How to Run a Low-Risk Trader Reactivation Pilot

Nobody at a brokerage gets fired for a $2,000 pilot that proved nothing, which is exactly why so many of them prove nothing. A trader reactivation campaign pilot earns its budget when it answers one narrow question about one slice of the dormant book, at a cost you already agreed you can lose. This guide covers how to scope one: the question, the list cap, the markets, the budget at $0.35 per minute, the success criteria and the go/no-go rule, all written down before the first dial.

It's written for the retention or ops lead who's been told to "test AI calling on the dormant accounts" and now needs a scope that compliance and finance will sign in one meeting. Topcalls runs the calls; the worksheet linked below is the scope.

Key Takeaways

• A pilot tests one question on two or three segments in one or two markets; a worked example of 1,000 accounts, 3 attempts and 1.5 minutes per attempt costs $1,575 in calling at $0.35 per minute.
• Hold out 10-20% of each segment and don't call them, so the pilot shows how many traders would have redeposited without a call.
• Under UK PECR regulation 21, a number on the TPS register for 28 days or more must not receive an unsolicited marketing call, so suppress before the list is uploaded anywhere.
• US telephone solicitations under 47 CFR 64.1200 stay between 8 a.m. and 9 p.m. in the called party's local time, and a do-not-call request must be honored for 5 years.
• Topcalls setup takes about 15 minutes once the list and script are ready, and most pilots fit in four to six weeks including the attribution window.
• Write a three-outcome go/no-go rule (go, adjust and rerun, no-go) before launch, so the review meeting reads results instead of debating them.

1. What does a trader reactivation campaign pilot need to prove?

A trader reactivation campaign pilot proves one thing: whether AI calls can bring a defined slice of dormant traders back to a deposit or a trade at a cost the brokerage accepts. Write the question, the reactivation event and the attribution window before the first dial. Logins, app opens and "interested" dispositions don't count as reactivation; they make a weak pilot look good and the scale decision wrong.

The useful shape is narrow. "Can AI calls bring funded traders who went quiet 90 to 180 days ago back to a deposit above $100 within 21 days of the last attempt?" has a yes or no answer. "Does AI calling work for us?" doesn't, and a pilot built on that question ends in a meeting where everyone reads the same numbers differently.

Name what's out of scope too. Offers, new markets, human handoff and inbound each deserve their own test; a pilot that tries all four usually proves none. And write two names down: the owner who runs it day to day and the decision maker who signs the go/no-go. If trader segmentation for reactivation is still an open question, settle it first, because the pilot inherits whatever segments you define.

2. How big should the pilot list be?

Cap the list at a number you can afford to be wrong about, then pick two or three segments from the dormant book after suppression. A practical floor is enough accounts that 20 reactivations would read as a signal rather than noise. Hold out 10-20% of each segment and don't call them, so you can see how many traders would have redeposited on their own.

Suppression comes before sizing, not after. Remove DNC entries, withdrawn consent, open complaints, blocked or under-review accounts, and anyone your compliance desk excludes. In the UK, regulation 21 of PECR bars unsolicited direct marketing calls to any number on the TPS register (with a grace period only for numbers listed less than 28 days before the call) and to anyone who has told you to stop. For US numbers, 47 CFR 64.1200 says a do-not-call request "must be honored for 5 years from the time the request is made", and you have at most ten business days to act on one. The full exclusion list is in the reactivation campaign suppression rules post.

Recently quiet and funded traders usually go first: accounts that passed KYC, deposited at least once, and stopped trading 60 to 180 days ago. They know the platform, and the consent basis is cleanest. Registered-never-funded accounts are a different campaign with a different consent question, so keep them out of a first pilot even though there are more of them.

One more field before the export: how the segment tag reaches the calling platform. A column in the CSV, or a CRM field that Smart Campaigns reads per contact. Without it you get one blended result and no idea which segment to scale.

The pilot scoping worksheet is a seven-section fill-in: the one question, the list cap and segments, markets and languages, success criteria, a budget table at $0.35 per minute, a timeline, and a three-outcome go/no-go rule with sign-off lines.

3. Which markets and languages should a pilot include?

One or two markets, not the whole footprint. Each market adds a language, a calling window in the trader's local time, a caller ID and a compliance review, and each of those can hide a problem the pilot then can't separate from the AI itself. A single market with a clean read beats three markets with a muddy one.

Use the language traders actually speak to your support desk in, not the one on the marketing site. Topcalls covers 32 languages end to end, so a Spanish-speaking book in Madrid and a Portuguese one in Lisbon can run in the same pilot with separate scripts; check the specific languages you need before committing. The customer reactivation page shows how a campaign switches language per contact.

Calling windows are set per market. In the US, 47 CFR 64.1200(c)(1) bars telephone solicitations before 8 a.m. or after 9 p.m. at the called party's location. In the UK, FCA COBS 4.8.2R permits an unsolicited real-time financial promotion only where "the recipient has an established existing client relationship with the firm and the relationship is such that the recipient envisages receiving" such calls, and the caller must identify the firm and the purpose at the outset. Have your compliance officer confirm which rule applies to each segment. Don't take a vendor's summary for it.

