direct_mail_marketing v1.0.0

name: direct_mail_marketing description: "Plan and run a direct mail campaign that earns its postage: choose postcard, letter, self-mailer, or EDDM, build the offer and copy, clean and target the list, work the print and postage cost math to a break-even rate, set a random holdout, measure incremental response, and run the follow-up sequence. Platform-free, no vendor account required. Ships a cost and break-even calculator, a campaign worksheet with a holdout plan, and a worked campaign from list to response report. Use when someone asks to plan a mail drop, postcards or letters to customers or prospects, Every Door Direct Mail, direct mail costs or response rates, or whether a mailing paid off."

Direct Mail Marketing

Purpose

Direct mail is bought reach with a long fuse. You pay for print, data, and postage before you learn anything, the piece lands days later, and responses arrive over weeks. That cost structure punishes guessing. This skill runs a campaign as a measured bet: a defined audience, one clear offer, cost math solved to a break-even rate before print, a random holdout so the result can be believed, and a follow-up sequence planned before the first piece drops.

It is platform-free on purpose. The same system works with a local printer and a post office counter, a mail house, or any sending service. Vendor tools can execute a drop, they do not replace the planning, the hygiene, or the measurement here.

Two numbers govern everything:

• Break-even rate: total campaign cost divided by margin per response, divided by pieces mailed. Below it, the drop loses money no matter how good the creative feels.
• Incremental lift: conversion in the mailed group minus conversion in the holdout group. Gross response counts people who would have bought anyway. Only lift is what the mail caused.

Workflow

1. Start with the goal, the offer, and the unit economics

• Name one goal for the drop: orders, donations, bookings, appointments, or qualified leads. A piece that asks for two things usually gets neither.
• Write the offer before any copy: what the recipient gets, what it costs them, and the deadline or limit that makes acting now rational. Weak offers are the most common reason a well printed piece fails, ahead of format and ahead of design.
• Fix value per response and margin percent, and be able to defend both. Margin per response is value times margin percent. A campaign judged on revenue instead of margin will call losing drops winners.
• Run the calculator in scripts/ early, with a rough postage figure, to learn whether the economics can work at all at a plausible response rate. If break-even needs a rate no list like yours has ever produced, fix the offer, the audience, or the format now, not after print.

2. Choose the format for the job

• Postcard: cheapest to produce, no envelope to open, the offer is visible in the mailbox. Best for one simple offer to a broad or local audience, reminders, and repeat drops. Limited room, so one message only.
• Letter in an envelope: the format for a longer argument, fundraising appeals, considered purchases, and anything that needs trust or detail. Costs more per piece in print, insertion, and often postage class, and it has to earn the open. Plan the envelope, letter, and reply device as one unit.
• Self-mailer: a folded piece that mails without an envelope, with room for more copy than a postcard. It must be folded and tabbed or sealed to postal rules for its size and class. Confirm the finished, folded size is what qualifies, not the flat sheet size.
• Every Door Direct Mail (EDDM): saturation mail to every address on chosen carrier routes, addressed to the postal customer, no list to buy or clean. Use it when geography is the targeting, typically a local business drawing from nearby routes. Pieces are flats: a piece qualifies when it is longer than 10.5 inches, or taller than 6.125 inches, or thicker than 0.25 inch, and no larger than 15 by 12 inches and 0.75 inch thick, up to 3.3 ounces. Common sizes are 6.25 by 9, 6.5 by 9, and 8.5 by 11 inches. A standard 4 by 6 postcard does not qualify.
• Match format to consideration level. Low consideration and local: postcard or EDDM. High consideration or high value: letter. In between, or a story to tell in panels: self-mailer.

3. Pick the mail class and check the minimums

• USPS Marketing Mail is the default for promotional drops. It is bulk only: 200 pieces or 50 pounds per mailing, a business mail permit is required, and there is no single-piece Marketing Mail rate. Delivery is slower and not guaranteed to a date.
• Presorted First-Class costs more and moves faster, and it is the class for time-sensitive pieces and mail that must be forwarded or returned. Presorted First-Class needs 500 pieces.
• EDDM Retail needs no permit and pays a flat per piece rate, with at least 200 and up to 5,000 pieces per day per ZIP Code. EDDM through a Business Mail Entry Unit is priced slightly lower per piece and suits larger drops.
• Rates change on a published schedule. Check the current USPS Notice 123 price list on the day you budget, record the date you checked, and pass that exact rate to the calculator. Rates quoted in any guide, including the references here, are a starting point for the conversation, not a budget.
• Postal anchors from Notice 123 effective July 12, 2026, to sanity check a quote against: Marketing Mail letter presort at the Mixed tier 0.467 dollars per piece, presorted First-Class letter at the Mixed tier 0.707, presorted First-Class postcard at the Mixed tier 0.495, EDDM Retail 0.26 per piece on the USPS EDDM page, single-piece First-Class letter 0.82, single-piece postcard 0.65. Denser, better sorted lists earn lower tiers than Mixed.

