Key Takeaways
- Vendor-reported match rates are measured on vendor-chosen samples. On 25 real company domains we owned, contact enrichment returned 16 verified contacts, a 64% hit rate rather than the 80-90% most planning assumes.
- Enrichment misses are not randomly distributed. Nonprofits and hospitality returned nothing, as did construction and small family-run firms at 30-50 employees, while every technology company on the list resolved.
- Enriching before you filter costs less than filtering on a proxy and enriching the survivors. On one list build, filter-first burned 168 credits with roughly three quarters spent on companies that never qualified.
- Instantly's documentation confirms no credits are charged when a verified work email is not found, so a miss rate costs you list size rather than money.
- The number worth tracking is cost per record that changed what you sent, not cost per record returned.
What data enrichment tools actually do
A data enrichment tool takes a thin record and fills in the fields you are missing. You hand it a company domain or a person's name, and it returns a work email, a job title, a headcount, an industry, a funding round, or a technology signal. Every product in the category does some version of that. What separates them is where the data comes from, how many sources get checked before the tool gives up, and what happens to your bill when a record comes back empty.
That last question gets almost no attention in buyer's guides. It is the one that decides your real cost.
Enrichment usually gets bought for one of two jobs. The first is repair: your CRM holds 4,000 contacts and a third of them carry a job title from three years ago. The second is construction: you have 200 target accounts and no named contact at any of them. Different jobs, different failure modes, and a tool that handles one well is frequently mediocre at the other.
The repair job never finishes. HubSpot's database decay tool puts the rate at about 22.5% a year, citing MarketingSherpa research that measured B2B data degrading 2.1% per month. Nothing you buy today stays accurate. Enrichment is a subscription to an ongoing maintenance problem rather than a purchase that closes it.
The cost of ignoring the problem is well documented. Gartner's data quality guidance states that poor data quality costs organizations at least $12.9 million a year on average. That figure describes enterprises, not a 15-person B2B firm. The mechanism scales down cleanly though: bad records produce wasted sends, misrouted leads, and sales time spent on companies that were never a fit.
The number every buyer's guide leads with
Search the term and you get roughly ten listicles ranking vendors by coverage and match rate. Nine of those ten are published by a company that sells enrichment.
Match rate is a real measurement, and the problem is not that vendors lie about it. The problem is who ran the test and on what. A vendor reporting 95% coverage measured that against a sample they selected. Point the same tool at your list and the number moves, sometimes by thirty points, because coverage is not spread evenly across the market. It clusters.
There is a second issue. Match rate counts records that came back with something in the field. It does not count records that came back with something correct, and it does not count records where the correct answer changed nothing about what you were going to send. Those two are what decide whether the spend earned anything.
What a real list actually returns
We ran 25 company domains we already owned through Instantly's SuperSearch enrichment, asking for one verified contact per company. Sixteen came back.
That is 64%, a long way under the 80% to 90% that most list planning assumes. The shortfall was not random. It clustered by company type. Nonprofits and hospitality returned nothing at all, as did construction firms and small family-run businesses in the 30 to 50 employee range. A rural health network came back empty. So did an air ambulance charity, a yacht club, a civil works contractor. Every technology company on the list resolved on the first pass.
That pattern is more useful than the headline number, because it tells you where enrichment quietly fails. Public contact data is thick around companies that hire in public and publish real leadership pages. It is thin everywhere else. If your ICP sits away from technology, plan on 60% to 70% and expect the gap to land exactly where you least want it.
Before you panic about the miss rate, one detail matters. Instantly's waterfall enrichment documentation states that no credits are consumed when a verified work email is not found for a lead. The published rates are 1 credit when Instantly finds the address itself and 2 or more when it arrives through a data partner, with named providers priced individually: BetterContact at 2 credits, ContactOut at 4. Under that model a 36% miss costs you nothing in money. It costs you in list size, which is the constraint that actually binds when you are trying to fill a calendar.
Where the budget actually leaks
Four failure modes account for most of the enrichment waste I see in B2B teams.
Filtering on a proxy, then enriching whatever survives. This is the expensive one and it looks like prudence. On one list build we had no headcount field, so we filtered on a proxy for company size, then enriched the 84 companies that passed. It cost 168 credits, and roughly three quarters of that went to companies that turned out not to qualify once real headcount arrived. Profile-enriching all 168 candidates first, at half a credit each, would have cost 84 credits and produced a clean list. When the qualifying attribute can be bought for well under a cent per record, approximating it is a false economy. Enrich first, filter second, then buy emails only for what is left.
