Key takeaways
- Activation stays where you already govern it. Untitled data is available through five à la carte access methods: portal, REST API, private PostgreSQL warehouse, SFTP, or custom delivery. No one on your team is required to log into Untitled.
- Untitled resolves an average of 45–60% of addressable U.S. traffic against an identity graph of 300+ million U.S. person-level records. Non-U.S. traffic must be excluded from the denominator before calculating a match rate.
- Resolution is primarily deterministic, anchored to a hashed email address. In production environments with sufficient traffic, Untitled’s matching accuracy typically falls in the 90–95% range.
- Untitled does not support federated querying against your warehouse. Where data minimization is a requirement, we can sync with your first-party data using hashed email addresses as the sole matching key, which matches records without transferring directly identifiable PII. If you maintain native UUIDs, we can sync against those instead, typically via activity backfill or identity pixels from an enterprise CDP.
- Coverage is limited to U.S. adults aged 18 and over. Untitled resolves U.S.-based visitors only.
- The delivery feed runs every 15 minutes, and Untitled performs a comprehensive structural update to the identity graph every 30 days.
Most enterprise data teams evaluating identity resolution are not shopping for another platform. They are consolidating. They have a data warehouse at the center of the stack, a Customer Data Platform, or a composable CDP alternative, that queries it rather than copying it, and a mandate to reduce the number of interfaces their team has to learn.
The question they bring to a vendor is not what can you do.
It is where do you sit, and what do I have to change to make room for you?
This post answers that question in technical detail. It covers how Untitled resolves anonymous website traffic, how the identity graph is validated, what the system genuinely cannot do, and, most importantly for teams running a composable architecture, the five ways Untitled data can reach your stack without anyone on your team logging into Untitled.
The short version: Untitled works alongside a warehouse-native CDP as a data source, not as a competing activation layer. We do not expect a large organization to onboard its established CDP into our platform. That would be the wrong architecture.
What Untitled is, and what it is not
Untitled is a target audience platform and identity resolution provider. A single JavaScript tag, known as the Untitled ID Tag, placed on your website matches anonymous visitor signals against an identity graph of 300+ million U.S. person-level records, returning a resolved profile with contact and demographic attributes attached.
What distinguishes Untitled in an enterprise context is not the resolution itself. Instead, the resolved data is treated as a data asset you own and route, rather than as fuel for a proprietary activation engine.
Untitled offers marketing destinations (email tools, social, paid media platforms), and they are entirely optional. If we are in your marketing stack for data, the data is accessible by whatever method fits your architecture.
It produces one specific input that most stacks cannot produce on their own: the identity of the anonymous visitor.
How identity resolution works
A standalone website typically identifies 1–2% of its traffic using first-party cookies and user-submitted data alone. Everything above that number requires an identity graph and a way to connect a browser event to a record inside it.
Untitled’s resolution process runs in three phases. Each builds on the one before it.
The three-phase resolution framework. Verification establishes a deterministic link between a browser event and a hashed email address; efficacy expands that link through the identity graph; accuracy reinforces it through real-world activation.
Verification: When the Untitled ID Tag executes, it observes the IP address, device and browser characteristics, cookie identifiers, and timestamped engagement activity. The system triangulates those signals, comparing the observed IP against known IP footprint history, validating the device-to-record relationship, and aligning timestamps to establish a deterministic link to a hashed email address (HEM).
The HEM is the foundational identity anchor of the graph and the reason cross-device linkage and offline-to-online reconciliation are possible at all.
Efficacy: Once a valid HEM linkage exists, the system maps the online identifiers to demographic, firmographic, cross-device, and behavioral attributes. Validation here is distributed rather than record-by-record: cookie sync partnerships, email engagement IP footprint validation, mobile movement data, and log-level timestamp validation are evaluated together for signal consensus.
Records are continuously assessed for reinforcement, drift, and stale attributes, and confidence scoring adjusts when signals weaken or contradict prior associations.
Accuracy: The highest confidence occurs after activation. When a record is activated in a marketing channel, converts, and is confirmed by your first-party data, that feedback improves local confidence and makes subsequent enrichment more precise.
Why accuracy depends on signal density, not just matching logic
In production environments with sufficient traffic and activation, Untitled’s matching accuracy typically falls in the 90–95% range. That figure is environment-dependent, and understanding why matters more than the number itself.
