WordStream's 2025 Retail and Shopping benchmark reports a $3.49 cost per click and a 3.83% conversion rate. The United States also ended duty-free de minimis treatment for covered imported goods valued at $800 or less effective August 29, 2025 (WordStream, 2025 benchmark; The White House, July 2025). Those ad numbers cover broad U.S. search campaigns, not future dropshipping sales, and I found no reliable universal margin, income, or success-rate figure in the sources I checked.
I am treating this as a data page and general business education. It is not a review, my personal verdict, or personal financial or legal advice.
I will only answer the question after every changing input has a current source and I can repeat every calculation.
Use the full protocol to put this model inside a wider system for choosing what to build.
Is dropshipping dead in 2026, or is that the wrong question?
I think the better question is whether your specific offer survives a model with every assumption written down. A screenshot can feed your excitement or your pessimism, but neither feeling proves the offer works.
On this page, dropshipping means you take a customer's order without holding the item, then a separate supplier sends it to the customer. I am only looking at per-order economics for a U.S.-facing offer, broad paid-ad benchmarks, and the cited 2025 import-policy change when it applies. I am not estimating organic demand, marketplace traffic, inventory-backed retail, or the economics of any named store.
I could not find a reliable public demand measure that answers whether a beginner should start dropshipping in 2026. The market-size estimates disagree sharply, and none of them tells you whether one offer has buyers or can acquire them at a workable cost. So I will not turn category size or search interest into a verdict they cannot support.
Which inputs decide whether the unit economics work?
Give every input its own line, then put the definition, source date, and limitation beside it. If a value only sounds typical, it does not go in my model.
I am going to label every assumption before I run the store math:
Unless I attach a citation, every number in this worked example is an assumption. I use one measured month with 100 completed orders and $60 in revenue received per completed order. Against that revenue, I put an $18 product charge, a $7 fulfillment and delivery charge, and $2,400 in promotion spend. The transaction charge is assumed at 3% of revenue, the loss allowance at 5%, and the chosen per-order contribution target at $12. I also use $300 in fixed overhead and 50 hours of owner time. Taxes and any product-specific duty stay outside the result because the sources do not supply them. None of these assumed values describes a typical store.
EXHIBIT 01
| Input | Sourced input or labeled assumption | Role in the model |
|---|---|---|
| Revenue received per completed order | Worked-example assumption: $60 per completed order before the separate loss allowance, with collected taxes and incomplete orders left out. | This is where I start the per-order calculation. |
| Product charge | Worked-example assumption: $18 per completed order, before any unknown product-specific duty. | I subtract it once for every completed order. |
| Fulfillment and delivery charge | Worked-example assumption: $7 per completed order under the same order scope. | I subtract it under that same order scope. |
| Promotion spend for the measured period | Worked-example assumption: $2,400 assigned to the same measured month and attribution method as the completed orders. | This is the top half of observed acquisition cost. |
| Completed orders under the same attribution method | Worked-example assumption: 100 completed orders, with canceled, incomplete, and out-of-period orders left out. | This is the bottom half of observed acquisition cost. |
| Customer acquisition charge | Worked-example assumption: $2,400 / 100 = $24 per completed order. For a benchmark proxy, not observed customer acquisition cost, WordStream reports Retail and Shopping CPC of $3.49 and conversion rate of 3.83%. Its wider sample covers 16,446 U.S. search campaigns running April 1, 2024, to March 31, 2025. The exact implied benchmark-conversion cost is $3.49 / 0.0383, displayed as $91.12 only after rounding the final result to cents (WordStream). | The example ties spend and completed orders to the same period. The cited proxy ties promotion spend to reported benchmark conversions, which may not be completed orders. |
| Transaction charge | Worked-example assumption: 3% of the assumed $60 revenue, so $60 x 0.03 = $1.80 per completed order. | I only subtract this when the observed or stated assumed terms apply. |
| Returns, refunds, and dispute allowance | Worked-example assumption: 5% of the assumed $60 revenue, so $60 x 0.05 = $3 per completed order. This is arithmetic, not a sourced loss rate. | This turns observed losses into a per-order allowance. |
| Per-order contribution after variable charges | Worked-example result: $60 - $18 - $7 - $24 - $1.80 - $3 = $6.20 per completed order. | This shows what is left before fixed overhead and owner time. |
| Chosen per-order contribution target | Worked-example assumption: $12 per completed order. This is a test target, not a sourced benchmark or recommendation. | I use this target to calculate the maximum acquisition allowance. |
| Fixed overhead for the measured period | Worked-example assumption: $300 for the same measured month, with the included costs defined before use. | This carries the model from per-order contribution to an operating result. |
| Owner time for the measured period | Worked-example assumption: 50 recorded hours in the same measured month, covering whichever tasks the operator decides to include. | This puts the labor burden beside the operating result where you can see it. |
How should the unit economics be calculated?
