Why We Built MarketingSoda Refine: The Data Quality Gap Nobody's Closing

HubSpot has 289,000 customers. They collectively manage hundreds of millions of contact records. And not one of them has a purpose-built tool for measuring the quality of that data.

There are tools that enrich data. Tools that deduplicate it. Tools that validate email addresses. Tools that flag formatting issues. Each solves one dimension of a problem that has at least seven. And none of them answer the question that RevOps teams actually need answered: across my entire database, how good is my data right now, and where exactly is it failing?

This is the story of why we built MarketingSoda Refine. Not a product walkthrough — we will get to that — but the structural gap in the HubSpot ecosystem that made building it feel less like a business opportunity and more like an obligation.

The Gap: 289,000 Customers, Zero Quality Layer

The HubSpot ecosystem is one of the most vibrant software marketplaces in B2B. The App Marketplace lists over 1,700 integrations. You can find tools for enrichment, for email verification, for workflow automation, for ABM targeting, for conversation intelligence, for proposal generation, for virtually any operational function a revenue team needs.

What you cannot find is a tool that scores your data quality across multiple dimensions, gives every record a grade, tracks that grade over time, and tells you precisely which fields on which records are dragging your database down.

This is not a niche need. Data quality underpins every revenue motion a HubSpot customer runs. Lead routing depends on accurate firmographics. Campaign performance depends on valid email addresses and current job titles. Lead scoring depends on fresh, complete records. Attribution depends on clean contact-to-company associations. Pipeline forecasting depends on all of the above.

When data quality is poor, every downstream system degrades. But the degradation is invisible because it does not produce error messages. It produces slightly worse open rates, slightly lower conversion rates, slightly less accurate forecasts. The compounding effect is enormous — Gartner estimates $12.9 million per organization per year — but no single metric screams "your data is broken." The problem hides in plain sight.

What HubSpot Gives You (and Where It Stops)

HubSpot is not ignoring data quality. The platform includes several features that address pieces of the problem. But there is a ceiling to what they do, and understanding that ceiling is essential context for why Refine exists.

Data Quality Command Center is a diagnostic dashboard available in Operations Hub Professional and Enterprise. It surfaces property completeness rates, identifies formatting inconsistencies, and flags duplicate records. It is a useful starting point for understanding the shape of your data problems. But it is exactly that — a starting point. Command Center is diagnostic, not prescriptive. It tells you that 37% of your contacts are missing a job title. It does not tell you which of those contacts matter most, how that percentage has trended over the past six months, or what the composite quality of the records that do have a job title actually is. A contact can have every field populated and still be a low-quality record if half those fields are outdated.

Breeze Intelligence is HubSpot's enrichment product, launched in early 2025 as the successor to HubSpot Insights. It enriches contact and company records using a single proprietary data source. Coverage is approximately 40% for a typical B2B database — higher for US technology contacts, significantly lower for EU, APAC, or non-tech verticals. Breeze solves one dimension of data quality (enrichment coverage) for a subset of records in a subset of geographies.

Duplicate Management is a built-in tool that surfaces exact-match duplicate pairs. It is useful but limited — "Jonathan Smith" and "Jon Smith" at the same company will not be flagged. Fuzzy matching, probabilistic deduplication, and merge conflict resolution are outside its scope.

Formatting Rules in Operations Hub allow you to standardize field values — capitalizing names, formatting phone numbers. This addresses validity and consistency for specific fields but does not assess or score the overall quality of a record.

Each of these tools does something valuable. None of them compose into a system. There is no unified quality score, no grade per record, no trend tracking, no automated remediation workflow that fires when a record drops below a threshold. HubSpot gives you diagnostic instruments. What is missing is the quality engine.

Why Enrichment Alone Is Not the Answer

The most common response to a data quality problem in the HubSpot ecosystem is to buy an enrichment tool. This makes intuitive sense: if fields are empty, fill them. If data is stale, refresh it.

