Build an internal video marketing baseline

What to take away

When assessing a video marketing benchmark, most teams quote a figure before checking what was measured, over what period, and on whose definition of a view or a lead.
A benchmark is only useful if the sample, the metric definition and the date are all stated. Without those three, treat the number as decoration.
Build your own comparison from your last two campaigns, then use published research only to sanity-check the range.
If a benchmark relies on personal data, check that the data behind it has an appropriate lawful basis. That does not, by itself, validate the performance figure.

Why borrowed benchmarks mislead

The common mistake is quoting a benchmark that answers a different question. Average cost per completed view means little if your campaign treats a three-second play as a view and the source counted a full watch.

Sample size matters too. Forty respondents drawn from every sector say little about a niche audience, and format and campaign length shift the number.

Dates are the third trap. A cost-per-view figure from 2021 is not a 2026 planning assumption. Name the publication date whenever you cite one.

A benchmark you cannot rerun is not evidence, so the measurement and reporting pillar sets out how to define each figure before comparing it.

Context sources are not performance benchmarks

The Office for National Statistics holds cultural identity data that can help describe an audience by language, religion and national identity. It can inform audience context, but it does not provide video campaign performance benchmarks.

GOV.UK money and tax pages cover corporation tax, VAT and PAYE obligations relevant to production businesses. They may help explain agency cost structures, but they cannot substantiate claims about video views, completed views or leads.

Build a local baseline first

Four figures to calculate

Cost per thousand impressions
Cost per completed view
Cost per qualified lead
Revenue attributed within ninety days

Start by writing down four definitions and keeping them for every campaign:

Viewthe platform's own counted view, with the platform named, because each platform counts a view differently.
Completed viewa play to the end of the video, or to a stated percentage of it if the platform reports quartiles.
Qualified leada form or call that meets the client's own qualification rule, not every form submission.
Conversionthe outcome the campaign was bought for, such as a booked call or a sale, recorded in the client's CRM.

Then build the baseline in five steps:

Pick your last two completed campaigns for the same client or product, each run for its full planned flight.
Export views and completed views from each platform's own analytics, such as YouTube Analytics, Meta Ads Manager or LinkedIn Campaign Manager, for the campaign dates only.
Take leads and conversions from the client's analytics and CRM, filtered to the campaign's tracked links or codes, and apply the qualified-lead rule.
Total the cost for each campaignproduction, media and agency fees, all on the same basis, with or without VAT but never mixed.
Calculate cost per view, cost per completed view, cost per qualified lead and conversion rate, and record them in one sheet with a row per campaign and a column per metric, plus the date range and the definitions used.

Update the sheet after every campaign and review it quarterly. Two data points are not a trend, but they are a defensible starting range. If research quoted by video marketing agencies falls outside that range by a wide margin, investigate before adopting it.

If the numbers were gathered inconsistently, the benchmark inherits the error. Fix the definitions first.

A worked example

For example, a video production company in Leeds runs two campaigns for the same client. Campaign one costs £12,000 in production and media and returns 400 qualified leads, so cost per lead is £30.

Campaign two costs £9,000 and returns 250 qualified leads, so cost per lead is £36. The internal range is therefore £30 to £36.

A published benchmark claims £18 per lead across the sector. Before chasing it, the team checks the definition. The source counted any form submission, including newsletter sign-ups, which the client excludes.

Adjusted to the client's definition, the same source implies roughly £40 per lead. The team's own £30 to £36 range now looks reasonable, and the reported saving was never real. All figures here are illustrative.

Questions to ask of any study

Who was sampled, and how were they recruited? What counted as a view, a lead or a conversion? When was the fieldwork carried out?

Was the research funded by a platform or vendor with an interest in the answer? If so, read the method section before the headline.

Finally, does the benchmark describe agencies or clients? A figure about agency margins cannot justify a client's media budget. Compare like with like.

Common questions

Can I use a benchmark without my own data?

Yes, but label it an external reference rather than a target. State the source, the date and the metric definition alongside it.

How many campaigns do I need before a range is meaningful?

Two gives a range, four gives a pattern. Below two, describe the figure as a single result, because one campaign cannot show variation.

Should benchmarks appear in client reports?

Only where they change a decision. A benchmark that alters no budget, brief or creative choice adds length without evidence.

What if the benchmark flatters us?

Check the definition before you celebrate. A favourable comparison usually means the two sources counted different things.