For influencer cost per engagement, count only the interaction types you agreed to measure, on the campaign content and within the reporting window. A useful default is likes, comments, saves and shares. If the brief prioritizes discussion or content people keep and pass on, report comments, saves and shares separately. Keep views, unique accounts, website visits and purchases outside both totals.
The calculation is campaign cost divided by eligible interactions. The work is deciding what makes an interaction eligible before you see the result.
Write the denominator next to the price
A report that gives CPE without a definition leaves the reader guessing. It could mean cost per like, cost per interaction, or cost per click. Each answers a different question.
Meta's Instagram insights documentation defines Interactions as actions people take with content, including likes, comments, saves and shares. It separately defines Accounts engaged as unique accounts that interacted. Those units cannot be swapped in a calculation.
Modash's guide to influencer marketing metrics includes several possible engagement types and warns that combined engagement rates merge different behaviors. For a cost report, the practical response is to name the included actions and keep their individual counts visible.
Use these two definitions when both help the campaign decision:
| Report label | Eligible interactions | Question it answers |
|---|---|---|
| CPE, likes + comments + saves + shares | All four recorded action types | What did each counted content interaction cost? |
| Cost per comment, save or share | Comments + saves + shares | What did each action in this narrower set cost? |
These are reporting recommendations. They are not platform-mandated CPE definitions. The narrower set does not prove purchase intent, positive sentiment or greater financial value. A critical comment still counts as a comment.
A hypothetical cost-per-engagement sheet
Suppose a brand commissions an organic Instagram campaign. It receives creator-supplied results for the agreed campaign posts, collected at the same post age. No posts in this example receive paid distribution. Every figure below is hypothetical, and all costs use USD.
Hypothetical campaign costs
| Included cost | USD |
|---|---|
| Creator fees | $2,400 |
| Product cost and delivery | $300 |
| Allocated campaign labor | $300 |
| Total cost used in both calculations | $3,000 |
Choose the cost basis before comparing results. A creator-fee-only CPE and a fully costed CPE need different labels. For costs that span several campaigns, record the allocation rule. Use the complete campaign cost ledger to decide which costs belong in your numerator.
Hypothetical eligible interactions
| Recorded action | Count | Broad definition | Narrow definition |
|---|---|---|---|
| Likes | 8,400 | 8,400 | Excluded |
| Comments | 400 | 400 | 400 |
| Saves | 700 | 700 | 700 |
| Shares | 500 | 500 | 500 |
| Eligible total | 10,000 | 1,600 |
Hypothetical calculations
| Calculation | Arithmetic | Result |
|---|---|---|
| CPE, likes + comments + saves + shares | $3,000 / 10,000 | $0.30 |
| Cost per comment, save or share | $3,000 / 1,600 | $1.88, rounded |
| Likes as a share of broad interactions | 8,400 / 10,000 | 84% |
The campaign has the same cost and results in both calculations. Removing likes changes the denominator. It does not reveal a new outcome or show that performance deteriorated.
The broad CPE is useful alongside the interaction mix. If most counted actions are likes, readers can see that without guessing how much weight likes received. The narrow figure lets the team track the selected actions without claiming they have equal business value.
Avoid adding arbitrary multipliers to make saves or shares worth several likes. That creates a weighted score. If your team needs one, publish its weights and call it cost per weighted point. Keep the unweighted counts beside it.
Keep these units outside the total
Views measure exposure or playback under the relevant platform definition. For example, Instagram defines Reel Views as starts and replays. Adding them to the denominator would turn this sheet into a mixture of playback and interaction costs.
Accounts engaged measures unique accounts. Adding it to likes and comments would combine people and actions, with overlap between them. A person who likes and comments contributes to both action counts. The resulting CPE is a cost per counted action, not a cost per unique person.
Website activity needs its own definition. Google Analytics explains that events can record page loads, link clicks, purchases and system occurrences. A total event count therefore cannot stand in for creator-content engagement. Nor should a website visit be added to the social click that led to it without an explicit reason to count both stages.
For a traffic brief, report cost per specified click alongside content CPE. For a sales brief, report cost per qualifying order. Name the event, data source and attribution rule separately. These measures answer different questions, so keep separate columns.
Make creators comparable before sorting by CPE
Use a short eligibility record for every report:
- Content scope. List the campaign post IDs or URLs. Exclude unrelated account activity.
- Time window. Record publication dates, collection dates and the measurement cutoff. Give posts comparable time to collect interactions.
- Distribution. State whether counts include organic delivery, paid delivery or both. Match the cost scope to that choice.
- Action definition. Name the included fields and preserve the source labels. Record any exclusions, such as identifiable creator replies.
- Evidence. Retain the supplied report or export and its date. Mark missing fields as unavailable.
For the timing decision, use a reporting window chosen before results arrive. A lifetime total for one creator and an early snapshot for another will produce misleading comparisons.
Instagram says public accounts can access insights for their own content. That does not make their private analytics available to a brand. Request the relevant creator-supplied report rather than assuming saves, shares or every other field can be collected publicly. Field availability and labels can vary; retain the report's original names.
If one creator cannot supply saves, do not enter zero. Either report that creator's CPE as incomplete or calculate an additional comparison using only fields available for everyone. Give that comparison a new label. Keep exclusions consistent, and never infer a removable spam count from a low-quality-looking aggregate.
Sum eligible interactions across campaign posts, then divide total included cost by that total. Do not take an unweighted average of individual post CPEs. If the eligible total is zero, show "not calculable" alongside the cost and zero actions.
Before the next campaign goes live, add the eligible action list, cost basis and cutoff to the reporting brief. Ask each creator to confirm which fields they can supply. That agreement makes the eventual CPE reproducible.
