Lab #02 found that three companies which looked like coverage gaps weren’t. A single department filter was hiding 72 senior contacts. That was three companies and one observation, so we ran it properly: twenty companies, five credits, one filter added at a time, counting exactly what each one costs.
The short version: two filters removed 94% of the senior contacts at ten companies. The department filter did nearly all of that damage, and among the people it removed were three Chief Revenue Officers, because a CRO is commonly tagged General Management rather than Sales. Separately, on a set of ten early stage companies, half of everyone tagged with founder seniority was not a founder.
Every figure on this page was pulled live via Lusha’s MCP tools on August 15, 2026, at a total cost of 5 credits. No contact fields were revealed. Companies and individuals are anonymised, and titles are reported because the titles are the finding.
Finding 1: two filters removed 94% of the population
Ten established B2B software companies. One search, run three times, adding a single filter each pass.
| Filter | Contacts returned | Change |
| VP and C-suite, any department, any country | 690 | baseline |
| Add: Sales department | 61 | −91% |
| Add: United States | 39 | −94% cumulative |
Nothing about the underlying data changed across those three rows. The same ten companies, the same session, minutes apart. The department filter alone removed nine out of every ten senior contacts.
That number on its own is interesting but not actionable. The useful question is which people it removed.
Finding 2: the department filter removes the Chief Revenue Officer
Three CROs appeared in the unfiltered baseline, at three separate companies in the set. All three are tagged department: General Management.
So applying a Sales department filter, the obvious first move when you are looking for sales leadership, excludes the senior sales person at the top of that function. Not because the record is missing. Because a CRO’s departmental classification usually reflects that the role sits in company leadership rather than inside the sales function.
The same pattern applied to several other revenue roles in the set: a VP of Global SMB, a Regional VP of Enterprise, and an Area VP were all tagged General Management despite carrying commercial responsibility in their titles.
This is not unique to any one data provider. Departmental classification is a judgment call applied at scale, and a C-level revenue role legitimately belongs to more than one department. The practical consequence is the part worth internalising:
If you filter for the Sales department and get a thin list, you have probably filtered out the person you actually wanted. Run the search once without the department filter and read the titles instead. On this set that was the difference between 39 results and 690.
Finding 3: half of “founder” seniority is not a founder
Lab #02 noticed two contacts labelled founder seniority whose titles were Founding Account Executive and Founding Sales Development Representative. Individual contributor roles, founder level label. We tested whether that generalises, using ten early stage companies where such titles are common.
Twenty-eight records came back tagged with founder seniority. The split was exact:
| Actual founders (14) | Not founders (14) |
| Founder Co-Founder Co-Founder and Chief Executive Officer Co-Founder and Chief Technology Officer | Founding Engineer (five records) Founding Designer (two records) Founding Product Marketer (two records) Founding Software Engineer Founding Product Designer Founding Data Scientist Founding Account Manager |
Fourteen and fourteen. The label appears to key on the word “Founding” in the title string, which means an early hire with a “Founding” prefix carries the same seniority value as the person who started the company.
This compounds with Finding 2 in a way worth spelling out. One of those fourteen non-founders is a Founding Account Manager, tagged department Sales. So at an early stage company, a search for Sales plus founder seniority surfaces an individual contributor as founder level. At an established company, the same department filter drops the actual CRO. The filter combination that looks like it should find sales leadership returns the wrong seniority tier at both ends of the company size range.
One more thing that set showed: of twenty-eight founder tagged records, exactly one sits in Sales. The founder population skews heavily technical, engineering, design, and product. If you are targeting founder led companies for a commercial conversation, seniority alone will not get you to a commercial contact.
Finding 4: two failure modes, opposite behaviour
During the second run we passed a country as “United States” instead of the two letter code. The response was an explicit, typed error naming the exact constraint and pointing to the tool that lists valid values, at zero credits.
Set that beside what Lab #02 found. A malformed filter value fails loudly and charges nothing. A malformed company name can match a real but unrelated company through substring matching, return a full and normal looking result set, and charge for the search.
Two kinds of bad input, opposite behaviours, same API. The practical rule that falls out of it:
- Structured filter values are self correcting. If you get them wrong you will know immediately and it costs nothing. Use the filter tools to list valid values rather than guessing.
- Free text inputs are not. Company names, and anything else typed rather than selected, can match something plausible and wrong. Sanity check the domain, size, and location of a returned company before trusting the result set.
What to do with this
Treat a thin result as a filter problem first. Remove one filter at a time and watch the count. On this set, removing a single department filter took the result from 39 to 690.
Do not filter by Sales department when you want sales leadership. Filter by seniority, then read the titles. The CRO is likely tagged General Management.
Read titles, not derived labels. Seniority and department are both derived fields. When a derived label disagrees with the title string, the title string is what a buyer would recognise.
Include founder explicitly when targeting early stage companies, and expect to filter the results by hand, since roughly half will be early hires rather than founders.
Use the filter tools to get valid values. They cost nothing, and structured values that come from them cannot silently match the wrong thing.
Method notes
- Two sets of ten companies. Findings 1, 2 and 4 used ten established B2B software companies. Finding 3 used ten early stage companies, chosen because “Founding” prefixed titles are common at that stage and absent at scale.
- Both sets are distinct from Lab #01 and Lab #02, and all companies are anonymised here.
- Total cost: 5 credits across four searches. Filter value lookups cost nothing. The malformed filter attempt cost nothing.
- No contact fields were revealed. All work was preview only, so no emails or phone numbers were retrieved or spent on.
- Finding 3 counts records, not people, and reports one session on one date. It is a measurement of a labelling pattern, not a general accuracy rate.
FAQ
Why does my contact search return so few results when I filter by department?
Because departmental classification is a derived field and senior commercial roles frequently sit outside the department you would expect. In this test, adding a Sales department filter to a VP and C-suite search removed 91% of results across ten companies, including three Chief Revenue Officers who are tagged General Management.
Is a CRO in the Sales department?
Often not, as a data classification. In this run, every Chief Revenue Officer returned was tagged General Management rather than Sales. If you filter for the Sales department when looking for sales leadership, you are likely to exclude the person at the top of that function.
Does founder seniority mean someone founded the company?
Not reliably. Across twenty eight founder tagged records at ten early stage companies, exactly half were not founders. They were Founding Engineers, Founding Designers, Founding Product Marketers, and one Founding Account Manager. The label appears to key on the word “Founding” in the title.
How do I tell whether an empty result is a coverage gap or a filter problem?
Remove one filter at a time and watch the count. Filters compound, and each one can independently exclude a record that exists. In this test the same search returned 690, then 61, then 39 results depending only on which filters were applied.
Will a bad search input tell me it was wrong?
It depends on the input type. A structured filter value that is invalid returns an explicit error at no cost. A free text value such as a company name can match something real but unrelated and return a normal looking result set. Verify free text matches; trust structured values once the API accepts them.
Go deeper
- claude.com/connectors/lusha, install Lusha directly from Claude
- www.lusha.com/campus/plays, ready to run prompts for verified prospecting and signals
- How AI agents pick your data tool, where the credits go and how to check before you spend
Want to run this against your own accounts? See what Lusha’s connector covers
