Where they already are

Teams default to the channels they already know. A founder from a sales background reaches for outbound; a marketer from a content team reaches for a blog and SEO. Neither is wrong, but neither was chosen. The outer ring, which Weinberg calls "The Outer Ring: What's Possible", exists to force one concrete idea for every channel before anything is ruled out. The best channel is often the one nobody in the room has used, precisely because competitors have not used it either.

His nineteen channels, in his order and with his headings:

1. Targeting Blogs
2. Publicity
3. Unconventional PR
4. Search Engine Marketing
5. Social and Display Ads
6. Offline Ads
7. Search Engine Optimization
8. Content Marketing
9. Email Marketing
10. Viral Marketing
11. Engineering as Marketing
12. Business Development
13. Sales
14. Affiliate Programs
15. Existing Platforms
16. Trade Shows
17. Offline Events
18. Speaking Engagements
19. Community Building

Then the rings move inward: "The Middle Ring: What's Probable" holds the few you will test, and "The Inner Ring: What's Working" holds the one that earned the focus. A channel can always move back out when it stops working.

A channel is only as good as its overlap with the people you chose. The ICP from session 1 is a research lead at a software firm with 5 to 15 researchers, five or more recorded interviews a week. So the useful question is not "which channel is hot" but "where do these people already spend attention, unprompted?"

For research leads that points at a short, specific list: practitioner research communities and Slack groups, a handful of research newsletters and blogs they read every week, and a few research conferences where they give and attend talks. Each one maps onto a Weinberg heading: Community Building, Targeting Blogs, Speaking Engagements. Search Engine Marketing fits less well, because few people search for a category they do not yet know exists.

Remember the trigger as well. A new head of research is reachable in a different place from a team that is starting a shared repository; the first reads leadership posts, the second asks peers for tool recommendations. One ICP can point at two different channels depending on which trigger you are chasing.

Every middle-ring test fits on an index card, and the card is written before the test starts:

1. Channel and idea. The Weinberg heading and the specific move: not "blogs" but "sponsor two named research newsletters for one issue each".
2. The one metric. Cost per qualified lead, where qualified means the lead matches the session 1 ICP. Ideally cost per closed customer too, but a month is usually too short for deals, so state which one you will judge on.
3. The budget cap. A number, under $1,000, and it is a ceiling, not a target.
4. The date. The day you read the result, about a month out.
5. The kill line. The result that ends the test, plus the minimum sample you need before a pass counts. Cadence uses above $150 per qualified lead, and 10 or more leads to pass.

The order matters. A kill line written after the results arrive is not a threshold; it is a story told about the results.

If no possible result would make you stop, the money bought reassurance, not information. Three common ways a test is rigged without anyone meaning to: the metric is chosen after the data arrives (clicks looked bad, so the team reports impressions); the budget has no cap, so a slow channel simply gets more money until it "works"; or the sample is too small to say anything and gets read as a win anyway.

That last one is why the figure has three outcomes, not two. Speaking at one conference produced four qualified leads at $100 each, under the kill line but far below the 10 leads a pass needs. Inconclusive is an honest verdict. It means either run it again with a bigger sample or park it; it never means promote it.

Time matters as much as money. Traction is reported as recommending that a startup spend about half its time on getting traction rather than only on the product; treat that as the authors' rule of thumb, reported here from summaries, not as a measured finding. For the statistics of test design, sample sizes and stopping rules, see Learn Experimentation with Phoebe ↗.

The rule says a prospect has to see or hear from you seven times before they act. It is repeated in sales training, agency decks and channel plans, and there is no primary study behind it. The trail that does exist is reported, not read: a 2010 blog post traced the idea to an 1859 item about the sixth sighting of an advertisement, and to Thomas Smith's 1885 list of twenty exposures. Neither is a measurement; both are advertising wisdom written as a list. And the number in the list is not even seven.

Teach it as folklore and keep the useful kernel: most people do not buy on first contact, so plan more than one touch. How many touches your buyers need is an empirical question about your own funnel, answered in your own data, not a constant you can import.

The line is usually given to the American retailer John Wanamaker. Quote Investigator, whose account was read at source, finds no firm evidence that he said it. The earliest Wanamaker-linked evidence it reports is from 1919, and second-hand; a version linked to the British soap maker Lord Leverhulme appears in 1931, also second-hand, through his son; and David Ogilvy in 1963 credited both men. So the honest form is "attributed to Wanamaker", or better, "an old saying".

The saying survives because the question behind it is real: which half? That question is attribution, deciding how much credit each channel deserves for a customer who touched several. It is a discipline of its own, with its own models and its own disagreements, and it lives in Learn Marketing Attribution with Phoebe ↗. Here, the point is narrower: you cannot find the wasted half of a channel you cannot see.