Adobe Target licenses get renewed on hope and cancelled on math: the platform pays for itself only when a steady stream of well-designed experiments and personalization activities compounds into measurable revenue. That makes partner selection unusual — you are not really hiring an implementer; you are hiring the operating rhythm of a testing program. The evaluation should look correspondingly different.
Configuration is a fortnight; the program is the engagement
Target's technical setup — implementation via the Platform Web SDK, audience integration, QA workflow — is genuinely modest work for a competent team. What separates partners is everything around it: hypothesis pipelines sourced from analytics and research, prioritization frameworks that balance impact against effort, statistical discipline that prevents false winners, and the enablement that leaves your team running the program independently. When you read proposals, weigh the program design sections double and the implementation sections half.
Five things to screen for
A velocity commitment. Ask candidates what test cadence they will get you to by month three, and what determines it. Mature practices talk about development capacity for test variants, QA turnaround and decision meetings — the real bottlenecks — rather than tool features. A program shipping fewer than four experiments a month rarely justifies its license.
Statistical literacy you can audit. Have the proposed lead explain, in plain language, how they decide when a test is done and how they guard against peeking. Ask how they handle a test that shows +2% at 80% confidence. The answers reveal whether wins will be real. Bonus signal: ask when they would deliberately NOT test something (low traffic, high implementation cost, ethical concerns) — judgment shows in refusals.
Audience strategy before widget strategy. Personalization value concentrates in a handful of high-leverage audiences — returning visitors with cart history, high-value segments from analytics, lifecycle stages. Partners who begin with an audience map and a personalization roadmap will outperform those who begin with homepage banner variations.
Analytics integration depth. Target decisions should read from your analytics categories and feed results back into them (A4T if you run Adobe Analytics). A partner fluent in that integration turns every test into durable audience intelligence; without it, tests are one-off verdicts.
An exit plan for themselves. The best Target partners architect their own redundancy: documented playbooks, trained internal owners, a program calendar your team can run. Ask what month twelve looks like — if the answer implies permanent dependence, keep looking.
Commercial shapes and numbers
Three engagement models dominate. Implementation-plus-enablement (six to ten weeks, $30,000–80,000) suits teams with internal optimization staff. The managed testing program ($8,000–25,000 monthly) delivers hypothesis-to-readout as a service and fits teams without dedicated practitioners. The hybrid — partner-led first two quarters, then staged handover — is usually the best value if the enablement is contractual rather than aspirational.
A one-question filter
End every partner conversation with: "Tell me about a test you were sure would win and didn't — and what you did next." Practitioners light up; this is the job. Vendors redirect to their win-rate slide. Experimentation is a discipline of being wrong efficiently, and the partners who understand that will build you a program that outlives the engagement.
Shortlist from firms with documented Adobe Target delivery, ask each for a program design against your traffic and team reality, and choose the one whose plan you could imagine running without them in a year.
Frequently Asked Questions
What does an Adobe Target partner engagement cost?
Implementation and enablement engagements run roughly $30,000–80,000 over six to ten weeks. Managed testing programs — hypothesis development through readouts — price at $8,000–25,000 monthly depending on test velocity and whether the partner also develops variant code.
How many tests should a Target program run monthly?
Mature programs on adequate traffic typically sustain four to twelve experiments monthly. The constraint is rarely the tool — it is variant development capacity, QA and decision cadence. Ask prospective partners what velocity they will commit to and which bottleneck they will fix first.
Do we need Adobe Analytics to use Adobe Target well?
No, but the combination is materially stronger. Analytics for Target (A4T) lets you analyze experiments in your standard reporting with consistent segments and metrics, and turns test results into reusable audience insight. Partners fluent in A4T extract more durable value per test.
When is a site's traffic too low for A/B testing?
Below roughly 10,000 conversions a year per testing surface, most A/B tests take too long to reach significance. Honest partners will steer low-traffic programs toward bigger, bolder changes, sequential testing approaches, or personalization rules grounded in research rather than dense experimentation calendars.