4. Build and clean the list

List quality moves response more than any copy change. Do hygiene in this order:

• Standardize and validate with CASS certified processing: correct formatting, add ZIP+4, verify the address is a real delivery point, flag the ones that are not.
• Update movers with NCOA processing, which compares the list against USPS change of address records covering the last 48 months. Marketing Mail discounts require Move Update processing within 95 days of the mailing. A list processed last year is not compliant and not clean.
• Dedupe on name plus address, and decide the household rule in advance: one piece per household, or one per person, on purpose.
• Suppress what should never mail: your own customers on a prospect acquisition drop, people who asked not to receive mail, deceased records, vacant addresses, and any segment the offer does not apply to.
• Record the counts at every step: in, standardized, moved and updated, duplicates removed, suppressed, final mailable. A large drop at any step is a finding about the list source, report it.
• Target after hygiene, not before. Segment the clean list by recency, past purchase or giving, and fit for this specific offer. House lists, people with an existing relationship, are a different population from prospect lists and must be forecast, mailed, and judged separately.
• EDDM skips this step by design, because there is no list. Its targeting is the route selection itself, using the USPS EDDM Online Tool to map routes and review route counts and Census based filters before committing.

5. Write the piece

• Lead with the offer and the recipient benefit, in plain words, in the first lines a scanning reader sees. On a postcard that is the headline side. In a letter it is the opening and the postscript, the two parts most often read.
• One call to action, stated the same way everywhere on the piece: one URL, one phone number, one code, one deadline. Every extra option dilutes response and breaks tracking.
• Give every cell its own tracking: a unique promo code, a unique URL or QR code that encodes the cell, and a unique phone number if calls matter. If you cannot tell which cell a response came from, you did not run a test, you ran two mailings.
• Personalize only with data you have verified. A wrong name or a wrong past purchase printed on a piece is worse than no personalization at all.
• Proof the physical facts like a printer will: finished size, weight, fold and tab placement, address block position, indicia or stamp area, and barcode clear space. A piece that fails a postal check at induction costs more than a proofing round.
• Keep claims honest and checkable. Discount terms, deadlines, and limits printed on the piece must match what the recipient actually gets.

6. Solve the money before print

Run the calculator with quoted, itemized figures, print, postage, data, and lettershop as separate lines, never a single blended per piece price you cannot audit:

python3 scripts/campaign_calc.py --quantity 10000 --print 0.15 --postage 0.467 \
  --data 0.02 --lettershop 0.05 --fixed 350 \
  --value-per-response 80 --margin-pct 60 --response-pct 2

• Read three outputs before approving anything: total cost, all-in cost per piece, and break-even rate.
• Forecast honestly. Your own last comparable drop is the best source. Published benchmarks are a distant second and are directional: the ANA and DMA Response Rate Report figures most often cited put house lists at about 5 to 9 percent response and prospect lists at about 2 to 5 percent, with wide variation by format and offer. Never forecast a prospect drop at a house list rate.
• If the forecast sits below break-even, change a real lever before mailing: a stronger offer, a cleaner or warmer list, a cheaper format, a larger quantity to spread fixed costs, or a higher value follow-on sale counted honestly in value per response. Hoping for a better rate is not a lever.
• Small lists change the math. Under the 200 piece Marketing Mail minimum there is no bulk rate to earn, and fixed costs dominate small drops. Price that reality instead of spreading a big campaign forecast over a tiny test.

7. Set the holdout and any A/B test before the list is finalized

• Hold out a random 5 to 10 percent of eligible records. Random means a mechanical draw, every Nth record after a random start or a seeded random sample. Never hand-pick, never use the worst records, never use one region. A biased holdout produces a confident wrong answer.
• Holdout records receive no piece from this drop. Other normal marketing continues for everyone, so the comparison isolates this drop.
• Keep the holdout large enough to read. On a small list, a 5 percent holdout may contain a handful of conversions and prove nothing. Say so in the plan and either raise the holdout share or accept that this drop will be judged on gross response with that limit stated.
• For an A/B test, change one variable only: offer, headline, format, or list segment, not two at once. Split randomly, mail both cells on the same day, and give each cell its own code and URL from step 5.
• Write the measurement plan before the drop, using references/campaign-worksheet.md: primary metric, match rule, window, report date, and the decision rule. The decision rule is stated in advance: scale only on positive incremental profit.

