Runway proves its value in marketing when you measure two things side by side: production efficiency and creative performance. Production KPIs show whether AI video is cheaper and faster per usable asset. Performance KPIs show whether those assets hold attention and drive results as well as traditionally produced video. A dashboard that tracks only one side will either overstate or understate what the tool is worth.
What is Runway and what do teams use it for?
Runway is a New York-based company that builds generative video models and an editing platform around them. Its Gen series of models create video from text prompts, images or existing clips, and the platform adds tools such as background removal, motion tracking, inpainting and camera controls. Film studios, agencies and in-house teams use it.
In marketing, common uses are short social clips, product shots in imagined settings, animated storyboards for client approval, B-roll for longer edits and quick visual concepts before a live shoot. Clip lengths and resolution limits depend on the model and plan, so check current specifications before planning a campaign.
Which production KPIs should you track?
Production metrics answer the question finance will ask: are we saving money or time? Track them per asset, not per tool subscription.
- Cost per usable asset: subscription share, credits used and editor hours divided by assets that actually went live.
- Generations per usable clip: how many attempts it took to get one acceptable result. This reveals hidden effort.
- Time from brief to approved asset, compared with your previous process for similar content.
- Revision rounds with stakeholders before approval.
- Share of assets rejected at brand or legal review.
The generations-per-clip metric is often the surprise. A clip that looks effortless may have taken 30 attempts and two hours of prompting. That is still cheaper than a shoot in many cases, but the number should be visible.
Which performance KPIs show creative value?
Performance metrics answer the marketer’s question: does the audience respond? Compare Runway-made assets with human-shot assets in the same campaign, placement and period.
- Thumb-stop or hook rate: three-second or two-second views divided by impressions, depending on platform.
- Average watch time and completion rate.
- Click-through rate and cost per click.
- Conversion rate and cost per acquisition from the click.
- Comment sentiment, especially reactions that call out AI-generated visuals.
How do you set up a fair comparison?
Label every asset at the point of creation. A naming convention such as campaign_concept_source_version, where source is AI, live or mixed, lets you group results later in Meta Ads Manager, TikTok Ads Manager or YouTube reports.
Then test like for like. If a Runway clip and a filmed clip carry different messages, you are testing the message, not the production method. Keep the hook, copy and offer identical where possible and change only the visuals.
Example test design
A travel brand promoting a coastal hotel might run four ads: two concepts, each in a filmed version and a Runway version built from location photos. Equal budgets for 10 days on Meta and TikTok give enough data to see whether the AI version holds attention and converts at a similar cost.
How long should you measure before deciding?
One test rarely settles the question. Generative video improves quickly, and your team’s prompting skill improves too, so early results often understate later ones. A sensible plan is a three-month review window with at least three matched tests per month.
Watch the trend in generations per usable clip. If it falls month by month, the team is learning and cost per asset will follow. If it stays flat or rises, the content type may simply be a poor fit for AI video, and it is better to move that budget back to filming. Also check fatigue: AI assets that win in week one sometimes decline faster, because audiences learn to spot a recognisable visual style.
What should the dashboard include?
Build it in Looker Studio, Power BI or your reporting tool of choice, pulling ad platform data and a simple production log kept in Sheets.
- Production panel: cost per usable asset and time to approval, AI versus traditional, by month.
- Performance panel: hook rate, watch time, CTR and CPA by source label.
- Volume panel: number of assets shipped per source, showing whether AI increased testing capacity.
- Notes: model or workflow changes, so jumps in quality can be explained.
What are the limits to keep in mind?
Generated video can show artefacts: hands, text, physics and product details are frequent weak spots. Never show a product feature in AI video that the real product does not have. Rights and disclosure also matter; review Runway’s terms for commercial use and platform rules on labelling AI-generated content.
How Alien Road applies Runway measurement
In a typical engagement we start by mapping the client’s current video spend and output: shoots per quarter, editor hours, assets delivered. That becomes the baseline. We then choose one content type, often social B-roll or concept animations, where AI video is most likely to help.
Setup includes the naming convention, a production log template and a test plan with matched pairs. Our editors use Runway alongside conventional tools, and every generated clip is reviewed for product accuracy and brand fit before it goes into an ad.
Monthly reports show both panels and a clear recommendation: expand AI video to more formats, keep it at the current level or reduce it. If AI assets save money but lose on CPA, the saving may not be worth it, and the report says so.
Where to start this week
Set up a production log with five columns: asset name, source, generations or shoot days, hours spent and approval date. Fill it in for every video asset from now on.
Then rename your live video ads to include a source label. Within a month you will have enough data for a first honest comparison, which is the starting point for any decision about Runway. Share the first results with both the creative and media teams, since each will notice different things.
Keep reading: Video Production · AI Marketing