Choosing Channels

A map before a list

Sort candidate channels on two axes. The first is ownership: owned (your website, newsletter, community, product), earned (press, word of mouth, a partner's recommendation, organic search results) and paid (ads, sponsorships, paid placements). Owned channels compound and are cheap at the margin but slow to build; paid channels are fast and stop when the money stops; earned channels are the most trusted and the least controllable.

The second axis is the job from Lesson 1.1. Some channels mainly build memory across a broad audience; others mainly capture people already in market. Binet and Field's IPA analysis found that broad reach was associated with both effectiveness and efficiency, that brand and activation channels together outperformed either alone, and that some online channels are better suited to brand building and others to activation. A plan made entirely of capture channels — search ads on your own brand name, retargeting, intent-list outbound — is a plan with no brand side at all.

Scoring candidates

For each candidate, answer three questions honestly. Reach: how many people in your ICP does this channel reach, and how often? A niche newsletter read by eight thousand of exactly your buyers can beat a broad platform reaching millions of people who are not. Cost: in money and, for a small team, in hours — content-heavy channels are rarely cheap in time. Fit: does someone on the team have the skill and appetite to do this channel well every week for a year? A channel nobody wants to run will be run badly.

The LinkedIn B2B Institute's 95-5 rule (Lesson 1.1) adds a fourth consideration: does this channel reach the 95% who are not in market, or only the 5% who are? You need some of each. The Ehrenberg-Bass finding that brands grow mainly through penetration suggests weighting toward channels that can reach many category buyers, not only the ones you already know.

Pick two or three. Doing fewer channels consistently beats doing many intermittently, for the same reason consistency beats novelty in creative (Lesson 2.1).

Testing a channel

Before committing to a channel, run a time-boxed test with three things written down in advance: the budget (money and hours), the duration, and the threshold that would justify continuing — usually expressed as cost per qualified conversation, compared with your current channels. Our suggested starting window for a paid test is six to eight weeks — a working convention, not a researched figure; owned channels like content or a newsletter need far longer before their effects are visible, and judging them on a six-week test kills them before they work.

Kill tests that miss the threshold without ceremony. Scale tests that pass, gradually, watching whether the cost per conversation holds as you spend more; do not assume the economics of the first small budget survive a larger one.

The efficiency illusion

The channel that reports the best numbers is often the one closest to the purchase, and proximity is not causation. Thomas Blake, Chris Nosko and Steven Tadelis ran large field experiments at eBay, switching paid search off in some markets, and found that the returns to paid search were a fraction of what conventional non-experimental estimates suggested. For brand-keyword ads — ads shown to people already searching for eBay — they found no measurable short-term benefit: those buyers arrived through the organic result instead. For non-brand keywords, ads helped new and infrequent users, but most spend went on frequent users whose purchases the ads did not change.

The lesson for a small team is not 'never buy search ads'. It is that a channel's attributed numbers measure who passed through it, not who it caused to buy. Before shifting budget toward a channel because its dashboard looks best, ask whether those buyers would have arrived anyway. Lesson 5.3 covers how to test that properly.