Resolution improves proportionally with site scale and logarithmically with session frequency. More tag fires produce more triangulation opportunities, which produce higher confidence.
The practical thresholds: sites below roughly 1,000 monthly unique visitors exhibit more variability, and performance becomes materially more stable above roughly 5,000 monthly uniques.
Untitled’s matching is primarily deterministic and rule-based. That is a deliberate trade. Deterministic logic performs strongly in high-signal environments and avoids speculative modeling, but in low-volume settings it can produce what we call extraneous linkages, technically valid matches formed from partial signals where insufficient data exists to discriminate confidently between candidates.
Two implications worth knowing before you evaluate:
- QA and dev environments are the worst case. They have low traffic, sporadic tag fires, and no repeat behavioral patterns. Anomalies observed there frequently do not persist in production.
- Small filtered exports exaggerate error. A system running at 93% accuracy still produces roughly 1–2 anomalies in a 20-record export. That looks alarming and is statistically unremarkable. Manual verification against public web searches is also not reliable ground truth, particularly for private companies and individuals with limited public presence.
What limits identity resolution?
These constraints are worth stating before match-rate math, because they change the denominator.
Untitled’s identity graph is built exclusively from U.S. consumer records, and resolves U.S.-based visitors only.
If half your traffic originates outside the United States, that half must be excluded before calculating a match rate. Untitled resolves an average of 45–60% of addressable U.S. traffic, a figure that is only meaningful against the addressable portion, never against total traffic.
The graph covers U.S. adults aged 18 and over. Beyond geography and age, a consented visitor may still go unresolved because they are not present in the dataset, because signals were weak or conflicting and did not meet the deterministic threshold, or because of browser conditions (private browsing modes, VPNs, and proxies all reduce resolvability).
Untitled does not support federated querying against your warehouse. We cannot read from BigQuery, Snowflake, or Databricks and match in place at this time. What we do support is hashed email matching, described further below.
Personal phone numbers are locked by default. Access is governed by permissible-use review, including TCPA requirements that vary by state.
Fill rates vary by brand and category. We do not quote a fill rate before seeing your data. Establishing that number is one of the explicit purposes of the evaluation period.
Does third-party cookie deprecation break this?
No, not for Untitled, and the reason is architectural rather than optimistic.
The deterioration of third-party cookies was aimed at cross-context tracking systems, mechanisms that follow a user between unrelated properties. That is not the mechanism Untitled’s resolution primarily depends on. The primary data points and steps that produce a resolution are first-party based: a first-party cookie initiates a process involving several other signals.
This gives Untitled some immunity to that decay. Third-party cookies still appear in the process, but they are not contingent match types required to produce a resolution.
There is observable evidence for this.
Untitled typically captures 15–18% more mobile web traffic than a conventional advertising platform pixel, like the Meta Pixel, because that pixel is a cross-context third-party tracker and is natively blocked in some mobile environments.
One thing this does not do is override your consent framework. The Untitled ID Tag follows the consent signal your CMP provides and will not fire where a visitor has explicitly opted out of applicable data collection.
What comes back on a resolved visitor?
Resolution begins within roughly 15 minutes to one hour of the tag going live. Returned attributes include first and last name, personal email address (business email is sometimes present as well, so both may appear on a single record), job information where known, demographics, household financials, company-level information, work phone, company and professional address, personal address, and social profile links. This is true person-level identification.
Every attribute is available as a filtering dimension: age range, income range, net worth, homeowner status, and consumer-side filters can be applied cross-dimensionally with business-side filters for advanced customer segmentation.
How audiences work in Untitled
An audience in Untitled is a set of rules, not a snapshot. It is filtering logic across attributes, page paths, session behavior, and source parameters. Any future visitor who matches those conditions enters the audience automatically, without a re-export.
The lookback window functions as segment expiration: a 30-day lookback produces a 30-day rolling segment. All-time audiences are available, and nothing collected is discarded.
For more advanced audience targeting, available filtering dimensions include UTM parameters (exact match or contains), referring source, page URL and page-path sequences, session time, session frequency, scroll depth, and first-party include/exclude lists. There is no limit on audience count; some customers maintain several hundred.
One implementation note that matters for teams with tag governance queues: install the tag once, set it to fire on all pages, and you are done.
