This article covers the principles of product optimization – the science of finding the best products and price points to succeed in ecommerce.

(or, why I decided against selling Moon Lights)

In brief: Dropshipping is about capitalizing on ecommerce trends. Finding trending products that work on your site with your inventory, brand and audience can be time-intensive and costly. Here’s a smarter approach to product optimization to make sure your time and money is going towards products your actual customers will buy. Skip to the end of the article for a list of resources and a helpful glossary.

“You gotta spend money to make money” is a universal truism, and for dropshippers, that seems to go double sometimes.

The blessing and curse of dropshipping is that as a shop owner who is unencumbered with owning and stocking physical inventory, the sky’s the limit when it comes to what you want to sell. Shopify and its awesome dropshipping inventory/fulfillment acquisition, Oberlo, make it only a few hours work to set up a store that sells pretty much anything you can imagine… which is a perfect example of what psychologist Barry Schwartz calls The Paradox of Choice. This usually applies to shoppers in a store with too many things to pick from, but what about the store owner in charge of stocking that store — do they face the same paradox?

The answer, dear dropshipper, is yes — and it’s one of the biggest and costliest time-sucks in running your dropshipping operation. To crib another cliche from the universal truism bank… there has to be a better way.

Picking Your Products as a Service (PYPaaS — just kidding)

The Ins and Outs of Inventory Selection

The typical profile of a dropshipper — whether it’s a mompreneur, a hobbyist who wants to monetize their knowledge, a young person learning about ecommerce and business or even a seasoned professional — is someone who wants to some income by running a store where data drives the decision making — pick a niche market by its size and viability, market your selected products to them, cull what fails and expand on what works.

Make more than you spend,

and spend less time doing it.

Expanding on what works is commonly called product optimization — the means by which a store owner (dropshipper or otherwise) finds the best inventory to stock to increase their AOV (average order value). Sometimes that means finding and working with trending products. Sometimes that means A/B testing through paid advertisements on search engines or social media. Ultimately, the goal is the same — make more than you spend, and spend less time doing it.

Image of one dollar bill
There’s a lot more where this came from, bucko — Photo by NeONBRAND on Unsplash

There is a thriving industry devoted to doing that work for you, in the form of blogs and vlogspaid services, and in-shop appsMany of these resources are great and valuable, but unless they’re taking your unique shop into account, the information they’re pushing isn’t always the golden-bullet for hands-free, high-margin sales endeavors.

Thumbnails of youtube videos about best products to sell
You know… these.

Specifically, the idea that some products are trending in general tends to cut across the best practices used by experienced dropshippers — stock a cohesive inventory from known suppliers, reduce ad costs through organic SEO by providing in-shop content on what you sell, and bank on repeat visitors being your highest yielding and least labor-intensive sells. If you have a well-cultivated audience that comes to your site to buy gardening supplies, for instance, then following a blogpost about “Top trending products to dropship this month” to the letter and suddenly shoehorning unrelated products like sunglasses, bluetooth speakers, yoga pants and off-brand diaper genies could do more harm than good. It can confuse your return visitors, cloud your content strategy, and wrestle away time you could better spend optimizing your product selection yourself.

Further, if you build an audience on one trend and then switch gears the next time a blog post suggests new items, you may be spending money on advertising to people who were drawn into your shop temporarily and are now less likely to find your marketing appealing. This isn’t to say following trending products is a bad thing, but rather, it’s only the first step on the way to higher sales and lower ad spends.

About that better way you mentioned…

The art of picking trend products that are right for your store

search screen for sunglasses in oberlo
Time to sell some sunglasses! Oh cool, there are one hundred and sixty three THOUSAND options.

So you’re ready to start working with trending products on your shop, but you don’t want to ‘spray and pray’ new and unrelated items into your inventory — we’re on the right track. Here are a few steps to take to start optimizing your inventory value — with either the inventory you already have or new and trending items — without having to pay out of pocket for every insight.

One store with bike cleats and dumbbells makes sense. Toss in a Bob Marley poster section and things start to get muddy.

Identify your customer base as a guiding light

This is a truly obvious point, but should be at the core of your decisions when introducing new items to an established customer base. If you’re a new store and you don’t have one (or you’re selling general purpose goods), there are many guides that can help you select a good niche. Some dropshippers elect to sell purely by the market value of a niche without much of a concern for their own stake in what they sell, while others prefer to select niches in areas close to their own heart so they can speak on the topic more passionately. One way or the other, your already-won customers are your lowest cost, highest value acquisitions, and are more statistically reliable to produce net-positive revenue for you going forward than building new audiences from scratch. So, even if that Luna Moon Lamp is trending EVERYWHERE, it might not be worth your time or the cohesive value proposition of your shop to try for a slice of that pie (or cheese, in the moon’s case).

