Most teams discover their intent data has a freshness problem about three months after they bought it. Reps keep receiving the same high-intent accounts, the accounts do not respond, and RevOps cannot explain whether the source is detecting current research or replaying old activity. By then, the renewal conversation is already close.
Intent data can fail in several less visible ways. A signal may be stale, drawn from a category that attracts many casual researchers, matched to the wrong company, or attributed to a person who is not part of the buying group. Each problem can make a source look precise in a dashboard while reducing its usefulness in a live sales motion.
The right question is not whether a provider has a large volume of signals. It is whether the source helps your team make better account decisions than it would make using firmographic filters, existing engagement data, and cold prospecting alone. That requires measuring signal quality at the point where it affects sales activity.
Start with the failure modes your team can see
Sales teams often experience poor data as a workflow problem before they recognize it as a data-quality problem. A rep opens the daily list and sees accounts that appeared yesterday, last week, and the week before. Another rep sees a company marked as researching a product category even though the company has no obvious reason to buy it. RevOps notices that the source produces many accounts but very few opportunities.
These observations are useful because they connect the data pipeline to an operational outcome. A source is not automatically valuable because it detects more activity. It is valuable when the activity is recent, relevant, correctly assigned to an account, and associated with a higher probability of a useful sales conversation.
Before comparing vendors, define the decision the signal is supposed to improve. Is it helping reps prioritize a named account list? Is it identifying accounts for outbound campaigns? Is it routing active research to an existing owner? Different use cases require different thresholds for freshness, coverage, and confidence.
Measure freshness as signal decay
Web activity has a half-life. A page visit or content interaction that happened yesterday may justify a timely, relevant touch. The same event from six weeks ago may only indicate that someone once had a problem related to your category. If your system treats both events as equivalent, the resulting score will overstate current buying readiness.
Ask the provider for event timestamps, not just a current intent score. Then group accounts by the age of their most recent qualifying signal and compare outcomes by age band. For example, you might compare accounts with activity in the last seven days, eight to 30 days, and more than 30 days. Measure reply rates, meetings, opportunities, or another defined sales outcome for each group.
A useful source should show some decay in predictive strength as the signal gets older. The exact curve will differ by product category. A short-cycle purchase may lose value quickly, while a complex enterprise evaluation may remain relevant for longer. What matters is that your scoring and routing rules reflect the observed curve.
Also measure refresh behavior at the account level. A provider may report a large number of new events while continuing to rank the same accounts because it is adding activity to an old record. That is not necessarily wrong, but it should be visible. Track the percentage of high-intent accounts that are new to the list, the percentage that reappear, and the number whose rank changes meaningfully from one period to the next.
One team ran this simple comparison after purchasing an intent source. Its “high-intent” list looked remarkably similar week over week, with no meaningful variation in the top accounts. The team initially assumed its market was simply stable. After checking event dates, it found that the source was recycling old activity rather than refreshing the account set at the expected rate. The problem was not the length of the list. It was the age and repetition of the evidence behind it.
Separate purchase intent from generic browsing
Category activity is not the same as purchase intent. People browse broad topics for research, education, compliance, hiring, troubleshooting, or general curiosity. A high volume of visits to content related to your category can therefore produce an impressive list with little buying relevance.
To test for this, split the source’s signals into narrower and broader topics. Compare outcome rates for accounts showing activity on product-specific pages, implementation topics, pricing or comparison content, and technical requirements against accounts showing only generic category consumption. You are looking for an incremental lift, not a large number of events.
Sampling bias can hide inside this problem. If the underlying panel over-represents a few popular publishers or content categories, the source may be especially good at finding companies that consume those materials, not necessarily companies that fit your ideal customer profile. Review signal distribution by topic, publisher type, geography, company size, and industry where the provider can disclose it.
Then compare that distribution with your actual target market. If most of the activity comes from a category that your customers rarely prioritize, the source may be measuring content consumption more than commercial interest. The appropriate response may be to adjust the topic taxonomy, reduce the weight of generic signals, or limit the source to a narrower account segment.
Low signal quality is not always a vendor problem. Sometimes the underlying web panel is solid but the ICP filter is too broad to give resolution. A source can accurately observe activity across a large market and still produce weak recommendations if your definition of a target account includes too many industries, sizes, use cases, or regions. Check the quality of the selection logic before concluding that the collection layer is defective.
Test account matching resolution
Intent only helps if activity is assigned to the right account. Much web activity begins as an IP address, a cookie, a device identifier, or an organization-level observation. Turning that observation into a company record involves matching and enrichment. The match can be exact, probable, shared, or unresolved.