8. Drop, then run the follow-up sequence

• Confirm induction: date the mail entered the postal system, quantity accepted, and any pieces rejected and why. The measurement window starts at first delivery, not at the print date and not at the drop date.
• Expect most responses inside 2 to 3 weeks of delivery for consumer offers. Business and high consideration offers can keep arriving for 6 to 8 weeks. Set the window to the sales cycle and do not judge a drop at day 4.
• Follow-up sequence, planned before the drop:
• Responders: deliver fast, confirm by the channel they chose, and put them into the normal welcome or fulfilment flow. Speed here is part of the campaign, not an afterthought.
• Engaged non-buyers, for example landing page visitors who did not convert where that is lawfully tracked: one prompt follow-up inside the response window while the piece is still in the home.
• Non-responders: one follow-up piece or a paired email or ad exposure inside about 3 weeks, carrying the same offer and deadline logic, coded as its own cell so its lift can be read separately. Do not silently fold follow-up responses into the first drop results.
• Holdout stays untouched by all follow-up tied to this drop until the window closes, or the measurement is destroyed.
• Watch deliverability signals during the window: returned pieces, undeliverable rate, and code or URL errors. A broken landing page found on day 10 is a campaign loss, not a footnote.

9. Report incremental results and decide

• Close the window on the planned date. Match responses two ways where possible: code or URL attribution, and a match-back of orders or donations against the mailed file. Report both, they rarely agree exactly and the gap is informative.
• Run the calculator in holdout mode with the observed rates:

python3 scripts/campaign_calc.py --quantity 10000 --print 0.15 --postage 0.467 \
  --data 0.02 --lettershop 0.05 --fixed 350 \
  --value-per-response 80 --margin-pct 60 --response-pct 2 \
  --mailed-pct 2.0 --holdout-pct 0.5 --holdout-size 526

• Report, in this order: pieces mailed, gross rate per cell, holdout rate, incremental lift, incremental cost per response, incremental profit, one sentence of learning, and the decision.
• Decide by the rule written in step 7: scale, retest with exactly one change, or stop. A drop at incremental break-even gets a retest with a stronger offer or a warmer list, not a bigger rollout.
• Log the campaign in a running record: list source, hygiene counts, cells, codes, costs, window, gross and incremental results. Forecasts for the next drop start from this log.

Output Contract

Deliver, for a planned campaign:

• A one page campaign plan: goal, offer with deadline, audience and list source, format and mail class with the reason, quantity, and drop date.
• The completed worksheet from references/campaign-worksheet.md, including hygiene counts, cell and holdout design with codes, and the measurement plan written before the drop.
• Calculator output pasted in full: total cost, all-in per piece, break-even responses and rate, and the forecast with its source named.
• Copy or a copy brief for the piece, with the single call to action and per cell tracking stated.

Deliver, for a completed campaign:

• A response report in the step 9 order, gross and incremental shown separately, with incremental profit as the headline number.
• A clear decision: scale, retest with one named change, or stop, with the next drop date if retesting.
• An updated campaign log entry the next forecast can start from.

Worked example of both, with real calculator output: references/worked-campaign.md.

Operating Rules

• Never quote a postage rate without a check date. Rates change, guides go stale, and a budget built on a stale rate fails at induction.
• Never present gross response as campaign profit. Without a holdout, say plainly that incrementality is unknown and the result is an upper bound.
• Never hand-pick a holdout, shrink it after seeing early results, or mail it a follow-up before the window closes.
• Never forecast a prospect list at house list rates, and never borrow a benchmark from a different format or offer without saying so.
• Never blend print, postage, data, and lettershop into one unauditable per piece price in a plan you approve.
• Never mail a list that has not had Move Update processing inside 95 days when claiming Marketing Mail discounts, and never skip dedupe and suppression to save a processing fee. Returned and wasted pieces cost more.
• Never print personalization from unverified data.
• Respect mail preference and opt-out requests, and follow the postal rules of the country you mail in. This skill uses USPS rules and US dollar math. Other countries have different classes, minimums, and hygiene systems, re-verify before applying any number elsewhere.
• No vendor is required. If a mail house or sending service executes the drop, this skill still owns the plan, the worksheet, the codes, and the report, and the vendor supplies itemized costs and induction confirmation into them.