*Note: It’s fine for shops to have multiple audiences! A customer base is often composed of smaller divisions of the same type of a shopper — stores that sell sporting goods should market to cyclists differently than they do to powerlifters. Identifying your customer base is about finding the best venn diagram for you that keeps your content and marketing obligations down and your average order value (AOV) up. One store with bike cleats and dumbbells makes sense. Toss in a Bob Marley poster section and things start to get muddy.

DAR.WIN logo

Add inventory incrementally to keep experiments clean

As with any data-driven project, control your variables. Adding new items all at once can make it difficult to investigate changes in your web traffic patterns. Are you getting a lot of new traffic? Increased bounce rate? Higher cart abandonment? Most web metrics are less clear-cut than last-click-attribution, and so if you’re doing a studious job of vetting new products for how they perform, it’s best to give your experiments as much focus as possible. This is crucial when adding items that seem to break with your standard inventory. A new set of bamboo-rim sunglasses showing up on your Print on Demand for New Dads shop might work out great, but it certainly isn’t the closest sibling to the rest of your inventory. Adding in sunglasses, ab rollers and pool floats all at once is basically negating the value of any traffic or sales metrics you produce in comparison to the previous state of the store.

So, take the time to add a new product intentionally, write meta copy about why it’s a good fit for your store (“Provides eye protection during diaper changes!”), and allow the product time to produce accurate metrics.

Goose your new stuff so customers actually see it…

If you already have a large inventory, new products can get lost in the shuffle, artificially suppressing the potential value of keeping that item available. Having a “New In The Shop” or “Hot Products” collection is an easy way to break a new product out of its standard collection or category. If you have a homepage slider featuring products you can select or a regular newsletter, these are louder outlets (meaning you are more aggressively putting your thumb on the scale in a featured product’s favor) but can also be an accelerant to pull attention to new stock. Keep in mind that any inorganic treatment an item gets (in other words, how you treat or feature it against how your standard stock performs) can affect the longer-term value of your metrics. Obviously selling inventory is always good, but if you’re running experiments for product optimization, artificially inflated presence of a new product in your shop can create a false impression about a product’s value.


… but keep your promotion focused.

You should be able to make some assumptions about distinct groups in your general customer base — demographics, locations, specific collections of interest. If you’re actively segmenting your customers for marketing purposes, try to keep promotion of items you’re experimenting with to those relevant segments. It will reduce the amount of marketing spend going towards an experimental effort and reduce friction from shoppers less likely to engage with new products of that type. This is the core of product optimization — expose new inventory to likely buyers and invest after you have evidence of engagement.

Customer partitioning can be done in a variety of means with varying levels of reliability (from customer surveys to unique sign-up forms to order history). DAR.WIN is an app available for Shopify that automates partitioning by assessing customer behavior patterns to generate unique Visitor Profiles for your shoppers. DAR.WIN users can install DAR.WIN for free, export lists of customers in each segment, and send targeted emails to that segment featuring products they’re most likely to buy (selected by machine learning algorithms). For product optimization purposes, DAR.WIN offers a Product Highlight feature that allows shop owners to feature a hand-picked product in line with automated recommendations. To learn more about how DAR.WIN makes product optimization painless, quick and cost-effective, watch for part two of this article or check out our YouTube channel for gems like this:

The narrator on this video sounds devastatingly handsome, doesn’t he?

NOW spend money — but only on your remaining unknowns

OK, the dads are buying the bamboo shades.

A real thing you can put on your face.

You’ve established that your experiment works (a target audience in your customer base is positively engaging with your new product selection). Now it’s time to experiment with volume, margins and bundling. Volume is the amount of a piece of inventory you can move (with dropshipping, your supplier is on the hook for fulfillment, so make sure that part of your supply chain is capable of fulfilling your needs). Margins are how much money you make each time you sell an item, cut against the cost to stock an item and market it.

*If you want to get incredibly good at balancing fulfillment capacity and optimal margins, play this paperclip factory game— but be warned, by clicking on that link, you’re about to lose hours becoming a paperclip magnate (but get a free B-School education in the process).

A burger might be sold at cost or even at a loss, but the selling a soda and fries for hundreds or thousands of times the cost to produce is the genius of working products together as a bundle.