Start with the revenue received for a completed order and subtract every variable charge tied to that same order. Before you fill in the result, decide how you will handle returns, refunds, disputes, taxes, and incomplete orders.
Per-order contribution after variable charges = revenue received minus product charge minus fulfillment and delivery charge minus customer acquisition charge minus transaction charge minus the loss allowance.
Observed acquisition charge per completed order = verified promotion spend divided by verified completed orders under the same attribution method.
Maximum acquisition allowance = revenue received minus every other sourced variable charge minus the chosen contribution target.
Total contribution after variable charges = per-order contribution after variable charges multiplied by verified completed orders for the same period.
Operating result before owner pay = total contribution after variable charges minus fixed overhead for the same period.
Owner-time return = operating result before owner pay divided by verified owner time for the same period.
Here is the full chain of math: $2,400 / 100 = $24 acquisition per completed order; $60 - $18 - $7 - $24 - $1.80 - $3 = $6.20 contribution per order; $6.20 x 100 = $620 total contribution; $620 - $300 = $320 before owner pay; and $320 / 50 = $6.40 per assumed owner hour. The maximum acquisition allowance is $60 - $18 - $7 - $1.80 - $3 - $12 = $18.20, and I check it with $18.20 + $18 + $7 + $1.80 + $3 + $12 = $60. The assumed $24 acquisition charge is over that allowance by $24 - $18.20 = $5.80. Taxes and product-specific duty are excluded, so this is not a net-income result.
If the worksheet gets a zero on the bottom of the fraction, it should label that problem instead of spitting out an acquisition or owner-time result.
Pick one order status, attribution method, time period, currency, and treatment of taxes before you type in a value. Keep refunds, disputes, promotion spend, goods costs, delivery charges, and overhead inside that same scope. A click, lead, reported conversion, placed order, and completed order are different events, so do not mix them. If you do not know how to treat an item, leave the result unknown instead of hiding the cost on another line.
Reproducible acquisition-cost stress test
Now I will stress-test the published ad inputs without pretending they describe a store:
This original layer uses only the published ad inputs I could verify. The sources do not give us the missing order revenue, product, fulfillment, return, transaction, overhead, or labor inputs, so this test cannot produce a store profit result.
I use 1,000 clicks in the first two rows only to show the arithmetic. That volume is an assumption, while the ad rates and campaign-sample details come from the cited sources.