Untitled collects every page URL where the tag fires, and every URL becomes an available filtering dimension automatically. There is no need to parameterize the tag or build a unique tagging structure for each page you want to segment on later.
Why we recommend against syncing everything into your CRM
We will say this unprompted, because it is where most implementations go wrong.
Do not sync all resolved identity data directly into your email platform or CRM.
The volume is too high, particularly for large, well-established sites, and it degrades the systems your operations depend on most. Deliverability suffers, record counts inflate against licensing tiers, and the signal-to-noise ratio in your CRM drops for every team touching it.
The audience layer exists precisely so that segmentation happens before anything reaches operationally critical systems. Narrow first, then sync. If your data architecture routes through a data warehouse and a CDP, this is even easier. The filtering can happen on your side, in the tooling your team already governs.
Where Untitled fits in a warehouse-native architecture
This is the central question for enterprise data teams, and the answer is deliberately unglamorous: Untitled is a well-behaved source inside an architecture that already exists.
Untitled is CDP-friendly. Using our marketing destinations and integrations is entirely optional. If we are there for data, that data is accessible by whichever of five methods fits your stack.
Five access methods, offered à la carte. Enterprise teams most often start with the private PostgreSQL warehouse or the REST API and route everything through their existing CDP.
A note on each:
- Untitled portal: The full UI for audience building and activation. Useful for teams who want a workspace, unnecessary for teams who don’t.
- REST API: Any data in Untitled can be queried and accessed programmatically through the Untitled REST API. This is the common choice for engineering-led stacks.
- Private PostgreSQL warehouse: We provision a dedicated database containing your Untitled data, and you build an ETL pipeline from it into your own warehouse or CDP. This is the most common enterprise pattern.
- SFTP: Configured file delivery, set up with your customer success contact, for established file-based workflows.
- Custom delivery formats: We can stream into a designated system. It is not where we recommend starting, but it exists when nothing else fits.
Delivery into a customer’s warehouse of choice is well-trodden ground for us, particularly for proof-of-concept work.
Three ways enterprises deploy Untitled
Across enterprise accounts, deployments tend to settle into one of three patterns. They are progressive, with most organizations starting at the first and moving as trust and integration depth increase.
Deployment maturity. None of these stages requires replacing an existing CDP.
How you join Untitled records to your own
Two mechanisms, and they can be used together.
Hashed email matching: The hashed email address is the native identity marker of the graph. If the only data you can share is a hashed email, Untitled can match on that alone.
For teams with data-minimization requirements, where syncing first-party PII records to a vendor is off the table, this is typically a solution that both compliance and governance teams are comfortable with.
Customer-designated UUID sync: This is an enterprise feature that lets Untitled capture an identifier your systems already generate and attach it to the resolution record, giving both sides a shared join key. Learn more about Customer-designated UUID syncs here.
Mechanically: the Untitled ID Tag waits for the page to load, extracts the configured identifier from the page environment, appends it to the pageview payload alongside Untitled’s internal external_id, and stores the result in a table you can query.
Identifiers can be pulled from DOM elements or from JavaScript objects such as a dataLayer, which covers the activity-backfill IDs most CDPs generate, order IDs, form instance IDs, partner identifiers, and campaign metadata. Joining is then a straightforward operation on external_id against the resolutions table.
Two practical constraints: identifiers must be available at page load to be extracted reliably, and configuration is per-client, requiring 5–7 business days once implementation details are provided. The resulting dataset updates approximately every 15 minutes.
Exclusion audiences and your first-party data
Enterprise brands almost always have people they cannot or should not contact, like existing customers, restricted parties, do-not-contact lists, and active opportunities owned by another team. Two capabilities work together here.
Custom uploads: Offline data can be brought into Untitled on whatever cadence you prefer, without requiring a standard SaaS connector. CRM exports, suppression lists, restricted-party lists, and existing customer files all work.
Include/exclude as a native filtering dimension: Used exclusionarily, this produces clean conquest audiences with existing customers and restricted individuals removed, so a first-time-buyer offer never reaches someone already loyal and buying frequently. Used inclusionarily, you can build audiences from an existing customer file using dimensions like lifetime value or order quantity.
Worth stating plainly: first-party data onboarding is not a requirement of the system. It makes audience building richer and helps considerably on the orchestration side, but Untitled functions without it. If your governance posture is to keep first-party data centralized in the warehouse, that is a supported architecture, not a workaround.
