Resonate Install: FAQ

Resonate Install delivers pre-scored, population-level consumer data into your own infrastructure, giving you individually addressable, predictive attributes on motivation, values, and behavior across the full U.S. consumer universe.

Below are answers to the most frequently asked questions about how the Install works. If you have additional questions, contact your Customer Success Manager or reach out to resonatesupport@resonate.com.

Jump to a question: 

• How is the Install different from Resonate Append?
• What ID types are supported?
• How big are the Output Files?
• What infrastructure do I need to use the Install?
• Why are Azure Blob Storage, GCS, and Box not supported?
• How often is the Install refreshed, and can I get faster refreshes?
• Why is each refresh a full replacement rather than an incremental update?
• Should I retrain my models with every wave?
• Why might attribute predictions for the same ID change between waves?
• Why are only some of Resonate's 15K+ attributes available via the Install?
• Why are some attributes redacted from certain records?
• Are sensitive attributes available for the rest of my universe?
• Are Install attributes deterministic or predictive?
• How does Resonate ensure the accuracy of predicted attributes?
• Are Install attributes privacy-safe? 

How is the Install different from Resonate Append? 

Both products deliver Resonate consumer attributes — the difference is where the work happens. The Install gives you a population-level file in your own environment; you do the joining, modeling, and activation in your own tools. Append takes your records, matches them in Resonate's environment, and returns enriched records to you. Many Install customers also use Append for specific record-level enrichment workflows. See the Install vs. Enrichment table in the Introduction to Resonate Install article for a side-by-side comparison.

What ID types are supported? 

Today, Resonate’s install supports: 

• HEM (Hashed Email — SHA256 default, SHA1, or MD5)
• MAID (Mobile Ad ID — IDFA and AAID)
• IPv4
• ZIP11

Each Install engagement is indexed by one or more ID types selected at contract initiation. Phone, IPv6, UID2, RampID, and CTV/ACR identifiers are on the roadmap — talk to your account team for timing. See the What's Included and How to Set it Up article for full format and normalization details.

How big are the Output Files? 

Full-catalog file sizes vary by ID type: 

Selecting an attribute subset rather than the full catalog produces substantially smaller files — file size scales approximately linearly with the number of attributes. Files at the higher end of the ranges above are delivered partitioned with a manifest file. 

What infrastructure do I need to use the Install? 

A cloud data warehouse (Snowflake, BigQuery, Databricks, Redshift, or comparable), enough storage for the licensed file size with retention headroom (plan for ~2× during validation windows), compute capable of querying at the relevant scale, and a data engineering or analytics function to operationalize the file. If you don't have this, Resonate's Append or SaaS products are better fits — your account team can help you decide. 

Why are Azure Blob Storage, GCS, and Box not supported? 

The Install is delivered via Snowflake share, Amazon S3, or SFTP. Azure Blob and GCS aren't supported as direct delivery destinations at launch; Box doesn't scale to Install file sizes. Customers operating in Azure or GCP environments should select SFTP delivery and stage files into their own cloud storage. (Note: Snowflake delivery is supported into Snowflake accounts hosted on Azure and GCP — Resonate handles the cross-cloud replication transparently. See the Install Process article for supported Snowflake regions. 

How often is the Install refreshed, and can I get faster refreshes? 

The Install refreshes every 8 weeks, anchored to your first delivery date and aligned to Resonate's wave release schedule. Each refresh arrives within five business days of the corresponding wave release. The cadence is fixed by the underlying U.S. Consumer Study production cycle and isn't configurable on a per-customer basis — it reflects how often the underlying data genuinely changes.

For use cases requiring real-time or near-real-time data (website personalization at high frequency, inbound channel response, IVR), Resonate's API-based delivery is the right architecture rather than a file Install. Talk to your account team if you're uncertain which fits your use case.

In between refreshes, Ignite offers the opportunity to onboard records and continuously update your insights into your customers’ and prospects’ profiles. 

Why is each refresh a full replacement rather than an incremental update? 