Ask how the provider distinguishes a corporate network from a home connection, a shared office, a cloud provider, a university, or an internet service provider. Ask how it handles subsidiaries, parent companies, branch offices, remote employees, and companies using shared infrastructure. These details affect whether a rep receives a useful account signal or contacts an organization that was never involved.
Measure match quality against a sample you can verify. Select accounts with recent activity and compare the provider’s company assignment with known website domains, CRM records, current ownership, and account headquarters. Track exact matches, plausible matches, wrong matches, and unresolved records. You do not need perfect resolution, but you do need to know how uncertainty is represented in the score and in the workflow.
Resolution also has a coverage tradeoff. A strict matching process may return fewer accounts but with higher confidence. A broader process may identify more activity while introducing more noise. The right balance depends on your sales capacity. A small team with limited research time may benefit more from a shorter list with clear account ownership than from a large list that requires manual investigation.
Check whether the signal belongs to your buyer
Even a fresh, correctly matched signal can come from the wrong person. A developer, student, job candidate, consultant, partner, or existing customer may research a topic related to your product. The account may be relevant, but the activity may not represent a buying motion.
Attribution should therefore be evaluated at two levels. First, does the activity belong to the right account? Second, does it provide evidence about a relevant buying group or business problem? Many sources cannot identify an individual, and that is acceptable if the account-level signal is paired with useful context. It becomes a problem when anonymous activity is presented with more certainty than the data supports.
Compare signal performance by account role and known engagement path where you have the data. Accounts that later reply, accept a meeting, or create an opportunity may show a different combination of topics, recency, and first-party engagement than accounts that never progress. Use those differences to adjust how you prioritize signals, rather than treating every event as an equal indication of intent.
Be careful with causality. A company may appear high intent because it was already well known to your team, actively engaged with your website, or included in an existing campaign. The intent source may be confirming activity rather than discovering a new opportunity. That confirmation can still be useful, but you should measure it separately from net-new account discovery.
Run an audit before you renew or buy
A short audit can reveal most serious quality problems without requiring a full data-science project. Use a recent sample of accounts and compare the source with your CRM and sales outcomes. Keep the evaluation window consistent, and define one primary outcome before looking at the results.
- Freshness: Record the age of the latest qualifying event and measure outcome rates by age band.
- List turnover: Compare this week’s high-intent accounts with last week’s list. Investigate repeated accounts and repeated events.
- Topic quality: Separate specific buying or implementation topics from generic category browsing.
- Account resolution: Verify company domains, parent-child relationships, geographic assignments, and ownership for a sample of records.
- Buyer relevance: Check whether the activity aligns with the roles, use cases, and buying groups your team actually serves.
- Incremental lift: Compare signal-led accounts with a similar cold or firmographic-only group using meetings, replies, opportunities, or another agreed outcome.
- Operational cost: Count how much manual research reps need before they can act on a signal.
Do not evaluate only precision. A highly selective source may identify a small number of good accounts while missing most of the market. Measure coverage alongside precision, then decide whether the tradeoff fits your team’s capacity. Also compare performance by segment. A source can be useful for mid-market technology accounts and weak for smaller professional-services firms, or the reverse.
There is a counterpoint to all of this measurement. Over-indexing on quality metrics can lead teams to wait for perfect data and never act. Some signal degradation is acceptable if the base rate of true positives is still higher than cold prospecting. The practical standard is not clean data in isolation. It is better prioritization at an acceptable operating cost.
For a small pilot, set a decision rule in advance: continue if signal-led accounts produce a meaningful lift in the chosen outcome, the lift holds across more than one time period, and reps can act without excessive verification. Stop or revise the program if list repetition, stale events, poor matching, or low buyer relevance erase that lift.
At Highperformr, our early-access pilot cohort has seen accounts ranked high intent convert to first meetings at 2.3x the baseline rate. That is an internal result, not a universal benchmark, which is why we recommend measuring the same relationship in your own sales motion. The useful question is whether your signal source improves decisions under your actual market, routing rules, and follow-up process.
Before renewing or buying, require evidence for these five points: the event is recent, the topic is specific, the account match is credible, the buyer relevance is defensible, and the outcome rate beats your existing baseline. If the source cannot expose enough detail to test those conditions, treat the lack of visibility as part of the buying decision.
If you want to pressure-test how we handle signal freshness and account resolution at Highperformr, reach out at [email protected].