EXHIBIT 02
| Test | Verified inputs | Calculation | Check the math |
|---|---|---|---|
| Cost of 1,000 all-industry Google clicks | Publisher's 2025 CPC: $5.26. Search Engine Land reports the prior-year comparator at $4.66. | 1,000 x $5.26 = $5,260; 1,000 x $4.66 = $4,660; change: $5,260 - $4,660 = $600. Exact percentage formula: $600 / $4,660 x 100. | $4,660 + $600 = $5,260. Search Engine Land reports a conflicting 2025 figure of $5.42 from the same dataset, so I use WordStream's own $5.26 as the current figure and keep the disagreement visible (WordStream; Search Engine Land). |
| Retail and Shopping benchmark-conversion proxy | CPC: $3.49. Conversion rate: 3.83%. Scenario: 1,000 clicks. | Click cost: 1,000 x $3.49 = $3,490. Modeled conversions at the reported rate: 1,000 x 0.0383 = 38.3. The exact implied cost is $3,490 / 38.3, displayed as $91.12 only after rounding the final result. | Algebraic check: ($3.49 / 0.0383) x (1,000 x 0.0383) = $3,490. A reported conversion may not be a completed dropshipping order (WordStream). |
| Facebook lead-cost change | 2025 cost per lead: $27.66. Prior figure: $22.87. | $27.66 - $22.87 = $4.79. Exact percentage formula: $4.79 / $22.87 x 100. | $22.87 + $4.79 = $27.66. This benchmark comes from 726 U.S. campaigns with a leads objective, not a dropshipping purchase cost (WordStream). |
The formulas keep the exact source inputs. A money display rounded to cents is for presentation, and I would never feed that rounded display back into the next calculation.
What should a beginner verify before spending money?
I would pause before spending until the offer, supplier terms, delivery promise, acquisition measurement, return rules, and downside limit all fit on the same page. Every outside input needs its source date and scope beside it.
- Save the dated supplier quote and terms for the exact item and destination. Write down what the quote includes, what can change, who handles a failed shipment, and which costs you still do not know. A marketplace listing or verbal estimate is not a landed-cost record for your store.
- Place test orders to the locations the offer will serve, then record the order, dispatch, and arrival timestamps. Keep the delivery promise shown to the buyer, tracking availability, damage, and failed delivery in that same record. Use what you observed instead of borrowing somebody else's general delivery-time claim.
- WordStream's 2025 Google report covers 16,446 U.S. search campaigns that ran from April 1, 2024, to March 31, 2025. It reports an all-industry CPC of $5.26 and Retail and Shopping CPC of $3.49 with a 3.83% conversion rate. Those are broad search benchmarks, not a forecast for your store (WordStream). Its Facebook report separately covers 554 traffic-objective campaigns and 726 leads-objective campaigns, and those objectives are not purchase campaigns (WordStream).
- Track returns, refunds, and disputes separately, and keep both the count and dollar loss for each. When you build an allowance, divide only by completed orders from the same period. Until the store has matching records, call any loss input an assumption.
- I could not verify a current platform-policy or jurisdiction-specific consumer-obligation source for this test. Check the current rules for the sales channel, advertising channel, buyer location, and seller location before you spend or accept orders. Keep any unresolved compliance cost outside the result and mark the model incomplete.
One policy input is current and verified: effective August 29, 2025, imported goods sent through means other than the international postal network and valued at $800 or less lost duty-free de minimis treatment. The sources give us no product-specific duty rate, so I cannot calculate the landed-cost effect from this evidence alone (The White House, July 2025).
When should a beginner choose another online path?
Write your decision rule before you touch the spreadsheet. It should include variable contribution, fixed overhead, owner time, and the downside boundary. If the sourced model misses your rule, change the offer, change how you plan to acquire customers, or walk away from the model. Do not massage the arithmetic.
If this option fails the test, compare it with legit online side hustles that need no car or degree under the same evidence standard.
Use the same measured period for every alternative. Start with the money: upfront cash, recurring cost, promotion cost, refund exposure, and fixed overhead. Then write down the path to a first customer, the delivery work, and the owner time, with the evidence quality beside all of it. Do not compare gross revenue from one path with contribution after costs from another. If an input is unknown, leave it unknown on both sides.
What original analysis does this page add?
I put every sourced input on the same completed-order basis, show the formulas, and make it clear which assumption moves the result. A public benchmark stays a benchmark. I will not turn it into a promise. The full method table is in the record below if you want to work through it.
