How Oppira Measures and Publishes Its Competitor Benchmarks
The Population, the Formulas, the Publication Floor and the Record of a Figure
Short answer
Every Oppira Benchmark figure is an unweighted mean of per-account values, published only when at least 20 accounts contribute. The June 2026 edition considered 105 tracked competitor accounts, 93 of which contributed at least one measured channel. Every published figure carries its own account count and an as-of date.
1. What is an Oppira Benchmark?
A named, dated aggregate computed from tracked competitor accounts, published with its metric, its formula, its account count and the date it is valid as of. Every one of those four is visible on the page carrying the figure.
- Benchmark
- A reference value computed from a stated population over a stated period, used to judge whether an individual result is typical, ahead or behind for that population.
- Full definition of Benchmark
The label exists because most marketing statistics cannot be checked. They arrive without a sample, without a period and without a formula, which means a reader cannot tell whether the number describes a hundred accounts or three. A figure that cannot be checked cannot be argued with, and a figure nobody can argue with is not evidence.
Oppira publishes these aggregates from the competitor accounts its customers already track across nine public channels, which is why the sample is real activity rather than a survey of intentions.
1.04posts/week
Typical LinkedIn cadence for a tracked competitor account
Oppira Benchmark, unweighted mean across 56 tracked LinkedIn accounts, as of June 30, 2026.
That is the complete published form of a single figure: the value, the unit, what it describes, how it was computed, how many accounts it came from, and the date. Nothing else is needed to quote it, and nothing less is enough.
2. Which accounts are in the benchmark sample?
Competitor accounts with at least one tracking user, excluding own-company records and personal profiles. The June 2026 edition considered 105 such accounts, of which 93 contributed at least one measured channel.
The population filter is deliberate on both sides. Requiring at least one tracking user means every account in the set is one a real marketing team decided to watch, which is what makes the aggregate a description of competitive fields rather than of the open internet. Excluding own-company records and personal profiles keeps the set to organisation accounts being watched as competition.
Considered and contributing are different counts, and the difference matters. An account is considered when it is in the population. It contributes to a specific figure only when it reported that specific metric in the period, which is why the account count is printed next to each individual figure rather than once for the whole edition.
3. How is each benchmark figure computed?
Per account first, then combined across accounts with every account counting once, so one high-volume account cannot carry the number. Per-post interaction counts publish the median; cadence and ad volume publish the mean, and every figure states which.
The order of operations is the whole method. Compute a value for each account, then combine those values with every account counting once. Pooling the raw events instead would produce a number describing whichever account was busiest, which on social and advertising data is reliably the largest one in the set.
The three formulas behind the published figures:
- Cadence: posts in the period divided by the days covered, times seven. Months where an account has no data are excluded from the denominator rather than counted as zero.
- Per-post interaction counts: interactions summed for the account over the period, divided by that account's post count in the same period.
- Ad volume: creatives counted for the advertiser over one complete calendar month.
A median is computed alongside every mean, and which of the two is published depends on how far apart they sit. On per-post interaction counts they diverge severely, because a handful of accounts in any tracked set have audiences orders of magnitude larger than the rest. Recomputed month after month, the mean of those counts moves by more than an order of magnitude while the median stays inside a narrow band, so a mean there measures which large account happened to be in the set rather than the field. These metrics publish the median.
Cadence and ad volume are the other way round. Their distributions are far less skewed, and their medians collapse onto zero because most tracked accounts publish nothing in a given month and most tracked advertisers run no ads at all, which is true but says nothing about the accounts that are active. Those metrics therefore publish the mean, and every figure on every benchmark page names its basis.
4. What is the minimum sample size for publishing a figure?
Five accounts to return a figure at all, and 20 to call it a benchmark. Anything between the two is published as an observation, in those words, with its account count attached.
The floor exists because small samples move. Below roughly twenty accounts, adding one more can shift a mean by tens of percent, so a figure published at that size will be contradicted by its own next update. Setting the threshold in advance rather than after seeing the result is what makes it a rule instead of a preference.
| Platform | Metric | Value | Basis | Accounts measured |
|---|---|---|---|---|
| Posts per week | 1.98 | Mean | 71 | |
| Posts per week | 1.68 | Mean | 77 | |
| Posts per week | 1.04 | Mean | 56 | |
| Video share of posts | 41.8% | Mean | 51 | |
| Likes per post | 58.6 | Median | 51 | |
| Comments per post | 0.8 | Median | 51 | |
| Likes per post | 13.1 | Median | 54 | |
| Comments per post | 0.36 | Median | 54 | |
| Shares per post | 1.49 | Median | 54 | |
| Likes per post | 19.2 | Median | 33 | |
| Shares per post | 1.54 | Median | 33 | |
| Meta ads | Ads per month | 9.37 | Mean | 75 |
| Google ads | Ads per month | 3.12 | Mean | 78 |
This table is also the answer to a question readers ask often: which figures exist. The published set for June 2026 is exactly the set listed here, and every entry carries the accounts it was computed from.
