Bad data is data that fails the decision.
Bad GTM data is information that is inaccurate, outdated, incomplete, duplicated, incorrectly connected or unsuitable for the action you intend to take.
A salesperson finds an old title. An email bounces. A physician is linked to a facility they have left. These deserve investigation. They do not, by themselves, establish the quality of an entire database.
But “data changes” should never become an excuse for recurring errors. The useful question is: How reliably does the information support your decisions, and how effectively does the provider correct what changes?
The record is wrong.
A former employer, an incorrect affiliation, a duplicate organization or an invalid address.
The record is insufficient.
A valid contact with no relevant buying role. A financial figure with no reporting period. A facility without its ownership context.
And a correct email address does not automatically make the person relevant to your campaign.
The U.S. Government Accountability Office assesses data reliability through accuracy, completeness and applicability to the intended purpose. U.S. GAO: Assessing Data Reliability ↗
The record is a snapshot.
The market keeps moving.
People change jobs. Practices relocate. Organizations acquire businesses. Clinicians add and leave affiliations. Email domains and recipient systems change.
There is a sequence between a real-world change and usable intelligence. Each step can introduce delay:
- 01 →Change happens
A person moves or an organization changes.
- 02 →Evidence appears
The change becomes observable.
- 03 →Data is collected
Available sources are gathered.
- 04 →Facts are checked
Conflicting information is reviewed.
- 05 →Records connect
Identity and relationships are reconciled.
- 06Product updates
The reviewed change reaches the user.
An executive may start a role before announcing it. A practice may relocate before directories catch up. A facility’s latest report may describe an earlier period.
Systems can also hold different, valid answers: a mailing address, a practice location and a corporate headquarters are not interchangeable.
A provider may achieve very high accuracy for a defined sample on a particular date. It cannot credibly promise every field will remain accurate indefinitely.
Track the error, the correction, its business impact and whether it recurs. A formatting issue is different from linking the wrong person to an account. Severity matters alongside frequency.
“95% accurate” is not enough.
Ninety-five percent of what?
Email addresses? Current employers? Returned records? Every contact requested? The answer changes the meaning of the claim.
Higher percentage. Fewer usable records.
Bar lengths show usable records as a share of the original 1,000 requested. These are fictional examples, not vendor benchmarks.
Provider A has the higher usable share among returned contacts. Provider B produces more usable contacts overall. Neither is automatically better: your requirements, risk tolerance and budget determine the fit.
Online reviews can reveal recurring problems. They rarely establish a consistent industry-wide “clean data” average. Use the same sample and definitions to benchmark your actual market.
Six practical tests
- Define usable before sampling. Executive outreach may need a current role and appropriate contact channel. Physician targeting may need specialty, practice affiliation, location and identity.
- Compare the same targets. Include multiple care settings, geographies and organization sizes, plus difficult records such as recent acquisitions and multiple affiliations. Test providers at approximately the same time.
- Separate coverage from correctness. Measure missing fields, wrong values, duplicates and relationship errors separately. Keep unresolved records separate from confirmed errors.
- Stress-test the scores. Ask what the score predicts and whether higher-scored groups perform better on that outcome under comparable conditions.
- Follow known changes. Recheck verified job moves, relocations and acquisitions after 30, 60 and 90 days, or longer where the stated window requires it. Note when evidence became available and when the product changed.
- Count the cleanup cost. Include manual research, duplicate removal, correction work and wasted outreach when comparing value.
Cost per usable record
(Data cost + validation and cleanup cost) ÷ records meeting your requirements
Apply the same requirements to every provider.
Three scores. Three different questions.
| Score | What to ask | What not to assume |
|---|---|---|
| Record confidence | How well is this information supported? | That every field has equal certainty. |
| Email reliability | How does this category perform for email? | Guaranteed inbox placement or replies. |
| Purchase propensity | Which accounts merit investigation? | Confirmed budget or intent to buy your product. |
Ask whether a score is a category, ranking or calibrated probability; what evidence informs it; and when it is recalculated. If higher-rated groups do not perform better over a meaningful sample, investigate.
Healthcare data is relationship data.
A clinician may work at several sites. A facility may belong to a larger system. Depending on the purchase, influence may sit with a corporate owner, IDN, GPO, FQHC administrative team or local leader.
An accurate address does not explain which relationship matters. An NPI helps establish identity, but does not alone establish current patient-facing activity, buying authority or every affiliation.
Evaluate the connections between records as carefully as the records themselves.
Transparent about the data.
Accountable for improving it.