Multiple brands, properties, and business units
This is one of the stronger enterprise patterns we see.
One long-tenured customer in the premium home and kitchen category runs four distinct brand properties under a single Untitled account: a direct-to-consumer eCommerce site, a commercial site selling into trade and institutional buyers, a connected-device product line stood up as its own brand, and a hospitality property.
Each property carries its own tag. All resolution lands in one place. The strategic value is cross-pollination. A customer who bought the high-end range but does not own the cookware is an obvious opportunity and an otherwise invisible one. Separate properties, separate audiences, one identity layer underneath.
The pattern generalizes cleanly to business units, sub-brands, regional properties, and acquired companies.
For organizations or performance advertising agencies using Untitled and managing many properties, the agency and multi-brand dashboard provides a consolidated view of tag health, activity trends, per-tag monthly resolution budgets, and user access across accounts.
How fresh is the data?
Two separate answers, and collapsing them produces a misleading picture.
Delivery cadence: The Untitled feed runs every 15 minutes. Larger sites take longer to sync all the way through. If you have a genuine sub-15-minute requirement, that warrants its own scoping conversation rather than an assumption.
Graph maintenance: Untitled performs a comprehensive structural update to the identity graph every 30 days. Between those updates, digital exhaust functions as a continuous verification system, the range of signals used to confirm a record is still valid, including email service feeds and real-time bidding feeds. Validation is built into the process rather than performed periodically against it.
The contrast worth drawing is with the traditional compiler model. Large legacy data brokers flow data one way, roll it up, and refresh on a quarterly cadence in theory and something closer to semi-annually in practice. Validation in that model leans heavily on third-party cookies, which provides no identity overlay and no meaningful cross-reference. We compete on validation methodology, not just recency.
Security and procurement
Briefly, because most readers evaluating this will hand it to a security team who wants primary documentation rather than a summary.
Untitled is a cloud-native SaaS platform hosted on AWS in United States regions, with no on-premises production infrastructure. Data in transit is encrypted using TLS 1.2 or higher; sensitive data at rest in databases and object storage is encrypted using AES-256, with keys managed through AWS KMS.
Multi-factor authentication is required for access to production systems, cloud infrastructure, and administrative tooling, under least-privilege role-based access control. Production data stores are backed up daily, with write-ahead logging supporting point-in-time recovery at approximately five-minute intervals.
What matters more than any single control: the documentation is public. The information security policy, observability and patch management policy with severity-based remediation SLAs, third-party subprocessor list, platform data flows, and the full technical detail of what the tag collects and which cookies it sets are all published in the Untitled knowledge base.
Enterprises generally expect to fight for this material. You should not have to.
Why we built Untitled this way
Untitled is eight years old. Before this platform existed, we operated as a full modern data stack consultancy, partners in the warehouse and ingestion ecosystem, with roughly 250+ modern data stack deployments delivered.
The platform exists because we kept delivering the same project. The recurring pattern was CDP-adjacent but fundamentally audience-centric, and it became unapproachable the moment identity graphing or unconventional data sources entered scope.
Untitled is the productized answer to that pattern, not a replacement for the modern data stack, but a component designed to behave well inside one. Our founding team’s background is in offline identity graphing, including research and development work on cookieless IP-to-household resolution and mobile device graph construction. The graphing technique is a core competency rather than an acquired capability.
That history is the reason the architecture looks the way it does.
We have been on your side of this table often enough to know that the vendor who insists on being the center of the stack is the vendor who eventually gets ripped out.
What an evaluation of Untitled looks like
The first 30 days with Untitled are a measurement period rather than a trial in the conventional sense. The goal is to establish three benchmarks: resolution rate against your addressable traffic, match and fill rates against your existing customer file, and the audience segments that become addressable as a result.
Days 0–3, setup. Stakeholder onboarding and access with MFA, tag placement, installation verification, and confirmation of your data access method. Initial data populates within approximately one hour; we recommend 3–7 days of accumulation before formal testing begins.
Days 3–10, data validation. Three parallel tracks: anonymous-to-known resolution validation, cross-over analysis against a verified set of your existing records, and incremental audience enrichment analysis.