Caller ID matters more than most pilots expect. A local number in each market connects better and looks less like spam, and a connect-rate shortfall in week one is more often a caller-ID or list problem than a script problem.

4. What does a trader reactivation pilot cost at $0.35 per minute?

Topcalls bills $0.35 per minute all-inclusive, covering the voice model, telephony, recording, transcription and analytics with no per-seat, setup or bundle fees, so the calling budget is minutes times 0.35. A pilot of 1,000 accounts, 3 attempts each and 1.5 minutes per attempt uses 4,500 minutes and costs $1,575 in calling. Human follow-up and data work come on top.

The $500 and $120 lines are placeholders; your loaded rates will differ. What shouldn't change is the structure: a ceiling above which the pilot stops regardless of how promising it looks, approved in writing by one name, plus a break-even count. Divide the total budget by finance's net revenue per reactivated trader. At $2,195 and, say, $300 net per returning trader, the pilot pays for itself at 8 traders, which is 0.8% of 1,000 dialed. Compare that with the rate target in the next section. The dormant trader revenue calculator runs the same maths on your own book size, and the cost per reactivated trader post goes deeper on the loaded number.

Offers are a budget line only if you're testing one, and only where the trader's jurisdiction allows it. ESMA's 2018 CFD measures, which reported that 74-89% of retail accounts lose money, also included "a restriction on the incentives offered to trade CFDs" for retail clients in the EU. Steven Maijoor, ESMA's Chair at the time, said the measures would "guarantee greater investor protection across the EU by ensuring a common minimum level of protection for retail investors". A first pilot is simpler without an offer anyway; the dormant trader reactivation offer post covers when one earns its place.

5. How do you set success criteria before the first dial?

Five numbers, written before launch: a reactivation rate target, a cost per reactivated trader ceiling, connect and conversation floors, a complaint and opt-out ceiling, and a minimum sample of connected calls before anyone reads the result. Set the rate target from your own past channels, not from a vendor benchmark. Targets set after the results arrive aren't targets. They're a story.

The rate target is reactivated traders divided by accounts dialed. Your best past channel, whether that was email, SMS or a human desk, is the honest reference point. If human callers got 2% on a similar segment, a pilot target of 2% at a lower cost per reactivation is a fair bar. The win-back campaigns for inactive trading users post compares the channels side by side.

Connect and conversation floors are early-warning numbers. You can read them in the first two days from real-time analytics, long before deposits show up. Connects short of the floor point at the list or the caller ID. Conversations short of the floor point at the opening line. Topcalls' 60%+ connect-rate lift is a first-party figure from its own campaigns, a reason to expect improvement over a manual desk rather than a target for your book.

Compliance sets the complaint ceiling, and it's the one number that pauses the pilot without a meeting. Write the sample floor too. Waiting for a few hundred connected calls before drawing a conclusion stops a 40-call Tuesday from getting the pilot cancelled for the wrong reason.

6. How long should a reactivation pilot run, and who decides?

Four to six weeks, including the attribution window, so the decision maker at the end is the same person who signed the scope. Prep is the longest stretch because compliance review and the CRM export sit with other teams; Topcalls setup itself takes about 15 minutes once the list and script are ready. Book the go/no-go meeting before launch, not after.

Run a soft launch first: a few hundred accounts over one or two days, so you can listen to recordings and fix the opening line before the full list burns through. Then dial the rest inside two weeks, retries included, so week one and week two are comparable. Nothing is read as final until the attribution window closes. If the answer is Go, a full AI voice agent programme for dormant trading accounts is typically live within about two weeks, so the pilot-to-scale gap stays short when the scope was clean.

Two rules make the table work. Decide now which segment scales first on a Go, usually the one with the lowest cost per reactivated trader unless compliance flags it. And on an Adjust, change one thing per rerun: opening line, retry count, calling window or segment mix, in an order you list before launch. Anything else turns the second batch into a hunch instead of a test.

7. When doesn't a reactivation pilot fit?

A pilot isn't the right move when the dormant pool after suppression is under a few hundred accounts, when consent for the segment you want to call is unresolved, or when nobody can define a reactivated trader as one event. In each case the pilot returns a number nobody can act on. Fix the input first; the pilot will still be there afterwards.

• Pool too small: With 300 accounts after suppression, even a strong 3% result is nine traders. Call them from a human desk and skip the pilot.
• Consent unresolved: If compliance can't name the basis for calling a segment in a given market, that segment stays out. A pilot doesn't settle a legal question.
• No deposit signal in the CRM: If deposits from MT4 or MT5 don't reach the CRM inside the attribution window, nothing can be attributed. Wire the deposit event first.
• The real question is the offer: If the debate inside the brokerage is about what to offer returning traders rather than whether to call them, run an offer test on a channel you already have.

Scope it on paper first. Then book a 30-minute call with Topcalls: we'll size your dormant book, estimate pilot minutes at $0.35 per minute for the segments and markets you pick, and map how deposits flow back from MT4/MT5 or your CRM so the go/no-go is measurable. A proposal follows within 48 hours.