Filters that fail silently. Some platforms ignore unrecognized filter keys without raising an error. Measured against a baseline query returning 363,014 leads, several plausible-looking size parameters returned exactly 363,014 again. The search ran without them and handed back an unfiltered result that looked like a successful narrow search. Build a list believing your size gate worked and you will pay to enrich every company size on the market. Audit this before spending: run the count once per filter key with that key removed, and if the total does not move, the key did nothing.
Buying fields that never change the message. Full-profile enrichment is cheap per row and easy to switch on for everything. If your campaign filters already guarantee industry, size, and revenue, paying again to have those returned buys back data you supplied. Keep the enrichment types that feed a real downstream decision, such as routing or cohort assignment, and switch off the rest.
Joining the results back on a raw domain string. Enrichment does not return the domain you uploaded. It returns the provider's resolved company record, so the same company comes back as nth.com, www.nth.com, or a subdomain, and the contact's email domain is frequently different again after a rebrand. A naive lookup drops those rows without telling you, and a silent drop means somebody receives the wrong segment's message. Normalize both sides to the registrable domain before matching, and write every failure to a review file instead of discarding it.
The categories, and which job each one fits
Single-source databases. One vendor, one index, one price. Fast and predictable, and the ceiling is whatever that vendor's index covers. Good for the construction job when your ICP sits inside their strong coverage. Our Clay and Apollo comparison walks through where that ceiling shows up in practice.
Waterfall aggregators. These query several providers in sequence and stop at the first verified hit, which lifts coverage above any single source and makes per-record cost variable rather than fixed. This is usually the right default for contact discovery. We cover the mechanics in more depth in our guide to waterfall enrichment.
Workflow platforms. Tools like Clay sit above the data layer and let you chain lookups, conditional logic, and AI research into one table. Powerful, and the pricing model is worth reading closely. Clay's own Actions and Data Credits documentation states that each fully enriched record typically costs 6 to 20 data credits depending on which data types you request and whether you run waterfalls across multiple providers, with phone numbers noted as expensive relative to emails. A twentyfold swing per record means your cost forecast depends entirely on which columns you switch on.
Point enrichers. Single-purpose products for one field, usually a technology signal, a funding event, or a job posting. Worth adding when that specific field drives your targeting, and worth checking for coverage before you build copy around it. Website technology scanning, for example, only sees what runs on a public page. Anything behind a login leaves no trace.
How to actually run the evaluation
Stop comparing published match rates and run a controlled test. It takes an afternoon.
Pull 50 companies from your own list that represent your actual mix, including the unglamorous segments. Run the same 50 through each shortlisted tool with identical field requests. Then measure four things.
Coverage on your sample, not theirs. Accuracy, which means manually verifying 15 returned records against the company's own site or LinkedIn, because a returned field and a correct field are different populations. Total spend across the run. And the number that actually matters: how many records came back with information that changed what you would have sent. A verified email for a company you were already going to contact with the same message is a rounding error. A title correction that routes a founder into a different sequence is the whole point.
Divide total spend by that last count. That is your cost per usable record, and it is the only figure that compares tools honestly across different pricing models.
Where enrichment sits in the system
Enrichment is plumbing. It does not book anything by itself, and buying a better tool does not fix a pipeline problem that lives somewhere else.
It matters because three other things depend on it. Your outbound targeting is only as good as the fields you filter on, which is why we treat data quality as part of the Outreach Engine rather than a separate purchase. Your CRM follow-up routes on those same fields, so a stale title sends a founder into a sequence written for an operations manager. And your reporting inherits every gap, which is how teams end up unable to explain why one segment converts and another does not.
If you want the fuller picture of how enrichment, routing, and follow-up wire together, our AI automation service page covers the build, and the AI automation playbook walks through the workflows we deploy most often. For the strategic version of the same question, B2B data enrichment covers why the data layer decides what the rest of the system can do.
Pick the tool that produces the lowest cost per usable record on your own list. Then spend the time you saved on what happens after the record lands, because that is where most B2B teams are actually losing conversations.
