i
Skippable in-stream
The workhorse format. Bought against incremental reach with hook-first creative, because the first five seconds decide whether you paid for anything at all.
- Target CPM & CPV buying
- Hook testing
- Sequenced messaging
Six ways to buy attention inside Google.
Each discipline below can be bought on its own. Bought together they share one audience definition, one frequency cap and one measurement framework — which is where most of the value actually sits.
The plan is an argument, not a spreadsheet
A video plan should answer three questions without hesitation: who is it reaching, how often, and why that combination is right for this commercial problem. Most plans we inherit answer only the first, and usually with an impression count.
We start from audience evidence — Google's reach planner, your first-party segments, category search trends and your own sales data — and build a reach curve before looking at a single format cost. That order matters. It stops the plan being shaped by whichever format happened to be cheap last quarter.
You receive a primary recommendation plus two alternative scenarios, each with an explicit statement of what it gives up. Choosing between them is a business decision, and it belongs with you.
Deliverables
Audience definitionEvidenced
Reach & frequency modelIncluded
Scenarios provided3
Phasing calendar52 wk
Share of searchQuarterly
Plan turnaround3 weeks
One video plan, not four competing ones
In-stream, in-feed, Shorts and bumpers all sell against the same attention and, left unmanaged, all reach the same people. We plan them as one line item with a shared frequency cap.
ii
In-feed & Shorts
Discovery-led placements where the thumbnail and first frame do the work. Particularly effective for considered purchases where people are actively researching.
- Vertical-first edits
- Thumbnail testing
- Shorts inventory
iii
Bumpers & non-skippable
Six and fifteen second formats used for frequency and reminder duty against audiences already reached, never as a campaign's primary vehicle.
- Frequency duty
- Reminder sequencing
- Reach top-up
iv
Exclusions & brand safety
Content suitability set deliberately rather than left on default, with channel and topic exclusion lists reviewed monthly and published to you.
- Inventory mode selection
- Channel exclusion lists
- Monthly placement audit
Prospecting, kept honest by separation
Demand Gen reaches people across Discover, Gmail and the YouTube feeds. It is genuinely effective, and it is also very good at quietly harvesting people who were already coming to you.
A
AudiencesCustomer Match & lookalikes
Built from your own first-party lists with proper consent provenance, then extended by lookalike segments rather than broad interest categories.
B
AssetsImage & video sets
Full asset sets per audience, produced in our studio, so the campaign is not running one image against four very different groups of people.
C
StructureHeld apart from Search
Separate campaigns, separate budgets, brand terms excluded, so Demand Gen cannot claim credit for conversions Search would have won anyway.
D
ProofMeasured on incrementality
Geo holdouts and lift studies rather than platform-attributed conversions, because attributed and incremental are not the same number.
Standing exclusions
Mobile app inventoryExcluded
Made-for-advertising sitesExcluded
Parked domainsExcluded
Placement reviewWeekly
Exclusion list sharedYes
The most misused inventory in Google Ads
Left on defaults, a Display campaign will happily spend a substantial share of its budget inside mobile games and on sites built purely to carry advertising. That is where Display's reputation comes from, and it is entirely avoidable.
We run Display with app inventory excluded by default, a maintained made-for-advertising exclusion list, and a weekly placement review that we publish rather than summarise. Handled that way it is an efficient reach and remarketing channel.
Dynamic remarketing sits here too, wired to the same product or service data as the rest of the account so it never advertises something unavailable.
Written for the format, not trimmed to fit
The single most common reason YouTube activity underperforms is that the creative was made for a different medium. A television advert cut to fifteen seconds is not a skippable in-stream ad.
i
Hook-first editing
Re-cutting existing footage so the proposition lands inside five seconds, with three or four hook variants from the same material for asset-level testing.
- Hook variants
- Asset-level reporting
- Existing footage reused
ii
Vertical & Shorts
Purpose-built vertical edits with safe-area awareness, captioning and on-screen text sized to be legible on a phone held at arm's length.
- 9:16 edits
- Burned-in captions
- Safe-area layout
iii
Display & Demand Gen sets
Full responsive asset sets across every required ratio, built from one design system so six months of output still looks like one brand.
- All ratios covered
- Modular templates
- Handover files included
iv
Production, where needed
Where there is no usable footage we shoot it — modestly, quickly, and briefed against the performance job rather than an awards entry.
- Single-day shoots
- Performance-briefed
- Rights handled
The department that can fire our own recommendations
Our analysts report to the managing director, not to the planning team. That structure exists so a finding of "this campaign is not contributing" reaches you unfiltered, even when it concerns a campaign we recommended and are earning a fee on.
Upper-funnel Google Ads activity is genuinely hard to measure with last-click reporting, which is exactly why so much of it goes unexamined. We design the measurement before the campaign launches, not after somebody asks whether it worked.
Studies are documented in plain English and handed over with the underlying data, so a future agency can reproduce the analysis rather than take our word for it.
- 01Google Brand Lift studies
Ad recall, awareness and consideration measured against a randomised control group, designed with a stated minimum detectable effect so a null result is informative.
- 02Geo holdout testing
Matched regions with and without activity, measured on actual sales or enquiries. The closest thing to a controlled experiment that advertising allows.
- 03Search lift analysis
Branded search volume and share of search tracked against burst timing, which for most brands is the fastest honest read on whether video worked.
- 04Budget reallocation
An annual response-curve exercise showing where each format stops paying back, and a recommendation on what to do with the money after that point.
Full service, project, or audit only.
Most clients appoint us on a rolling annual retainer covering planning, campaign management, creative and measurement. We also run standalone projects — a launch plan, a Brand Lift study, an audit of your incumbent agency's video activity — and an advisory retainer for in-house teams running their own campaigns.
Discuss an engagement