Short version: dropshipping is a retail business where you own the customer, the refund, and the chargeback but not the inventory. Digital marketing is a service business where you sell your time and judgment to someone else who owns those problems. They get compared because a dropshipping store dies without traffic, so the skills overlap — but the cash flow, the risk, and the daily work are not the same job. If you need revenue inside 60 days and a $1,500 loss would hurt, take the service side; you can be paid before you have spent anything meaningful. The store is only worth building if you can watch that same $1,500 disappear into ad tests that taught you something and still have cash left for the refunds that show up a month after the sale.
The one-line difference
In dropshipping, the money flows to you first and the problems arrive later — the customer pays on Tuesday, the supplier ships on Friday, the package is late on day 19, and the dispute lands in week five. Every one of those is yours. In digital marketing you invoice for work performed; if the client’s product is bad, that is their inventory problem, not yours. The tradeoff is ownership. A store compounds: the product page you rewrote in March is still converting in November while you sleep. A client roster does not. I lost my largest retainer because the marketing director who hired me took a job somewhere else — the work was fine, the numbers were fine, the sponsor was gone — and I was back to pitching with nothing carried over.
The difference that only shows up on paperwork: dropshipping makes you a merchant of record. When a card is disputed the money leaves your account first and you argue afterward, and the letter about your dispute ratio arrives with your business name on it. The card networks, the tax authority, and the consumer protection regulator are all looking at you. A freelance marketer running someone else’s ad account never touches any of that — it is the client’s name on every one of those documents.
Comparing on cost
Store side, checked against the official pricing pages in August 2026:
- Shopify Basic: $19/month billed annually on shopify.com/pricing as of 2026 (monthly billing runs higher). If you use a third-party payment gateway instead of Shopify Payments, Basic adds a 2% fee per order on top of whatever that gateway charges. On a $25 order that is $0.50 gone before your gateway has taken its own cut, and it lives in a settings page you configure once and never look at again.
- Amazon, if you list there too: the Professional plan is $39.99/month and the Individual plan is $0.99 per item sold. Referral fees run 5% to 45% by category, but 15% is the most common rate — it covers Home and Kitchen, Sports and Outdoors, Toys and Games, Tools, Office Products and more — with a $0.30 per-item minimum in most categories (sell.amazon.com/pricing, 2026).
- Traffic: this is the real budget. WordStream’s 2026 Google Ads benchmark study, built from 13,474 US search campaigns running April 2025 through March 2026, put the median cost per click at $5.42 across all industries. That is a cross-industry median, not an ecommerce number — retail keywords typically clear well under it and legal or insurance terms well over — but use it as a sanity check. If your product sells for $29 at a 40% margin, you have roughly $11.60 of gross profit and about two clicks of room before the order is underwater.
Service side, the fixed costs are close to nothing: a laptop, and optionally a credential. Google’s Skillshop certifications are free. Meta’s exams are paid and priced per exam through its Pearson VUE store — the vouchers listed there in August 2026 were $49 for the Associate level and $75 for Professional, with prices varying by country, so check the store for your region rather than trusting a blog. Neither certification wins clients on its own. One documented account you improved does.
Comparing on requirements
Dropshipping requires: a registered business entity your payment processor will underwrite, sales tax registration wherever you have obligations, a supplier who will actually answer messages, a return address that is not your apartment, and an ad account that has not been restricted. The processor is the gate people trip on. Underwriting looks at your category, your expected volume, and your dispute exposure, and a brand-new store selling a high-refund category can get a rolling reserve placed on it — meaning a percentage of your revenue is held for a set period before it reaches your bank. Plan your first two months as if that hold exists, because you will not find out until after your first real sales day.
Digital marketing requires: proof. Not a course certificate — a before/after you can describe in specifics. “Cut cost per purchase from $41 to $26 over six weeks by killing 14 of 19 ad sets and rebuilding the creative” gets a meeting. “Certified in Meta Ads” does not. The practical bootstrap is running your own small store or a friend’s account at cost for 60 days so you have real account access to point at. Ask for read-only access to stay on after the engagement ends, in writing, while they still like you — once you are off the account you cannot screenshot anything and the case study evaporates.
What only shows up once you commit
Three things I did not understand until I was already running orders.