The Resonate consumer universe is dynamic — IDs are added and removed between waves as the identity graph evolves, and attribute predictions for the same ID may shift to reflect updated signal. A full replacement is the simplest, most reliable way to keep your environment in sync with the current state of the data. See the Operating the Recurring Refresh article for pipeline patterns that handle full-replacement deliveries efficiently.

Should I retrain my models with every wave? 

That's a judgment call for your data science and analytics teams. Resonate's recommendation is to align retraining with refresh receipt for any model that uses Install attributes as features, but the right cadence depends on your model's sensitivity to feature drift. Some teams retrain every wave; others establish a quarterly or semi-annual retraining rhythm and treat intermediate waves as feature updates without full retraining. 

Why might attribute predictions for the same ID change between waves? 

Each refresh reflects the most recent wave of survey data and updated modeling. A customer scored as "Frequent Traveler = TRUE" in one wave may be scored differently in the next as their behavioral and motivational signal evolves. This is the point of refreshing — your predictions stay current with consumer reality rather than reflecting a static snapshot. See Operating the Recurring Refresh for more on what variation is normal between waves. 

Why are only some of Resonate's 15K+ attributes available via the Install? 

The Install catalog includes a curated subset of Resonate's full 15K+ attribute universe. Like Resonate’s Enrichment products, the Install requires a higher standard for modeled accuracy and consistency than the Ignite Platform, which has built-in governance flags for statistically weak insights in real time. The Install subset is the set of attributes that meet reliability and scale thresholds for use outside the platform. 

Why are some attributes redacted from certain records? 

Predictions for sensitive attributes — including those covering racial or ethnic origin, religious or philosophical beliefs, union membership, citizenship or immigration status, sexual orientation, and health information — aren't delivered for individuals residing in specific specified states. This is enforced at file generation and applies to the Output File only. Records aren't excluded from the file; the sensitive Survey_Value_Keys for those records are either listed in a Redacted column (Standard structure) or appear as NULL (Pivoted structure). See the Sensitive Data article for the full state list and how the restriction appears in your file. 

Are sensitive attributes available for the rest of my universe? 

Yes. The restriction applies only to individuals residing in the restricted states. For all other records, sensitive attributes are delivered alongside everything else. See the Sensitive Data article for the full state list and how the restriction appears in your file. 

Are Install attributes deterministic or predictive? 

Predictive. Install attributes are modeled — they describe what people are likely to think, feel, and do, not literal facts about specific individuals. Predictive attribution starts with a foundation of deterministic inputs (self-reported survey data from the U.S. Consumer Study, validated offline identity data, and 30B+ daily observed online behaviors) and uses machine learning to scale that understanding across the full U.S. adult population.

Deterministic data tells you who someone is — useful for compliance and identity verification. Predictive enrichment tells you why they act and what they'll do next — purpose-built for marketing scale, segmentation, modeling, and personalization. The Install delivers the latter at the granularity of individual records across the full addressable U.S. population.

How does Resonate ensure the accuracy of predicted attributes? 

Models are anchored in the U.S. Consumer Study (the foundation of self-reported motivations, values, and behaviors linked to real individuals), validated against 30B+ daily observed online behaviors plus offline sources like U.S. Census and credit bureau records, scaled across the U.S. adult population using Resonate's proprietary AI infrastructure (rAI), and weighted to census baselines with built-in quality thresholds for statistical confidence.

A modeled attribute isn't a literal truth about one individual — it's a statistically reliable signal across millions. In rare cases an individual's predicted traits won't align perfectly with what you know about them; that's expected and by design. The power of modeled data is behavioral similarity at scale, which deterministic data can't reach.

Are Install attributes privacy-safe? 

Yes. Resonate is privacy-first by design, SOC-2 certified, and operates in compliance with U.S. privacy laws. Install attributes are modeled using pseudonymous identifiers and ethically sourced behavioral data. The Install is also architecturally privacy-safe in a way that matters for many customers: no first-party data leaves your environment, because no first-party data is involved in production. Resonate delivers a population-level file; you do the joining on your side. Learn more about Resonate's privacy and compliance posture in the Trust Center.