# SaaS Keyword Cost by Length

> Cost per conversion by keyword word count in B2B SaaS Google Ads, September 2026: £124 for three-word keywords, £136 for five-plus, across 14 accounts.

Source: https://saasadsstudio.com/research/saas-keyword-cost-by-length/  
Author: The SaaS Ads Studio  
Published: 2026-09-16  
Updated: 2026-09-16

---

## Key figures
- **£124**: Cost per conversion, 3-word keywords (GBP)
- **£136**: Cost per conversion, 5-plus-word keywords (GBP)
- **12,589**: Non-brand keywords analysed
- **£1,998,000**: Ad spend covered (GBP)
- **14**: SaaS accounts in the study

A three-word keyword is the cheapest thing to buy in a B2B SaaS search account, and the gap between it and everything else is narrower than most keyword advice implies. This page sets out what a non-brand keyword costs per conversion by its word count, where those figures come from, and what they cannot be used to claim.

## Cost per conversion by keyword length {#cost-by-length}

Average cost per conversion runs £124 for three-word keywords, £130 for four-word keywords, £132 for two-word keywords, £135 for single words and £136 for keywords of five words or more. All five figures are in pounds sterling, they are Atlas account data, 14 accounts, and they are published on the [SaaS Ads Studio Google Ads keyword generator page](https://saasadsstudio.com/tools/google-ads-keyword-generator/), where they were read on 16 September 2026.

| Keyword length | Cost per conversion (GBP), [Atlas account data, 14 accounts](https://saasadsstudio.com/tools/google-ads-keyword-generator/), read 16 September 2026 |
| --- | --- |
| 1 word | £135 |
| 2 words | £132 |
| 3 words | £124 |
| 4 words | £130 |
| 5 or more words | £136 |

Every band in that table sits between £124 and £136, which is the part worth carrying away: word count moves cost per conversion, and it does not move it by much. Both ends of the range are Atlas account data, 14 accounts, [read on the source page](https://saasadsstudio.com/tools/google-ads-keyword-generator/) on 16 September 2026.

![Bar chart of SaaS Google Ads cost per conversion by keyword length, from £124 for three-word keywords to £136 for keywords of five words or more](/images/blog/saas-keyword-cost-by-length-cost-by-length.svg "Drawn from zero, so the bars show how small the spread between the bands actually is")

## What the study covers {#what-the-study-covers}

The analysis covers 12,589 non-brand keywords, £1,998,000 of ad spend in pounds sterling and more than 15,000 conversions, across 14 SaaS Google Ads accounts, with brand terms excluded. Those are Atlas account data, 14 accounts, and the [keyword generator page](https://saasadsstudio.com/tools/google-ads-keyword-generator/) carries the same study, read on 16 September 2026.

![Four figures describing the study: 12,589 non-brand keywords, £1,998,000 of ad spend, more than 15,000 conversions and 14 SaaS accounts](/images/blog/saas-keyword-cost-by-length-scope.svg "The four scope numbers that belong beside any figure lifted from this page")

Two details matter if you are citing this. The [source page](https://saasadsstudio.com/tools/google-ads-keyword-generator/) states the spend rounded, as £2m, read 16 September 2026; the figure in our own quarterly figures file is £1,998,000, and that is the one to quote. And the dataset is 14 accounts. Until 13 September 2026 the tool page described it as 100+ SaaS brands, which overstated the study several times over. If you are holding the older number, replace it.

## Why word count is not search length {#match-types}

A keyword's word count is not the length of what somebody typed. Google Ads matches keywords to search terms through three match types, and [broad match is the default applied to every keyword unless you specify another](https://support.google.com/google-ads/answer/7478529?hl=en), read on 16 September 2026. The same page states that broader match types capture all the queries of the narrower ones and more, so a two-word broad match keyword can be paid for by a long, specific search. The bands above describe the keywords an account bought, not the searches it was charged for.

That distinction is why a keyword length benchmark cannot be set beside a study of search query length, and why it says nothing about long-tail search volume. It is a statement about how accounts were built.