The problem is that enrichment solves one dimension of quality — field population — while leaving six others unaddressed. And the market has largely accepted this framing. Enrichment providers position themselves as data quality solutions. RevOps teams buy enrichment and check the "data quality" box. The gap persists.

Consider what enrichment does not solve:

Accuracy. Enrichment fills fields. It does not verify that the values it fills are correct. A provider that returns a job title of "VP of Marketing" for a contact who was promoted to CMO three months ago has populated the field inaccurately. The record looks complete. It is not accurate. Accuracy requires plausibility checking — detecting obviously fake names, disposable email domains, impossible phone formats, and values that do not pass a basic sanity test.

Freshness. Enrichment runs once. Data decays continuously. A record enriched 14 months ago has experienced a full cycle of B2B contact decay — job changes, company changes, phone reassignments. Without time-decay tracking that measures when each field was last verified and flags records that have aged past a freshness threshold, enrichment is a one-time intervention in an ongoing process.

Validity. A phone number field that contains "TBD" or an email field that contains "test@test.com" has a value. That value is not valid. Format validation — checking that phone numbers match expected patterns, that emails resolve to real domains, that postal codes match stated regions — is a separate quality dimension from whether a field is populated at all.

Consistency. A contact with a job title of "CEO" and a seniority field of "Individual Contributor" has two populated fields that contradict each other. A contact whose state field says "California" and whose area code is 212 has a cross-field inconsistency. Consistency checking requires examining relationships between fields, not just individual field values.

Uniqueness. Duplicate records fracture attribution, inflate list counts, and produce duplicate outreach that damages brand perception. Enrichment does not deduplicate. In many cases, enrichment makes duplication worse — enriching both copies of a duplicate record makes each copy look more legitimate independently, reducing the likelihood that a manual reviewer will catch the duplication.

Completeness scoring with priority weighting. Not all missing fields matter equally. A missing email address on a contact you intend to include in a nurture campaign is a blocking gap. A missing phone number on a contact you only reach via email is an inconvenience. Completeness scoring that weights fields by their operational importance produces a far more actionable signal than a simple "percentage of fields populated" metric.

This is the fundamental insight that led to Refine. The HubSpot ecosystem does not need another enrichment tool. It needs a quality engine that measures all seven dimensions, scores every record, and gives RevOps teams an objective, trackable baseline for their database health.

The Competitive Landscape: Everyone Solves One Piece

We did not build Refine because no one is working on data quality. We built it because everyone is working on a fragment of it, and no one is assembling the fragments into a system.

Breeze Intelligence provides single-source enrichment with approximately 40% coverage. No quality scoring, no deduplication, no freshness tracking. Credits expire monthly.

Clay is a powerful waterfall enrichment orchestration platform. It can sequence across 50+ data providers and is widely used by growth teams and agencies. But Clay is an enrichment workflow builder, not a quality scoring system. It fills fields. It does not grade records, track quality trends, or tell you which records in your database are operationally reliable and which are not. The Pro tier starts at $800/month and the learning curve is steep.

Apollo and Cognism are enrichment databases with strong coverage in specific geographies — Apollo in North America, Cognism in EMEA. Both are excellent at what they do. Neither measures data quality. Neither deduplicates. Neither tracks freshness or scores records across multiple dimensions.

ZoomInfo is the largest B2B data provider by revenue. Its database is deep and its coverage is broad. It is also priced for enterprise buyers — plans start north of $15,000 per year — and is architected as a standalone database, not a HubSpot-native quality layer. For teams running their revenue operations inside HubSpot, ZoomInfo is a data source, not an operations platform.

Insycle is the closest existing tool to what Refine does. It provides bulk data cleansing, deduplication, and standardization for HubSpot. Insycle is a genuinely useful tool for data remediation. Where it stops is scoring. Insycle cleans data but does not grade it. There is no per-record quality score, no composite quality metric, no automated quality monitoring that tells you whether your database is improving or degrading over time.

Koalify provides data quality monitoring and alerting. It flags issues but does not remediate them — no enrichment, no deduplication, no standardization.

