Third-Party Segment Selection for Industrial B2B Buyers

On a six dollar working CPM, a third-party business segment carrying a two dollar data fee is a third of your media budget spent on a list you are not allowed to inspect. Sometimes that trade is worth making. For South African industrial and energy targeting it almost never is, because the panels that produce a segment called something like energy sector decision maker for this market are thin enough that the output is inference stacked on inference, and you are paying a per-impression fee for it on every single impression served. Where the money actually goes: Data fees are charged per thousand impressions, not per member of the audience, which means the cost scales with how much media you buy rather than with how good the list is. A segment that halves your addressable pool and adds thirty per cent to your effective CPM has to improve qualified response by well over a third before it breaks even, and most do not. The first thing I do on any engagement is rebuild that arithmetic for the segments already running, with the real fees from your platform's billing detail rather than the rate card. How I test a segment: Three cells on identical inventory, identical creative and identical flight dates: the segment, a contextual or account-list alternative, and an untargeted control drawn from the same supply. Judged on qualified session rate and enquiry rate, never on click-through, which in display measures accidental taps and nothing else. Minimum readable test is a few weeks and a real budget, and I will tell you before we start whether your volume can produce a result worth acting on. If it cannot, that is the finding. What tends to survive: Intent data with a disclosed methodology, where you can see which content topics drove the surge and at which company, is a different product from a modelled demographic segment and is the category that most often earns its fee. Firmographic segments used for exclusion rather than inclusion often pay for themselves too, since removing consumer inventory is worth more than adding an uncertain audience. What tends not to survive is anything sold as reaching decision makers in a named industry without an explanation of how membership was determined. That market got smaller when Oracle closed its advertising business on 30 September 2024 and its BlueKai and Datalogix segments went with it, which stranded a great many media plans and taught the industry less than it should have. Excluded: First-party data work, which is a separate engagement and usually the better place to spend the same money. Contract negotiation with data providers. Any billing arrangement tied to a share of data spend; I am frequently recommending you spend less on it, and I will not be paid in a way that argues against that advice. Say no to me if: Your buying is entirely in South Africa and your target is a few hundred named companies. In that case the honest answer is to skip third-party data altogether and target the account list directly, and I would rather say that now than take a fee to say it in six weeks. Also say no if the segments were chosen by a partner who is paid on the data spend and that relationship is not up for discussion. The audit will produce an uncomfortable number and nothing will change. What you receive: A per-segment economic read with the break-even response lift each one requires, a test design you can run without me, and a written recommendation on which to keep, which to convert into exclusions and which to cancel, with the notice periods for each.