Days 10–30, utility testing. Identity continuity and session behavior across sessions and devices, downstream activation in your existing channels, and personalization readiness.
The organizing question throughout is a simple one: could I have gotten this from my current provider?
Where the answer is no, that is net-new coverage, and net-new coverage is usually where the business case lives.
Frequently asked questions
No. Untitled is a data source that feeds your CDP, not a competing activation layer. We do not expect a large organization to onboard its established CDP into our platform. Untitled data can reach your stack through the REST API, a private PostgreSQL warehouse, SFTP, or custom delivery, so activation stays in the tooling your team already governs.
No. The Untitled portal is one access method among several and is entirely optional. Enterprise teams commonly adopt Untitled purely as a data layer, using the API or a provisioned PostgreSQL database and routing everything through their existing warehouse and CDP.
Untitled does not support federated querying against a customer's warehouse. What Untitled does support is hashed email matching — the hashed email address is the native identity marker of the graph, so if the only data you can share is a hashed email, that is sufficient to match. This satisfies most data-minimization requirements without moving raw records.
Untitled resolves an average of 45–60% of addressable U.S. traffic. The qualifier matters: the identity graph is built exclusively from U.S. consumer records, so non-U.S. traffic must be excluded from the denominator before calculating a match rate. A site with 100,000 monthly visitors where half the traffic is U.S.-based has 50,000 addressable visitors, and would expect roughly 22,500 to 30,000 resolved profiles.
Fill rates for individual attributes vary by brand and category, and we do not quote a fill rate before seeing your data.
No. Untitled deploys a JavaScript tag, not a tracking pixel. It is a single line of code that functions as a tag container, with several endpoints underneath it working together to produce a resolution. The architectural distinction that matters to a security reviewer is that a conventional advertising pixel is a cross-context, third-party tracking mechanism, while Untitled's resolution path is first-party based.
The script structure, the data it sends, and every cookie it sets are documented publicly in the Untitled knowledge base.
Install the tag once and set it to fire on all pages. Untitled collects every page URL where the tag fires, and every URL automatically becomes an available filtering dimension. There is no requirement to parameterize the tag or build a unique tagging structure per page — which means a single change-control ticket rather than an ongoing backlog.
Third-party cookie deterioration was aimed at cross-context tracking systems, which is not the mechanism Untitled's resolution primarily depends on. The primary path is first-party based, giving Untitled some immunity to that decay. Third-party cookies still appear in the process but are not contingent match types required to produce a resolution.
Supporting evidence: Untitled typically captures 15–18% more mobile web traffic than a conventional advertising platform pixel, which is natively blocked in mobile environments as a cross-context tracker.
Two separate cadences. The Untitled delivery feed runs every 15 minutes, though larger sites take longer to sync all the way through. Separately, Untitled performs a comprehensive structural update to the identity graph every 30 days, with continuous validation between those updates from digital exhaust signals including email service feeds and real-time bidding feeds.
No. Untitled's identity graph is built exclusively from U.S. consumer records covering adults aged 18 and over, and resolves U.S.-based visitors only. Brands with significant international traffic should exclude that portion when modeling expected resolution volume.
Yes. Each property carries its own tag while resolution lands in one place, which makes cross-property audience building possible. The pattern applies equally to business units, sub-brands, regional properties, and acquired companies. A multi-brand dashboard consolidates tag health, activity trends, per-tag resolution budgets, and user access across accounts.
Leveraging Untitled through the composable CDP
If you are mid-migration to a warehouse-native architecture or composable CDP and trying to work out whether an identity layer fits without disrupting what you have built, the fastest path is a scoping conversation about your specific stack. We’ll ask questions like which access method fits, what the join strategy looks like, and what your addressable traffic actually is.
Book a demo with our team, or start with the technical documentation in the Untitled knowledge base, which covers the tag structure, cookies and storage, identity resolution methodology, data delivery options, and our full procurement and security documentation.
Check Out Other Articles
- Person-Level Website Visitor Identification: How to See the Real People Behind Anonymous Traffic
- The Future of Website Visitor Identification: 7 Trends Reshaping How Brands Identify Anonymous Traffic in 2026
- Customer Data Platform Alternatives: Why Marketing Teams Are Moving to a Composable Stack
- How Identity Resolution Unlocks Customer Insights
- Website Visitor Tracking Software
- Understanding First-Party Data
