Why SaaS accounts leak differently from other accounts
An ecommerce store loses money on the wrong product word. A SaaS company loses it on the wrong person. Someone searching "crm software" may be a buyer, a student writing a comparison, an employee looking for the login page, or a developer after an open source option. All four trigger the same keyword, and only the first one ever pays.
The cost is also lopsided. Conversion volume is low, often a few dozen demo requests a month, so a handful of expensive junk clicks changes cost per demo visibly. And because sales cycles are long, a bad exclusion does not show up for weeks, which is why approval and undo matter more here than speed.
Five requirements before you pay
1. Intent classes, not word lists. The usual SaaS waste comes in recurring families: "login", "support", "pricing" for a competitor, "salary", "jobs", "course", "template", "open source", "free". A tool should group the queries of your report into such families so you decide once per family, with the money each family burned shown next to it. A flat list of 900 rows is not a tool, it is a spreadsheet.
2. A conflict check against your converting terms. If "crm for contractors" produced two demos, no exclusion containing "contractors" should pass quietly. The tool must show which converting queries a proposed negative would touch, before you accept it.
3. Approval, not auto-apply. With few conversions, the data behind any rule is thin. Require a queue where you accept, edit or reject each proposal, and a log of who approved what. The scripts versus software comparison explains why silent automation is the wrong default.
4. Phrase and exact control. Broad-match negatives are rarely safe in a SaaS vocabulary, where one word carries several meanings. You want proposals in the narrowest form that stops the waste. The match types guide shows the difference.
5. A trial on your own export. Never judge from a vendor sample. The one-hour test plan gives you pass and fail criteria.
A worked example with numbers
Take a project management SaaS spending 12,000 USD a month on search, with 60 demo requests, so 200 USD per demo. The 90 day search terms report is exported and the zero conversion rows are grouped. The figures below are illustrative, so replace them with your own export.
| Family | Example queries | 90 day cost | Action |
|---|---|---|---|
| Job seekers | project manager jobs, salary | 1,150 USD | Block as phrase |
| Students and courses | project management course, pdf | 940 USD | Block as phrase |
| Login and support | app login, help center | 620 USD | Block as exact |
| Free and open source | open source tracker, free gantt | 1,380 USD | Block as phrase |
| Held back by conflict check | project software for contractors | 410 USD | Keep, it converted twice |
The four blocked families total 4,090 USD over 90 days, about 1,360 USD a month, roughly 11 percent of spend. The fifth family looks bad on cost alone and would have been blocked by a keyword filter. The conflict check keeps it because two demos came from it. If the freed 1,360 USD buys clicks at the same cost per demo as the rest of the account, that is about seven extra demos a month. A Growth plan costs 149 USD a month, so the software is a small line next to the number it moves. The ROI guide shows how to run the same break-even math with your own figures.
Buying table for a SaaS team
| Requirement | Pass | Walk away |
|---|---|---|
| Grouping | Families with cost per family | Only a flat list |
| Safety | Shows touched converting terms | No conversion check |
| Control | Approval queue and undo | Auto-applies by default |
| Match type | Phrase and exact proposed | Broad only |
| Cost | Plan price below one month of found waste | Price hidden until checkout |
If your SaaS also targets leads with forms rather than demos, the lead generation page covers that setup, and the pricing page lists what each plan includes. Teams with several products or portfolios in one account often track them in a planning tool such as Portfoliohub, so each product line has its own spend target to review against.
For the underlying mechanics, Google documents negatives in its negative keywords help.
How to roll it out in the first month
In week one, run the 90 day export and block only the families where every query has zero conversions and the cost is clear, usually job seekers, courses and login or support terms. In week two, review the families that need judgment, such as free, open source and competitor names, one at a time in the approval queue. In week three, check cost per demo and the lead quality your sales team reports, not click volume. In week four, extend to the next campaign set and keep the log, so a drop in demos can be traced to a specific exclusion and undone.
The common mistake is to treat the first report as a cleanup and then stop. SaaS vocabulary keeps shifting as you add features and landing pages, and Google keeps matching new query forms to old keywords. A short weekly review beats a big quarterly one, because each exclusion is made on fresh data and a wrong call is caught within days.
Questions B2B SaaS buyers ask
Is negative keyword software worth it for a small SaaS account? It is when found waste in a 90 day export exceeds a few months of the plan price. Run your own export through the demo first. If the ranked list shows less than that, you do not need to pay.
Should I block competitor names? Only the families that show zero conversions in your own data. Some competitor searches convert as comparison shoppers. Review them in the approval queue rather than with a blanket rule.
How often should a SaaS account be reviewed? Weekly for the first month after any campaign change, then every two weeks. Low conversion volume means the weekly sample is small, so group by family instead of reacting to single queries.
Can it run on both Google Ads and Bing? Yes, on Growth and above. See the Microsoft Ads negative keyword tool page.