## What this study does not tell you {#limits}

- The period is not published. The source page gives no date range for the spend, so this page does not state one either, and a figure without a period should be quoted as such.
- It is B2B SaaS, and it is 14 accounts charged in pounds sterling, per the [source page](https://saasadsstudio.com/tools/google-ads-keyword-generator/) read on 16 September 2026. It is not a cross-industry benchmark and should not be presented as one.
- A conversion is whatever each account counts as one: a trial start, a demo request, a signup. Cost per conversion is not cost per customer, and the two differ by whatever your trial to paid rate is.
- Brand terms are excluded, so none of this describes what bidding on your own name costs.
- It describes spend that happened, not an experiment. Nothing here shows what would happen to your cost per conversion if you moved budget from one band to another.
- Keyword length is not keyword quality. A cheap band is no reason to add a bad three-word keyword, and no band protects you from matching to searches you did not want.

## Method {#method}

Every figure on this page is first-party: Atlas account data, 14 accounts, aggregated from the Google Ads accounts this company runs for B2B SaaS customers. A vendor's figure about itself is a claim rather than an independent statistic, and that rule applies to ours, so check it against your own account before you rely on it.

No figure on this page is derived. There are no ratios, no index numbers and no currency conversions: the five band figures and the four scope figures are published values, read off the [keyword generator page](https://saasadsstudio.com/tools/google-ads-keyword-generator/) on 16 September 2026 and reproduced unchanged. The one place this page differs from the source is that the spend is given in full, £1,998,000, where the source rounds it to £2m.

The figures in the [keyword length study](https://saasadsstudio.com/tools/google-ads-keyword-generator/), read 16 September 2026, are produced by a fixed quarterly script rather than by hand, and that script applies two floors before anything is published: at least five accounts in any cell, and no single account contributing more than 40% of that cell's spend. Those floors are this company's publication policy rather than findings of the study. A cell failing either floor is dropped rather than softened, no customer is named, no single account's numbers appear, and no cell is recut or combined afterwards.

The Google Ads Help material cited here is definitional rather than statistical. It describes how match types behave, not what anything costs.

## What this means if you are running ads {#what-it-means}

If your non-brand ad groups are built mostly of single words and five-word-plus phrases, these bands say you are sitting at the dear end of a narrow range, and your search term report is where the reason will be visible. Atlas builds and runs SaaS Google Ads campaigns inside limits you set and changes you approve, and you can see what it would build in your account at [app.saasadsstudio.com](https://app.saasadsstudio.com) or on the [Google Ads for SaaS](https://saasadsstudio.com/google-ads-for-saas/) page.

## Questions people ask about this

**Which keyword length is cheapest in B2B SaaS Google Ads?**

Three-word non-brand keywords, at £124 per conversion in this study, with four-word keywords next at £130. All the figures are in pounds sterling and are Atlas account data, 14 accounts, read on the source page on 16 September 2026.

**Do these figures include brand keywords?**

No. Brand terms are excluded from all 12,589 keywords in the study, so nothing here describes what bidding on your own company name costs. Brand traffic is usually far cheaper and belongs in a separate comparison.

**What period does the study cover?**

The source page does not publish a date range, so this page does not state one. It will be added when the quarterly figures file carries it.

**Is this the same as the 100+ SaaS brands figure?**

No. The study is 14 SaaS accounts. A claim of 100+ brands appeared on the keyword generator tool page until 13 September 2026 and overstated the dataset. Quote 14 accounts.

## Sources
1. [Atlas account data, 14 SaaS accounts: keyword length study, published on the SaaS Ads Studio Google Ads keyword generator page](https://saasadsstudio.com/tools/google-ads-keyword-generator/), read 2026-09-16
2. [Google Ads Help, About keyword matching options](https://support.google.com/google-ads/answer/7478529?hl=en), read 2026-09-16

## Cite this page
Free to quote with a link back. Figures last checked 2026-09-16.

> The SaaS Ads Studio, "SaaS Keyword Cost by Length", 2026. https://saasadsstudio.com/research/saas-keyword-cost-by-length/