ORIGINAL ANALYSIS TABLE
EXHIBIT 03
| Original-analysis asset | Method | Evidence rule |
|---|---|---|
| Comparable per-order model | Put every cited input on the same completed-order basis. | Define revenue, each variable charge, order status, attribution method, currency, period, and exclusions once. I reject any input that uses a different scope. |
| Sensitivity table | Change one sourced input, then run the result again. | The sources give us no defensible low, middle, and high cases for a dropshipping store. Use observed store data when you have it. Otherwise, label every case as an assumption and never call that range typical. |
| Break-even chart | Plot the maximum acquisition allowance against sourced order inputs. | I only plot inputs with a visible source date or an assumption label, keep the units consistent, and hold the chart back when a required cost is unknown. |
| Operating-result and owner-time table | Carry contribution through fixed overhead, then put the result beside verified owner time. | Use one period for overhead and time, spell out what each includes, and keep owner pay separate. Only compare them with completed orders and contribution from that same period. |
| Assumption audit | Mark every input as observed, estimated, excluded, or still unknown. | The observed outside inputs are published ad benchmarks and the August 2025 de minimis policy change. The actual order revenue, product and delivery charges, completed orders, transaction costs, losses, overhead, owner time, and product-specific duty are unknown. I exclude vendor claims about margins, income, and success rates because the source notes rate them weak and contradictory (WordStream; The White House). |
How does this page stay current?
Whenever I refresh the page, I need to record what changed, which source changed it, and whether the conclusion moved. If I silently swap a number, you cannot audit the analysis. The full log is in the record below.
PAGE MAINTENANCE RECORD
EXHIBIT 04
| Update-log field | Entry |
|---|---|
| Review date | July 10, 2026 gap-close review. |
| Sources added, replaced, or removed | My initial baseline uses WordStream's 2025 Google benchmarks, Search Engine Land's report of the same dataset, WordStream's 2025 Facebook benchmarks, and the White House de minimis fact sheet. |
| Inputs changed | The baseline records Google's publisher-reported all-industry CPC at $5.26 for 2025, Search Engine Land's prior-year comparator at $4.66 and its conflicting current figure at $5.42, Retail and Shopping CPC at $3.49 with a 3.83% conversion rate, Facebook lead CPL at $27.66 versus $22.87, and the removal of duty-free treatment for covered imports valued at $800 or less effective August 29, 2025 (WordStream Google; Search Engine Land; WordStream Facebook; The White House). |
| Calculation impact | At 1,000 clicks, the built $4.66 to $5.26 comparison moves modeled spend from $4,660 to $5,260, a $600 difference. The Retail and Shopping proxy models 38.3 conversions from 1,000 clicks at the reported 3.83% rate and a $3,490 click cost. You can see the exact formulas and checks above. |
| Remaining limitations | The sources give us no trustworthy universal margin, income, or success rate, and no store-specific revenue, goods cost, delivery, completed-order, fee, return, overhead, labor, or product-duty inputs. That means the model cannot calculate per-order contribution or operating profit for a real store. |
| Hydration review | July 10, 2026. I added the acquisition stress test with each source kept separate, plus the policy boundary. Every unit-economics input missing from P0-08 is now either labeled as an assumption or named as missing. |
What are readers asking?
Is dropshipping still worth trying in 2026?
Maybe, but the verified evidence cannot give every store the same yes or no. It tells us to test paid acquisition against current benchmarks, and it tells us that covered low-value imports lost duty-free de minimis treatment on August 29, 2025. It gives us no trustworthy common margin, income, or success rate. To answer for one store, I need its order revenue, goods and delivery charges, observed acquisition cost per completed order, fees, losses, overhead, labor, and applicable duty (WordStream; The White House).
Should a beginner go all in on ecommerce?
Not from this page alone. Run a bounded test, use sourced inputs, write your stop rule first, and make sure the calculation still works when an assumption moves.
What should a beginner do instead if the numbers fail?
Use the same worksheet on another path. I would keep the required resources, path to a first customer, downside limit, and evidence quality visible before choosing.
WORK WITH KEN
I built the research and checks behind this page as one system. I can build the business version around the way your team works.