tied to the question, such as relevant replies or visits to a linked page. For website traffic, Google's campaign URL guidance explains how UTM parameters identify referring campaigns in Analytics. Use distinct content labels for test and comparison links.

Competitor experiment note

Observed pattern and source: (post links)
Alternative explanation: (what else could explain it)
Our hypothesis: (why our audience might respond)
Single change: (opening, structure, example, or call to action)
Comparison posts: (matching usual posts)
Primary metric and reading age: (one measure, fixed time)
Baseline and useful improvement: (what would justify repeating)
Publishing slots: (dates for test and comparison posts)
Owner and review date: (name, date)
Decision: repeat / revise / stop, because (evidence)

Put the publishing slots and measurement reminders into your content calendar. Set the useful improvement before results arrive, based on the effort the change requires. Record what happened even when the test is inconclusive.

Spend 15 minutes each week checking new candidates alongside your weekly trend review. Repeat the full analysis monthly with the same five accounts and a fresh 30-day window. Reconsider the account list quarterly, or when your audience or offer changes.

How Krevaya fits this routine

Krevaya can handle ongoing monitoring after the first review. Account Spy tracks Instagram, Threads, and YouTube accounts and alerts when a post reaches about three times that account's usual engagement. Alerts arrive in the app and through the competitor.breakout webhook.

Account Spy is included from the Plus plan at $29 a month. Its alerts use the account's own recent baseline; keep that product baseline separate from the manual median calculated in this guide. An alert gives you a post to inspect and a reason to return to the sheet.

Analytics & Benchmarks shows your engagement beside creators you track, and post and account insights flow back from Meta within the hour. Use those numbers to check your own experiments and keep the next review grounded in measured results.

Track the five accounts you selected, inspect the actual post behind each useful alert, and record the reader need worth testing. Bring your own material to the experiment, then assign its publication slot and review date.

Frequently asked questions

How do I do a social media competitor analysis?

Choose five relevant accounts on one platform and review posts published during the last 30 days. Record a consistent sample in one sheet, including post links, dates, formats, visible metrics, and context. Compare posts against each account's own baseline, inspect the strongest patterns, and turn two observations into original experiments with an owner, a metric, and a review date.

Which metrics matter in social media competitor analysis?

Start with a metric consistently visible on the posts you compare, such as likes plus comments where both counts are available. Record post age, format, sample size, and each account's median alongside it. Treat unavailable data as missing, never as zero. Read visible comments for relevant questions, and measure your own experiments with a metric tied to their purpose.

Can I analyze social media competitors for free?

Yes. A free spreadsheet and what is visible on public profiles are enough for a limited review. Choose five accounts and sample up to 12 posts from each across 30 days. Save links, dates, content notes, and consistently available metrics. The main cost is your time, and the resulting sheet should lead to specific experiments you can run yourself.

How often should I analyze social media competitors?

For a small team, a monthly review of a fresh 30-day window is a practical starting point. Spend about 15 minutes each week noting new candidates and checking experiments already underway. Keep the same accounts and sampling method so reviews remain comparable. Reconsider the account list quarterly, or sooner if your audience, offer, or main publishing platform changes.

How do I compare accounts with different follower counts?

Compare how unusual a post is for its own account. Divide a consistently visible metric by that account's median for comparable posts in the same format and age band. Keep the sample size beside the multiple, and skip division when the median is zero. Use the result to choose posts to inspect; it does not establish which account is better.

Sources & editorial details

Examples, diagrams, and downloadable worksheets are editorial teaching materials. They do not report customer results. Source-check dates appear below where recorded.

Created with AI assistance and published by Krevaya. Send corrections to [email protected].