Influencer Performance Report 2026

• Bookings marked Live or Completed, launched in 2025 or 2026, drawn from 10,533 records in our client databases.
• D2C brands across 14 industries, from children's furniture to e-bikes. Brands and creators are anonymised.
• More than half of collaborations run without a dedicated link. Chapter 1 explores the cost.

Share of the 1,975 live collaborations with each field recorded.

Source: Ad Specialist booking data: 1,975 collaborations marked Live or Completed, launched in 2025–2026. Exported 24 August 2026.

Three in four collaborations use a discount code, but fewer than half have a dedicated link. Codes count only people who enter them. Someone who watches, searches for the store and buys without the code is missed. Someone finding the code on a coupon site is credited to the creator without ever seeing the content. Our guide to influencer marketing KPIs and tracking explains a complete setup.

Actual views are recorded for only 30% of collaborations. Much of the channel is therefore assessed using pre-launch estimates rather than what actually happened.

The gap is uneven and depends heavily on the channel.

Share of collaborations by channel with a tracking link and reported actual views.

Source: Ad Specialist booking data: 1,972 collaborations on Instagram, YouTube and podcasts, launched in 2025–2026.

Podcasts report downloads rather than views. Their low rate of reported views reflects the metric, not a lack of measurement.

A channel where 85% of bookings have links and actual views can be managed by new-customer results. Where only 13% have actual views, decisions lean on intuition and discount redemptions. Instagram still attracts substantial budget because reach is cheap. Chapter 2 asks whether that also means cheaper new customers.

Our IDA tracking setup addresses this gap: a dedicated link for each booking and attribution beyond coupon counts.

Effective cost per thousand impressions: total booking cost divided by actual views, multiplied by 1,000. We calculate the median per brand, then the median across brands, so one large client doesn't dominate. See calculating CPM for what the metric does and doesn't tell you.

Source: Ad Specialist booking data: collaborations with costs and reported actual views, launched in 2025–2026. Each brand has at least ten bookings per channel and five with both values.

Without controlling for brand, the dataset medians are €35 for Instagram, €54 for YouTube and €73 for podcasts.

Three brands have at least ten collaborations with recorded costs on both Instagram and YouTube. Comparing within each brand removes industry and product differences.

Instagram is 30–50% cheaper per thousand views for all three brands. Yet YouTube brings substantially cheaper new customers for two: half the cost for meal kits and one-third for e-mobility. Instagram leads only for children's furniture, where it is the main channel with 113 collaborations, while YouTube has 16 test bookings.

Format helps explain this. A 60–90-second YouTube video integration is the dataset's most common booking. It explains the product, stays online and can be found months later. A Story disappears after 24 hours. Explore our YouTube influencer marketing guide and TikTok vs. YouTube vs. Instagram for platform comparisons and cases.

Podcast advertising plays a different role, with a median order value of around €1,140: expensive products and fewer, larger purchases. Only two brands provide sufficient data, so podcasts remain a limited part of this comparison.

Collaborations with a recorded compensation model. Medians per group, without brand controls, because only two brands use affiliate deals in this dataset.

A ROAS of 7 versus 2 looks decisive, but two caveats matter. Affiliate deals cover only two brands, both with inexpensive products and high repeat-purchase rates, where creators can recommend with little risk. Affiliate bookings are also small, with a median cost of €85: many small creators, each with limited reach. ROAS is high, but absolute volume per booking is low.

Fixed fees buy access to creators who won't work on commission, creating scope to scale. The practical conclusion is a hybrid: commission for breadth, fixed fees for larger creators.

• Of 851 bookings with a recorded booking type, 384 are repeat bookings and 467 are first bookings.
• Out of 782 creators with at least one live collaboration in 2025 or 2026.
• Involve creators booked more than once during the period.

The State of German Influencer Marketing 2026 survey reports 42% of collaborations as one-off deals. A performance agency's booking data looks different: creators who work are booked again, and those repeat relationships account for most volume. Our case studies show how these always-on programmes are built.

This reflects different operating models. Measuring creators by new customers produces a result after the first collaboration and guides rebooking. Booking primarily for reach provides less of that evidence, so the search continues.