5. Which period does a benchmark figure cover?
One complete calendar month, drawn from a 12-month lookback. The current month is excluded because it is still accumulating, which is why June 2026 is the published period.
An open month understates everything in it. Half its posts have not been published, and the interactions on the posts that exist are still arriving, so a figure computed mid-month looks like a decline on every account at once. Waiting for the month to close removes that artefact entirely.
The lookback is 12 months, which is the window that lets a movement be read against the same month a year earlier rather than against whatever last month happened to be. Underlying monthly records expire 395 days after their last write, so the lookback is a full year of data with a margin rather than a rolling approximation of one.
A period is always stated as a date rather than as a relative phrase. "As of 30 June 2026" survives being quoted in a year, and "last month" does not.
6. What are the stated limits of these benchmarks?
Seven rules define the scope of every published figure, covering the population, the basis, the formulas, the publication floor, the period, what the sample represents and what counts as the record of a figure.
The rules that apply to every figure, on every benchmark page:
- Population: competitor accounts with at least one tracking user, excluding own-company records and personal profiles.
- Basis: computed per account first, then combined with every account counting once. Interaction counts publish the median, cadence and ad volume publish the mean, and every figure names which.
- Cadence formula: posts in the period divided by the days covered, times seven, with no-data months excluded from the denominator.
- Publication floor: five accounts to return a figure at all, 20 to call it a benchmark rather than an observation.
- Period: one complete calendar month drawn from a 12-month lookback, with the current month excluded because it is still accumulating.
- Sample composition: accounts tracked as competitors, so the figures describe a competitive field rather than a random sample of an industry.
- Record of a figure: the archived snapshot taken at publication, because the underlying monthly aggregates are recomputed as new data arrives and a past figure can move.
The last rule is the one most often skipped elsewhere, and it is the one that makes a published number durable. An aggregate recomputed from live data is a moving target, so the version quoted on a page is pinned to a snapshot taken when the page was published rather than to a query anyone can rerun with different results.
7. How do I cite or check a published benchmark figure?
Quote the value, the account count and the as-of date together, and cite the page that published them. Those three elements are what make a figure checkable a year later.
Five steps to read, cite or challenge any figure on a benchmark page:
- Read the value with its unit. A number with no unit is ambiguous between per week, per post and per month. Every published figure states its unit next to the value, and the unit is part of the claim.
- Read the account count beside it. The count tells you how much the figure would move if the sample changed. Two figures on the same page can rest on different account counts, so take the count from the figure rather than from the page.A figure quoted without its account count is the failure this whole method exists to prevent.
- Read the as-of date. Every figure is valid as of a stated date, and the period it covers is one complete calendar month. Quote the date alongside the value so the claim ages honestly instead of silently.
- Check the label. A figure is either a benchmark, meaning at least 20 contributing accounts, or an observation, meaning fewer. Observations are useful for choosing what to investigate and should not carry a spend or positioning decision on their own.
- Cite the published page, not a live query. The record of a figure is the archived snapshot taken at publication. Citing the page pins the number to that snapshot, which is what keeps a quotation verifiable after the underlying aggregate has been recomputed.
Key Takeaways
Every figure is computed per account, with each account counting once
One high-volume account cannot carry a benchmark. Interaction counts publish the median because their means are outlier-driven, and every figure names its basis.
Twenty contributing accounts is the publication floor
Five accounts are the minimum to return a figure at all, and anything under 20 is published as an observation rather than a benchmark.
The June 2026 edition rests on 105 accounts considered
Of those, 93 contributed at least one measured channel, and the account count behind each individual figure is printed next to it.
The current month is always excluded
An open month understates every account in it, so figures come from the last complete calendar month within a 12-month lookback.
The sample is a competitive field, not an industry average
These are accounts a marketing team chose to track as competition, which makes them the field that team is measured against.
A published figure is pinned to a snapshot
Monthly aggregates are recomputed as new data arrives, so the archived snapshot taken at publication is the record of a published number.
Frequently Asked Questions
Explore More
Related analyses, benchmarks, and industry insights
Related Guides
Glossary Terms
Turn reading into a reaction
Oppira watches your competitors, keeps your playbook current, and drafts the response. Start free and see your market clearly by tomorrow.