Hazel is the AI coworker for Northbeam. Reads your campaign and ad performance, attribution models, and windows, then answers business questions in plain English. No dashboards to build.
June 12, 2026
Hazel is an AI coworker that connects to your Northbeam account and answers business questions about your marketing attribution the way an analyst would. Once connected, Hazel ingests your campaign-level and ad-level performance, the attribution models Northbeam computes (Northbeam custom, last-click, and the rest), your attribution windows (1-day click, 7-day click, and others), and the full metric catalog (spend, attributed revenue, CPO, ROAS, orders), then refreshes daily. Ask Hazel about CPO by campaign on Meta, how 7-day-click accrual revenue compares to platform-reported numbers, or where Northbeam-modeled performance diverges most from what each platform claims, and you get an answer with the numbers behind it. Hazel joins Northbeam to your Shopify orders and your ad platforms, so you can reconcile Northbeam-modeled performance against platform-reported numbers and against Shopify-truth revenue, by attribution model and window, in one conversation.
Hazel ingests your Northbeam data and answers business questions conversationally:
Hazel keeps the full historical record, so model-over-model and window-over-window trend analyses are one prompt.
These are the kinds of questions Hazel answers about your Northbeam data:
How Hazel connects to Northbeam. Hazel connects through Northbeam's data API using credentials you authorize. There is nothing to install in your ad accounts and no manual export to schedule. Once the connection is live, Hazel begins ingesting your performance and attribution data automatically.
What syncs. Campaign-level and ad-level performance (spend, attributed revenue, CPO, ROAS, orders, and the rest of the metric catalog), the attribution models Northbeam computes, your attribution windows and the accounting mode behind each, and Northbeam's dimensional breakdowns. Performance lands in long format with reporting date, platform, status, model, and window attached, so you can pivot from a platform total down to a single ad.
How often. Daily, automatically.
How attribution windows work. Northbeam reports the same campaign or ad through multiple attribution models and windows at once. Hazel keeps each combination distinct, so a question about 1-day click CPO and a question about 7-day click accrual revenue return different numbers from the same ad, and you always know which model and window produced a figure.
Auth. Hazel connects via Northbeam's data API with credentials you authorize. Hazel only reads your data. It never writes back to your Northbeam account.
Looking for a Northbeam MCP for analytics? Northbeam ships its own API and data exports for getting performance and attribution data out of the platform. Those are great if you want to pull Northbeam data into your own warehouse and write the analysis logic yourself.
Hazel sits between your Northbeam data and your team's questions. Hazel is not another attribution model. It is the AI layer that reads your Northbeam data and joins it with your other sources (Shopify, Meta Ads, Google Ads, TikTok Ads), then answers business questions across all of them in plain English. Rather than wiring an LLM directly to Northbeam's API and reconciling models, windows, and platform-reported numbers by hand, you ask the question and Hazel works it the way an analyst would, with the attribution model and window made explicit in the answer.
Hazel MCP. Hazel also exposes its own MCP endpoint, so your team can connect Claude, ChatGPT, or any MCP-compatible agent directly to Hazel's analytical store. Book a call to set it up.
Yes. Hazel connects to your Northbeam account through Northbeam's data API and ingests campaign and ad performance, attribution models, attribution windows, and the full metric catalog (CPO, ROAS, spend, attributed revenue, orders) across Meta, Google, TikTok, Snap, and your other platforms.
No. Hazel does not compute its own attribution. It reads the models Northbeam already computes (Northbeam custom, last-click, and the rest) and answers questions across them, then joins those numbers to your Shopify orders and your ad platforms so you can reconcile modeled performance against platform-reported and Shopify-truth revenue.
Triple Whale, like Northbeam's own dashboards, is an attribution-first product you log into and read. Hazel is the AI coworker on top of your data. It reads your Northbeam attribution, your Shopify orders, and your ad platforms, then answers cross-source business questions in plain English. You do not learn a dashboard or pick a view in advance. You ask the question and Hazel works it across every source you have connected.
Daily.
No. Hazel only reads. We never write to your Northbeam account or change any setting.
Northbeam computes modeled, multi-touch attribution, which often disagrees with the last-click numbers each ad platform reports for itself. Hazel keeps both numbers and shows them side by side, so you can see exactly where and why the gap is largest, by campaign, platform, model, and window.
Yes. Hazel ingests every model and window Northbeam computes, so you can ask the same CPO or ROAS question through Northbeam custom vs. last-click, or through a 1-day click vs. 7-day click window, and see how far the answers move.
Yes. This is the most common pattern. Hazel joins Northbeam-modeled performance to Shopify net revenue and to platform-reported numbers from Meta Ads, Google Ads, and TikTok Ads, so you can compare all three in a single answer.
Hazel connects with read-only access using credentials you authorize. We never write to your account. Your data is encrypted in transit and at rest, and Hazel does not share your data with other customers or use it for model training.
Hazel is a flat annual fee with unlimited seats and unlimited connectors, no per-seat or per-source pricing. Pricing details.
Book a call and we'll walk through your data, the questions you want to answer, and what the rollout looks like.
Hazel helps your team uncover new revenue, catch problems early, and make smarter, faster decisions before you even know what to ask.