Status of all 7,425 booking records dated 2025 or 2026, from first contact to publication.

Source: Ad Specialist booking data: all records dated 2025 or 2026, across all statuses. Exported 24 August 2026.

Across all years since 2021, the decline rate is 31% and the live rate 42%. Outreach was broader in 2025 and 2026.

Four in ten creators contacted decline. Others don't reply or remain in negotiation. Overall, each live collaboration takes almost four contacts on average, including research, outreach, briefing and follow-up.

If you're building influencer marketing in-house, this is the workload that matters. Time goes into the three creators who declined as well as the one you book.

Actual views relative to the pre-launch forecast. 635 collaborations since 2021 with both values, excluding zero values left unmaintained after launch.

Source: Ad Specialist booking data: Live or Completed collaborations with forecast and actual views above zero, launched by 1 September 2026.

The median actual-to-forecast ratio is 1.07. Including the dataset's 141 zeros brings it to 1.00 and increases the share below 50% to 23%. The export doesn't reveal whether these zeros are measured results or unmaintained fields.

Six in ten collaborations meet or beat their forecast, with the median portfolio result 7% above plan. That's encouraging for professional forecasting, but individual bookings vary widely: only four in ten fall within 20% of plan, one in six delivers more than double, and one in five delivers less than 80%.

Putting your entire budget behind one forecast is a bet. Booking ten creators and planning with a range gives the portfolio room to deliver.

Four campaign planning rules grounded in this study's data.

Every placement gets a link. Every booking gets a dedicated tracking link and runs through our attribution model. The discount code remains an offer, rather than the sole measurement tool. We assess each creator by new customers, beyond views.

Start with YouTube for products that need explanation. A 60–90-second integration explains the product and stays online. For two of three comparable brands, it delivers cheaper new customers than Instagram despite a higher CPM. Explore YouTube influencer marketing

Combine commission and fixed fees. Commission for breadth, fixed fees for larger creators. Affiliate delivers the highest ROAS, while fixed fees buy reach from creators who won't work on commission.

Rebook what works, and keep testing. 78% of volume comes from creators booked more than once. Assessing new-customer results tells you whom to rebook after the first collaboration. Our case studies show this in practice.

How we went about it

Export from our client databases. 10,533 booking records as of 24 August 2026. Each represents a collaboration with channel, costs, views, conversions and tracking status. Brands and creators were anonymised before analysis.

Filtered to live collaborations in 2025 and 2026. 1,975 records marked Live or Completed. Chapter 5 includes all statuses from the same period; Chapter 6 uses all years since 2021 to obtain enough records containing both forecast and actual views.

Channel comparisons controlled by brand. We calculate the median for each brand with at least ten collaborations in a channel, then report the median of those brand medians. Every brand has equal weight.

Medians, with no imputed values. All metrics use medians to limit distortion by outliers. Missing values stay missing; a zero counts only when measured.

Our own analysis. We implemented and documented the calculations in a script so every figure can be reproduced with the next export.

• Dataset: 1,975 live collaborations launched in 2025 or 2026, from 10,533 booking records
• Brands: 16 anonymised D2C brands across 14 industries
• Creators: 782 across Instagram, YouTube, podcasts and Twitch, anonymised
• Metrics: effective CPM, new-customer cost, ROAS, order value, tracking fields, booking type, status, forecast and actual views
• As of 9 September 2026. Author: Moritz Lambrecht, Co-Founder & CEO

Every study has limits. Here are five that matter.

• The data comes from our clients. It describes performance influencer marketing for D2C brands, rather than the market as a whole.
• One children's furniture brand accounts for 37% of live collaborations. Brand-controlled metrics adjust for this; the simple percentages in Chapters 1 and 4 do not.
• Not every brand maintains every field. Blank cells mean “not collected”, not zero. Each figure therefore states its own sample.
• Conversions and revenue reflect attribution in clients' systems, usually through codes or tracking links. They systematically undercount collaborations without links.
• Affiliate and podcast results each rely on two brands. They are observations, not benchmarks.