The shipping clock is yours, not your supplier’s. Under the FTC’s Mail, Internet, or Telephone Order Merchandise Rule, a seller must have a reasonable basis to expect shipment within the advertised timeframe, or within 30 days if no timeframe is stated. If you learn you cannot make it, you have to seek the buyer’s consent to the delay or refund promptly on your own initiative — the customer does not have to ask. Suppliers quote processing in their business days and they do not count their own factory holidays. The week around Lunar New Year is the one that got me: a SKU advertised as shipping in 3 business days went out 17 calendar days after payment, and every buyer in that window had a specific delivery date sitting in their confirmation email. Your storefront quotes a calendar date to a buyer. That gap is where the rule bites, and “my supplier was late” is not a defense. Read the FTC business guide once, properly, before you set delivery estimates.
Disputes are weighted by dollars, not by count. This changes which orders you fight for. Visa’s Acquirer Monitoring Program reduced its “excessive” merchant ratio threshold from 2.2% to 1.5%, effective April 1, 2026. Ten disputed $9 orders and one disputed $180 order are not the same event to your acquirer. In practice that means you assemble tracking and delivery evidence for the $180 one and you refund the $9 complaints on the spot, the same day, before they ever become disputes — the $6 you save by arguing is not worth the ratio. Enforcement reaches you through your acquirer, so the thresholds that actually apply to your account are the ones they tell you, not the ones a blog quotes.
Refunds cost more than the refund. When you refund a $6 item, you lose $6 of goods plus the ad spend that bought the click plus the payment processing, and on most platforms the processing fee is not returned. At a $12 acquisition cost, that refund is a $20 event. Track refund rate per SKU, not store-wide — one bad product usually carries the whole number. A single phone mount of mine came back roughly four times as often as anything else in the catalog and dragged the store average to a figure that looked like a store-wide quality problem. It was one listing. Delisting it fixed the metric in a week.
On the marketing side, the equivalent lesson: conversion tracking breaks silently. Change a theme, swap a checkout app, and the pixel stops firing while spend keeps flowing at full speed. Nothing alerts you — the dashboard shows zero conversions, which looks exactly like a bad week. Twice now I have found a pixel that died on a Thursday and only caught it the following Monday. Put the platform’s reported conversions next to the real order count in your admin every Monday, and if they disagree by more than about 5%, find out why before you touch a single budget.
Exactly who to contact and what to ask
Four conversations, in this order, before you spend on ads:
- Your supplier or sourcing agent (the account rep on WhatsApp or the platform chat, not the generic storefront inbox). Ask four things: average dispatch time in calendar days measured from when payment clears; the defect and replacement rate on this specific SKU over the last 90 days; whether they will ship blind with your packing slip and no promotional inserts; and what physically happens to a unit a customer returns. Get the answers in writing in the chat.
- Your payment provider’s risk or underwriting team — Shopify Payments support or Stripe support through the dashboard, and ask to be routed to risk. Ask: what dispute ratio applies to my account, which monitoring program thresholds does my acquirer enforce, and is there a rolling reserve or payout hold on a new account in my category?
- Amazon Seller Support, via a Seller Central case, before you list. Ask which referral fee category a specific ASIN falls into. Categories that look obvious frequently are not, and the difference between 8% and 15% decides whether the listing works.
- An accountant licensed in your state, for sales tax registration and nexus. This is the one place to pay for an hour of professional time rather than reading forums; requirements vary by state and change, and I am describing operations here, not giving tax or legal advice.
Which one to pick
Pick digital marketing services if losing $1,500 would change your month. You can bill your first client before you spend anything meaningful, and every account you touch teaches you the exact skill a store needs later. The ceiling is your hours — and not all of those hours are billable. Expect something like a third of the week to go to work you cannot invoice: the monthly report nobody reads, the call where someone asks why last Tuesday dipped, four days of waiting on approval for an ad that took twenty minutes to write.
Pick dropshipping if you can fund roughly 60 days of testing plus a refund buffer without touching money you need, and if you would rather debug a supplier than manage a client. Budget for the store to be unprofitable while you look for the product. Most tested products do not work — of the eleven I ran properly, two paid for the other nine — and the two were not the ones I expected. The skill that decides it is killing a test at $80 of spend instead of nursing it to $400 because the last three days looked slightly better.
If you are doing both, sequence them: run services first, use the income to fund the store, and treat the store as the place you test ideas you would not risk on a client’s budget. That order costs you nothing you were not already earning.