The pattern is consistent. Every tool in the ecosystem does one thing well. Enrichment tools enrich. Dedup tools deduplicate. Validation tools validate. Monitoring tools monitor. None of them compose these capabilities into a unified quality score that a RevOps leader can present in a board meeting and say: "Our database quality is a B+, up from a C last quarter. Here is where we improved and here is what we are working on next."

The Seven Dimensions of Data Quality

Refine scores every contact record in your HubSpot database across seven weighted dimensions. The composite score produces an A through F grade. Here is what each dimension measures and why it matters.

Completeness (20% weight). What percentage of operationally important fields are populated? Refine distinguishes between required fields — email, company name, job title — and optional fields, weighting accordingly. A record missing an email address scores lower on completeness than a record missing a LinkedIn URL, because the operational impact is different.

Accuracy (20% weight). Are the values in populated fields plausible and correct? Accuracy checks include detecting obviously fake names (e.g., "Test User," "asdf asdf"), identifying disposable email domains, flagging phone numbers with impossible formats, and cross-referencing field values against known plausibility constraints. A complete record with inaccurate values is more dangerous than an incomplete record, because it creates false confidence.

Freshness (15% weight). When was each field last verified or updated? Refine applies time-decay brackets — data verified within 90 days scores highest, 91-180 days scores lower, 181-365 days lower still, and data older than a year is flagged as high decay risk. This dimension surfaces the records that look complete but are silently aging into inaccuracy.

Consistency (15% weight). Do field values agree with each other? Cross-field consistency checks catch contradictions: a "CEO" job title paired with an "Individual Contributor" seniority, a California mailing address with a New York area code, a company listed as a 10-person startup with an enterprise-tier HubSpot license. Inconsistency flags records that need human review even when individual fields look reasonable.

Uniqueness (10% weight). Is this record a duplicate? Refine uses probabilistic record linkage — moving beyond exact-match deduplication to catch fuzzy matches, nickname variations, and records that share a company and similar name but differ in email domain due to an acquisition or rebrand. Duplicate clusters are identified and surfaced for merge or suppression.

Validity (10% weight). Do field values conform to expected formats? Email syntax, phone number patterns, postal code formats, URL structures — validity checking ensures that populated fields contain structurally correct values, not just non-empty ones.

Enrichment Coverage (10% weight). What percentage of the record has been enriched by a third-party provider? This dimension measures how much of the record came from verified external sources versus manual entry, form fills, or imports. Higher enrichment coverage correlates with higher accuracy, because provider-enriched data is typically more current than self-reported data.

The composite score is not an abstraction. It is the answer to a question that every RevOps team asks and no existing tool answers: how good is this record, really?

Why HubSpot-Native Matters

There is a design decision embedded in Refine that is worth explaining, because it is a deliberate constraint we chose and it shapes the product.

Refine operates natively inside HubSpot. It reads from HubSpot properties, writes scores back to HubSpot custom properties, and triggers remediation through HubSpot workflows. There is no separate application to log into, no external database to sync, no CSV exports to reconcile.

This matters for three reasons.

Operational simplicity. The teams that need data quality tooling most acutely — mid-market RevOps teams running lean, often a team of one or two — are the teams least able to manage another integration. Every external tool is another login, another sync to monitor, another failure point in the data pipeline. By operating inside HubSpot, Refine eliminates the integration tax entirely.

Real-time scoring. Because Refine reads directly from HubSpot, quality scores update as records change. When a workflow enriches a contact, the quality score recalculates. When a contact's email bounces, the quality score reflects it. This is fundamentally different from tools that pull a snapshot, analyze it externally, and push results back on a schedule. Scheduled syncs create a window where your quality scores are stale — which is ironic for a tool that is supposed to measure staleness.

Workflow integration. HubSpot Workflows are the automation layer that RevOps teams already use for lead routing, lifecycle management, and campaign enrollment. By writing quality scores to HubSpot properties, Refine enables workflow triggers based on data quality. You can build a workflow that suppresses contacts with a grade below C from campaign enrollment. You can route contacts with a grade of D or F to an enrichment queue. You can trigger a re-enrichment workflow when a contact's freshness score drops below a threshold. The quality score becomes an operational input, not just a reporting metric.

