Startup Acquisition Channel Pairs: What 406 Cases Combine
Download all 28 pairwise overlaps among eight normalized acquisition-channel families across 406 startup cases.
The most common acquisition-channel combination in this snapshot is Search + Organic: 68 of 406 cases, or 16.7%, mention both families. Organic + Video follows at 64 cases, while Video + Referrals appears in 62.
These results describe how acquisition channels are co-mentioned across a convenience sample. They do not show which channel came first, how much a company spent, or which channel worked best. This descriptive convenience sample is not a success forecast.
ProvenStartups is the Organization author and publisher.
Where should you start?
What does the snapshot show?
The 2026-09-22 snapshot contains 406 cases classified into eight regex-normalized channel families. Because one case can contain multiple families, the single-family counts overlap:
- ·Organic: 169
- ·Search: 167
- ·Video/editorial: 162
- ·Referrals: 149
- ·Paid: 122
- ·Communities: 97
- ·Outbound/sales: 71
- ·Product-led/free: 48
Organic and Search are the most frequently represented families, followed closely by Video/editorial. Product-led/free is the least frequently represented family in the snapshot.
The counts should not be added together to estimate the number of unique cases. A single project can appear in multiple family counts and many pair combinations.
Which channel pairs appear most often?
The ten most common unordered pairs are below. “Overlap” is the number of cases containing both families. “Share” is overlap divided by 406. Jaccard is the intersection divided by the union of the two family sets.
| Pair | Overlap | Share | Jaccard |
|---|---|---|---|
| Search + Organic | 68 | 16.7% | 25.4% |
| Organic + Video/editorial | 64 | 15.8% | 24.0% |
| Video/editorial + Referrals | 62 | 15.3% | 24.9% |
| Search + Video/editorial | 61 | 15.0% | 22.8% |
| Video/editorial + Paid | 56 | 13.8% | 24.6% |
| Organic + Paid | 54 | 13.3% | 22.8% |
| Organic + Communities | 53 | 13.1% | 24.9% |
| Search + Paid | 51 | 12.6% | 21.4% |
| Search + Referrals | 51 | 12.6% | 19.2% |
| Organic + Referrals | 50 | 12.3% | 18.7% |
Search + Organic leads by raw overlap and share. Video/editorial appears in four of the ten highest-overlap pairs, connecting with Organic, Referrals, Search, and Paid.
The Jaccard values add context. A pair can have a lower raw overlap but still be relatively concentrated within the combined presence of its two families. For example, Organic + Communities has 53 overlapping cases and a 24.9% Jaccard value, while Search + Referrals has 51 overlapping cases and a 19.2% Jaccard value.
What are the least common pairs?
The lowest listed overlaps are:
- ·Communities + Outbound/sales: 11 cases, or 2.7%
- ·Outbound/sales + Product-led/free: 11 cases, or 2.7%
- ·Paid + Product-led/free: 11 cases, or 2.7%
These are descriptive counts, not evidence that the combinations are incompatible or ineffective. They may reflect the composition and language of this sample, the normalization rules, or the relative frequency of the underlying families.
How should the pair results be interpreted?
A pair means that two normalized families were present in the same case. It does not mean that the channels were used at the same time or for the same purpose.
Co-mention is not sequence, spend, effectiveness, or causality.
For example, Search + Organic indicates that both families were identified in one case. It does not establish whether search preceded organic discovery, whether one channel supported the other, whether either channel generated customers, or whether the project invested equally in both.
The snapshot also does not distinguish between a brief mention and a central acquisition strategy. A family may be present because it was described as a channel, an experiment, a supporting tactic, or part of a broader growth narrative.
How were the pairs counted?
The data was normalized into eight regex-based families. Each case was then represented as a set of the families assigned to it.
For every case, each family was counted at most once. The same rule was applied to pairs: a case contributes once to a given unordered pair, even if the source text contains multiple mentions of the same family.
The analysis generated all 28 unordered intersections among the eight families, then sorted the results by overlap. Since there are eight families, the number of possible unordered pairs is:
8 × 7 ÷ 2 = 28
The overlap for a pair is the size of the intersection between the two family sets. The share is:
overlap ÷ 406
Jaccard similarity is:
intersection ÷ union
This makes the pair results reproducible while preserving the fact that cases can belong to multiple families.
What are the limits of regex normalization?
Regex normalization provides a consistent way to group related language, but it can misclassify text. A phrase may match a family without representing a meaningful acquisition activity, or a relevant activity may be described in language that the rules do not capture.
The eight families are therefore analytical groupings, not perfect measurements of channel strategy. Results depend on the source cases, the available wording, and the normalization rules used for this snapshot.
The sample is also a convenience sample. It should not be treated as representative of all startups, industries, geographies, stages, or business models. The descriptive convenience sample is not a success forecast.
How can the snapshot be verified?
The public files verify all 28 published pair rows. Full regeneration requires the 406 coded source records and should follow the same sequence:
- 1.Start with the 406 cases in the 2026-09-22 snapshot.
- 2.Apply the regex normalization rules that map source language into the eight channel families.
- 3.Convert each case’s assigned families into a set.
- 4.Count each family once per case.
- 5.Generate all 28 unordered family pairs.
- 6.Count each pair once per case using set intersection.
- 7.Calculate each pair’s share using 406 as the denominator.
- 8.Calculate Jaccard as intersection divided by union.
- 9.Sort the pairs by overlap.
The downloadable files provide the pair output and the accompanying manifest:
For additional context, see the related Startup Acquisition Channel Benchmark. You can also browse Projects or review How it works.
What should readers do with these results?
Use the pair ranking as a map of observed co-occurrence, not as a recommended growth playbook. The strongest signal in this snapshot is that Search and Organic frequently appear together, while Video/editorial connects with several other high-frequency families.
A practical next step is to inspect the underlying cases behind a pair and ask what the co-mention represents: discovery, distribution, conversion, retention, or simply narrative context. That additional review is necessary before making strategic decisions.
Frequently Asked Questions
What is the most common pair?
Search + Organic is the most common pair, appearing in 68 of 406 cases, or 16.7%.
Does the ranking show which channel performs best?
No. The ranking shows co-mention only. It does not measure conversion, revenue, customer quality, cost, sequence, or effectiveness.
Why do the single-family counts exceed 406 when added together?
Cases can contain multiple channel families. The counts are overlapping set memberships, so adding them double-counts cases that include more than one family.
What does a Jaccard score measure here?
Jaccard measures the size of a pair’s intersection relative to its union: the number of cases containing both families divided by the number containing at least one of them.