01 · Affiliate Managers
Stop Exporting CSVs: How Affiliate Managers Run Their Day by Just Asking Affise
Stop exporting CSVs. Start asking questions.
Every affiliate manager knows the Monday morning routine. You open Affise, pull last week's numbers, export them to a spreadsheet, compare them to the week before, and figure out which partners are up, which are down, and who's gone silent. Then you do it all over again to check for trafficback issues and make sure payouts are ready. By the time you finally have the full picture, the whole morning is gone.
The Affise MCP changes that. Instead of clicking through dashboards and exporting data, you ask a question in plain English and get the answer back in seconds. It connects your AI assistant (Claude, Cursor, Continue.dev) directly to Affise, so the assistant can query your live Affise data on your behalf.
This article shows you what that actually looks like, with real examples of the tasks you handle every day as an affiliate manager.
Meet the Affise MCP
The Affise MCP is a bridge between Affise and AI tools that support the Model Context Protocol. You type a question in plain English. The AI figures out which Affise data to pull, fetches it, and returns a clear answer. No API calls, no exports, no pivot tables.
It works with Claude (claude.ai), Cursor, and Continue.dev. The setup takes about two minutes via OAuth. Once connected, your AI assistant can see your live Affise data for the duration of the session.
What You Can Do
Morning Portfolio Scan
Instead of opening Affise, exporting stats, and loading them into a spreadsheet, you ask:
›"Top 20 partners by revenue last week vs the week before."
›"Top 10 partners by clicks last month."
The AI pulls stats for both periods, ranks partners, and shows the delta. You see immediately who grew, who dropped, and who stopped sending traffic. A scan that used to take 30 minutes of export-and-eyeball work takes a few seconds.
The 360° Partner Deep-Dive
The portfolio scan tells you who to look at. The next question is always why, and answering it used to mean re-running the same export five times with a different grouping each time: by offer, then by country, then by device, then by day, then by sub ID. With the MCP, you just ask for the whole picture, one angle after another, in the same conversation:
›"Full performance for partner 3554 last 30 days: clicks, conversions, revenue, and CR."
›"Now break that partner down by offer, then by country, then by device and OS."
›"Show partner 3554's daily trend for the last 30 days."
›"Conversions by goal for partner 3554 last month, and the top sub2 and sub5 values."
Each answer slices the same partner along a different dimension: offer, geo, device, OS, browser, goal, sub ID, and time — so you build a complete profile without leaving the chat. Affise exposes all of these as stat dimensions, and the AI picks the right one for each question.
When you have enough angles, let the assistant aggregate them into a single read:
›"/analyze_stats on partner 3554's last 30 days across all those breakdowns."
The analysis pulls the threads together: which offers carry the partner's volume, which geos convert and which leak, whether performance is trending up or cooling off, and which sub IDs (campaigns, creatives) actually drive quality. For the full quality picture you can pull the partner's other facets into the same conversation too:
›"For partner 3554, also show trafficback reasons, retention rate, and time-to-action last 30 days."
By the time you finish, you have a 360° view of one partner: volume, quality, geo mix, device mix, and trend aggregated from every dimension Affise tracks, assembled in a couple of minutes instead of a morning of exports.
Conversion Checks Before Payouts Go Out
When payout day arrives, you can pull raw conversion records to see exactly what cleared and what did not:
›"Approved conversions from yesterday for partner 193 in the US."
›"Last 100 conversions for partner 3554 with payouts and statuses."
The affise_conversions_raw tool returns per-event data: status, payout, affiliate, geo, and device. This means you can reconcile what you track against what the partner expects before invoices go out, without switching tools or writing a single query.
Tracking Links on Demand
When a new partner asks for a tracking link, or when you are setting up a new campaign, you no longer need to navigate through the Affise interface to build the URL:
›"Generate a tracking link for offer 95119 and affiliate 3554 with sub1: campaign_x and sub2: creative_y."
https://track.demo-network.com/click?pid=3554&offer_id=95119&sub1=campaign_x&sub2=creative_yThe AI builds the link with the correct offer, affiliate, and sub ID parameters and returns it ready to copy. If the partner or offer details are not in your head, you can look them up first:
›"Get partner 3554 detail."
The AI fetches the partner record, and you confirm you have the right person before generating the link.
Trafficback: Who Is Sending Junk and Why
Trafficback is one of the harder things to investigate in Affise because the data lives across multiple dimensions. You can ask:
›"Show trafficback for the last 7 days, broken down by reason."
The AI returns a breakdown by reason with each reason's share of the wasted traffic, plus an analysis summary. From there you can slice the same query further by geo, offer, or partner to see who and where the waste concentrates. You can follow up in the same chat:
›"/analyze_trafficback on the last result."
That analysis highlights which partners to investigate further, which reasons suggest technical issues versus policy violations, and where to focus your cleanup effort.
Sub ID Attribution and Retention
When a partner asks which of their campaigns are actually converting, or you want to understand which creatives are bringing quality traffic, sub ID breakdowns answer the question directly:
›"Show retention rate for last 30 days by offer."
›"Break down conversions last month by sub2 for partner 3554."
You can also look at time-to-action patterns to understand conversion quality across different traffic sources. The AI handles the dimensional breakdowns and surfaces which sub IDs are performing, so you can have an informed conversation with the partner about where to scale and where to cut.
Results You Can See, Not Just Read
In clients that support MCP UI apps, the answers don't arrive as a wall of text; instead, they render as interactive components right inside the chat. A portfolio scan or a partner deep-dive comes back as a sortable, paginated stats table you can eyeball in a second. A generated tracking link renders as a copy-ready card. An offer lookup shows up as scannable offer cards. In clients without UI support, the exact same data comes back as clean text, so you lose nothing either way.
Here is the same "top partners" answer as it actually renders. Click a column to sort, type to filter:
From Ad Hoc to a Repeatable Workflow
The MCP is not a one-shot query tool. It is a conversation. After the initial answer, you can follow up, dig deeper, or pivot to a related question without starting over.
You might ask for last week's top partners, notice one of them dropped 40%, then immediately ask for their recent conversions to see if it is a volume issue or a quality issue. The AI holds the context from the previous answers, so you can ask "how do their recent conversions look?" and it understands which partner you mean.
Over time, you will develop a set of questions you run every Monday morning or every Friday before payout is approved. Those questions become part of your daily routine, and the routine gets faster because the AI already knows what you tend to ask next.
Getting Connected
Setup takes about two minutes. Open the quickstart guide for step-by-step instructions for Claude, Cursor, and Continue.dev.
For Claude, you add a custom connector pointing to https://mcp.affise.com/mcp and authenticate with one click via OAuth in your browser. After that, your AI assistant has access to your Affise data for the session.
For teams, a static API key option is also available if you prefer not to use OAuth.
Start Asking
If you manage an affiliate program on Affise, the data you need is already there. The MCP gives you a faster way to get at it: ask in plain English, get the answer, act on it.
Open the quickstart guide and connect in the next two minutes. Your Monday morning will look different next week.