← All playbooks·Cross-media measurement
Playbooks · Cross-media measurement
Choose a single measurement currency for a deal
This job produces the single, contractually binding currency source and metric set that buyer and seller both use to measure delivery on a deal. It feeds the make-good and reconciliation math the moment actual delivery is checked against the guarantee.
What you need first
- Seller's delivery numbers for the flight from their ad server or set-top-box feed
- Buyer's independent measurement numbers for the same flight, market and audience, from the panel or big data source
- MRC accreditation documentation for each candidate source, including exactly which metrics the accreditation covers
- Variance history between the two sources on the last several comparable campaigns
- Panel sample size and composition for the specific target audience, from the panel provider's technical documentation
The procedure
- Pull matched-period numbers from every candidate source for the same flight, market and target audience, producing a side-by-side comparison table
- Confirm MRC accreditation status and scope for each candidate source, producing an accredited/non-accredited shortlist
- Calculate the variance between sources on reach, GRPs and frequency, producing a bias direction and percentage
- Cross-check that variance against the last six comparable campaigns, producing a historical tolerance band
- Match each source's methodology, panel or big data, to this deal's audience and device mix, producing a fit note that flags known undercoverage
- Select the single currency and, if switching from the source used to plan the deal, calculate a conversion factor to reconcile prior estimates, producing the number to carry into billing
- Write the chosen currency, its provider and the metric definitions into the IO, producing the binding reference for delivery and make-goods
- Get written sign-off from the counterpart referencing the same source and period, producing a countersigned agreement
Worked through with numbers
Netherlands, 4-week cross-media flight, target 25-54. Panel source (accredited, 5,200 households, 2,850 effective for 25-54): reach 61.4%, GRPs 196, frequency 196/61.4 = 3.2. Seller's STB big data feed (not accredited): reach 74.8%, GRPs 168, frequency 168/74.8 = 2.25. Reach variance: 74.8 - 61.4 = 13.4pp, or 13.4/61.4 = 21.8% relative. GRP variance: (196-168)/196 = 14.3%, with panel higher on GRPs and big data higher on reach. Last six campaigns averaged 9pp reach variance and 6% GRP variance, so this flight's 21.8% is outside tolerance and gets flagged in the file. Panel wins the selection on accreditation and audience fit despite the flag. Conversion factor to reconcile the media plan, built pre-flight on big data estimates, to panel currency at settlement: 196/168 = 1.167, a 16.7% uplift applied to big-data-based forecasts going forward. Read it as: the deal bills on 196 GRPs and 61.4% reach, and any prior plan built on the 168/74.8 numbers needs the 1.167 factor applied before comparing to the guarantee.
Where it goes wrong
- Don't average the two sources into a blended figure. Neither party can reproduce or accredit a blend, so make-goods have no ground truth to point to.
- Don't lock the currency after the flight has started. Set it before insertion so delivery and make-good math are undisputed from day one.
- Don't treat MRC accreditation as covering every metric the source reports. Check the accreditation letter names the exact reach or GRP definition used in this deal, not just impressions.
- Don't rely on total panel size. Check the effective sample for the specific target segment, since a 5,200-household panel can drop under 1,000 once filtered to a narrow demo.
How to know it is right
The chosen currency, provider and metric definitions read identically in both parties' contract paperwork, and the variance and conversion factor used to get there are on file for the next make-good conversation.
Terms used