Waterfall Enrichment, Built In

Refine includes waterfall enrichment because quality scoring without remediation is a dashboard that makes you feel bad.

The enrichment engine cascades through multiple data providers — Clearbit, Apollo, ZoomInfo, Hunter, Cognism, and others — in sequence. If the first provider cannot fill a field, the second provider tries, then the third. Coverage rates with three or more providers routinely reach 80-90%, compared to the 40% ceiling of single-source tools.

Refine orchestrates the cascade intelligently, applying field-level survivorship rules and quality scoring to every enrichment decision.

After enrichment runs, quality scores recalculate. You can see exactly how much each enrichment pass improved your database quality — not just how many fields it filled, but how the composite scores moved across all seven dimensions.

What Comes After Clean Data

Refine is the first module in the MarketingSoda platform. It is not the last.

Signal is the second module — an ABM and intent data layer that identifies which accounts in your database are showing buying signals. Signal depends on Refine because intent data applied to a dirty database produces dirty intent signals. If your ICP definition is built on inaccurate firmographics, your intent scoring inherits that inaccuracy. Clean data first, then signal.

Scope is the third module — ICP targeting and segmentation that uses quality-scored, signal-enriched data to build high-precision campaign audiences. Scope depends on both Refine and Signal because segmentation is only as precise as the data and signals it operates on.

The sequence is deliberate. Every marketing automation capability in the market assumes clean data as a precondition but does not provide it. We are building the precondition first.

The Economics

Refine starts at $99/month for Starter (up to 10,000 contacts), $349/month for Pro (up to 50,000 contacts), and $999/month for Scale (unlimited contacts with dedicated support).

To put that in context: the average cost of poor data quality is $12.9 million per organization per year. Even at the Scale tier, Refine costs less than $12,000 annually. The ROI math does not require heroic assumptions. If Refine prevents a single misrouted enterprise lead from aging in a default queue — a lead that would have converted at a 21x higher rate if contacted within five minutes — the tool has paid for itself for the quarter.

The more relevant economic comparison is against the assembled cost of solving each quality dimension separately. An enrichment tool at $300-800/month. A dedup tool at $100-300/month. A validation service at $50-200/month. A monitoring tool at $100-400/month. The total stack cost for partial coverage across multiple tools routinely exceeds what Refine costs for unified coverage across all seven dimensions.

Why Now

Three things converged to make this the right moment to build Refine.

HubSpot's market has matured past the "just get data in" phase. When HubSpot had 50,000 customers, most were in growth mode — acquiring contacts as fast as possible, worrying about quality later. At 289,000 customers, a significant and growing portion of the base has large, mature databases where quality is the binding constraint on campaign performance, not volume.

Breeze Intelligence established the category but left the gap visible. By launching Breeze and positioning it as HubSpot's data quality answer, HubSpot validated the need. By limiting it to single-source enrichment without quality scoring, HubSpot also made the gap between what customers need and what they can get inside the ecosystem more visible than ever.

The RevOps role has professionalized. Five years ago, "data quality" was an IT concern or an ops afterthought. Today, RevOps is a strategic function with budget authority, board visibility, and accountability for metrics that directly depend on data quality. The buyer for a product like Refine now exists in a way that it did not three years ago.

Get Your Baseline

Every data quality improvement starts with knowing where you stand. Refine provides a free database health audit — connect your HubSpot via OAuth, and receive an A through F grade distribution across your contact database in under 60 seconds. No data is extracted or stored. No credit card required.

If you are a RevOps team managing a HubSpot database and you have ever wondered whether your data quality is actually as bad as you suspect, the audit will give you a concrete, quantified answer.

Run your free audit: marketingsoda.ai/audit

Join the Refine waitlist: marketingsoda.ai

We are launching to select companies in mid-2026. The waitlist is where early access begins.