Intent.Health combines public, licensed and proprietary data, connecting organization, executive and HCP intelligence with affiliations, Decision Networks, intent signals, propensity mapping and contextual intelligence.
Users have reported strong experiences compared with other providers. We value that feedback. It does not mean every field is 100% accurate at every moment.
Our responsibility is to keep improving how information is collected, connected, assessed and updated. We strive to offer one of the cleanest and most actionable healthcare datasets in the market. That ambition should be tested against customers’ actual use cases.
Different attributes. Different update windows.
Intent.Health’s stated windows for changes to appear are shown below. The bars share a 180-day scale; they show the stated range, not an average.
When did the event occur? When did the source become available? When was the product updated?
These are operating windows, not fixed guarantees for every record. Source availability and verification affect timing. Financial, clinical and quality data may describe an earlier reporting period even when newly updated.
Available does not mean advisable.
Intent.Health reports the following email performance levels. These are reported performance figures, not guarantees for individual campaigns.
Reported deliverability. Continue to monitor your own results.
Reported deliverability. Use greater caution in selection and testing.
Not suited to mass email outreach. Treat as reference or exclusion information.
Sender reputation, authentication, sending patterns, recipient systems and message content also affect performance. Compare results using consistent definitions and conditions.
Accepted by a recipient server is not the same as placed in the inbox. Neither guarantees engagement. Verification does not establish permission to contact someone.
For delivery factors and complaint management, see Gmail sender guidelines ↗.
Why show low reliability addresses?
Visibility can help you avoid buying the same unsuitable address again from another source. A record may be useful for identity while remaining unsuitable for campaign activation.
Low reliability classifications may reflect historical complaint or blocking risk. Recognize that risk, exclude these addresses from mass outreach and do not treat their presence in another database as proof of quality.
Context helps you act.
It does not erase uncertainty.
Ishi, Intent.Health’s contextual AI, helps interpret available intelligence about an account, its people, relationships and signals.
AI does not make an old source current by summarizing it. Confirm high-stakes assumptions: who controls the purchase, whether a role is current and whether a signal represents a relevant opportunity.
Transparency in data. Transparency in pricing.
Customers should understand what they are buying, what it supports, where limitations exist and how access is priced. Intent.Health publishes its offers and pricing so teams can assess commercial fit alongside capability.
We apply the same principle to data: clear reliability categories, update windows and limitations are more useful than an unsustainable promise of perfection.
Explore published plans and pricing ↗Your provider evaluation checklist.
Tick each step once you have evidence. This tracks evaluation preparation, not a provider quality score. Your selections stay in this tab and reset when reloaded.
0 of 8 evaluation steps complete
The print button opens your browser’s print dialog. Choose “Save as PDF” to keep a copy.
Put the data to your test.
Show us your target segments and the attributes your team needs. Evaluate Intent.Health against the work you actually do.
Book a data walkthrough ↗Or email growth@intent.health.
Frequently asked questions
Can a GTM data provider guarantee 100% accuracy?
A provider may demonstrate complete accuracy within a particular sample at a particular time. That does not establish permanent accuracy across its entire database. People, organizations and contact details continue changing.
Does a bounced email mean the provider’s data is bad?
It is a reason to investigate. The cause may be an invalid address, a changed mailbox, a temporary rejection or a sender-related issue. Look at the failure reason, reliability category and pattern across a comparable sample.
Is verified email the same as guaranteed inbox placement?
No. Address verification, successful delivery and inbox placement are different measures. Verification also does not establish permission to contact someone or guarantee engagement.
How should I assess a data provider’s reliability scores?
Ask what each score predicts. Compare outcomes across score groups using similar audiences and conditions. Higher-rated groups should perform better on the specific outcome the score describes.
How often should healthcare data be refreshed?
There is no single interval for every field. Job changes, affiliations, ownership records and financial reports depend on different source and publication cycles. Evaluate performance for each important attribute.
Does unchanged data mean stale data?
No. Some facts remain stable for years. Test whether known changes appear, and whether verification dates and source periods are clear.
Why is relationship data important in healthcare GTM?
An accurate contact can still be the wrong person for a purchase. Affiliations, ownership and purchasing relationships help explain the wider decision environment.
Does Intent.Health claim 100% accurate data?
No. Intent.Health combines public, licensed and proprietary data and continuously works to improve its offering. Its approach emphasizes reliability categories, update processes, contextual intelligence and transparent expectations.
Why retain low reliability email records?
They can help users recognize an unsuitable address and avoid reacquiring it elsewhere. Treat them as reference or exclusion information, not as recommendations for mass outreach.