Finally, we have bundling, which we can use as a catch-all term for how you frame your product’s sale proposition in light of your other inventory. Your bundling approach may involve recommended items (another thing DAR.WIN is built to do), it may involve chaining items into suggested sets during checkout, or it may even mean taking a product and selling it with other items under a single SKU. The idea is that once your meta copy is perfected and you feel the product has been introduced successfully to your audience, what is the highest volume/highest margin way to get it into carts — again, increasing your Average Order Value.

*Bundling and Margin science to achieve your best overall AOV is a topic worth a full course rather than just a mention in an article. The most famous example is fast food value meals. A burger might be sold at cost or even at a loss, but the selling a soda and fries — which cost pennies to stock and move at the volume they sell for in each value meal — at hundreds or thousands of times to cost to produce is the genius of working products together as a bundle. Where one individual item’s margin might be depressed, the bundle as a whole achieves a higher AOV than any item sold individually. Your ongoing experiments can include these kinds of trials — maybe the bamboo sunglasses can only sell at a 5% margin, but pairing them with sunscreen or a sunglasses case and pushing the value of those items higher may mean more absolute revenue than you’d make otherwise.

The title of this section is ‘NOW spend money’ — but we haven’t yet. Essentially, the investment now is classic A/B testing, which we’ve managed to avoid until this point. We avoided it because A/B testing works best when the variables are identified and minimized and costs can be kept down. Our goal is to achieve the sales knowledge produced in such tests, not throw as much money as we can to an ad buy. Obvious trials include selling singleton versions of your new inventory vs. bundled versions, different price points (again, balancing the volume of sales you can generate against the margin being won from those sales), and even different audience segment targeting (if an item is unisex, like sunglasses, try a few ads appealing more specifically to dads, fitness women, etc. to discover new market penetrance). If you need help getting started with A/B testing, specifically on Facebook (a common marketplace for these kinds of experiments), this article is a great place to start.

*For dropshippers looking for an advanced strategy, look into multivariate testing — an emerging discipline in ecommerce that condenses the time and sample size required to make quality conclusions on ad tests while running several variable at once. Since there are exponentially more factors at play in multivariate testing, data are usually provided through computational processing like machine learning.


Annnnd stop the ads or wag the dog (I promise that’ll make sense)

Trends are, by definition, things that artificially swell in interest for a set period of time, only to ebb later. If you’re adding new items based on trend articles (a totally reasonable thing to do), pay attention to when your items fall into the ebb phase naturally. This is super easy for seasonal items. Whether or not you leave them listed and live in your inventory is one matter, but make sure that your marketing spend isn’t fighting an uphill battle to sell pool floats in November.

The exception to this rule is if you’ve landed the pretty rare status as the place to go for the thing. That means you have organic listings, word of mouth, steady and/or growing traffic, and ongoing sales for your experimental item. Dropshipping, at its core, is about experimenting to find what works — sometimes, as in this example, success find you in spite of what you thought your store was. An example might be a local farmstand that sells produce, jarred preserves and then decides to start selling honey sticks. The sticks are high margin, have staying power, and are easier to move than the other stuff that has less of a shelf life. That farm shop can stay close to home when honey season (if there is one) sails by — or they can revamp the entire stand to capitalize on their new mountain of liquid gold. Either one leaves money on the table — the question is which is a better bet for the goals of the vendor (an exit strategy to sell the stand at a period of inflated value? A dynasty stand that lasts 30 years on stable margins? These examples are both applicable to online store).

Liquid Gold! Photo by Amelia Bartlett on Unsplash

While the breakout success of an experimental item doesn’t require that you change from a baby clothing store to a sunglasses shop, for instance, you may find yourself in a position where that’s what makes sense. Brand reconstruction is time consuming and costly, so this should be considered a pretty rare case — but if the sales are growing and the market seems steady, it may be the best decision to make based on your experiments. Famously, it’s questioned whether the dog wags its tail or the tail wags the dog — in this case, you’ve optimized your product to the point where it’s wagging the rest of your store. Make sure you have a plan for that possibility.

Product Optimization = Store Optimization

What we’ve discovered is that there are ways to add experimental items — often trending, often outside of the typical profile of our shop’s standing inventory — in a way that can save time and money as opposed to brute force methods. We’ve also learned that products can be optimized — in the discovery of their best segment of your customer base, in their sweet-spot pricing and bundling options — through procedural experimentation. And, hopefully, it’s become apparent that this technique applies not just to your new inventory selections, but to your entire store as a whole. Using apps that leverage machine learning help you make better data-driven decisions about design, copy, pricing, customer profiling, ad strategies and full-blown branding shifts. As dropshippers, data-driven is what we are looking for to make intelligent business decisions (and is the hard of product optimization) that yield higher net profits. Poring over data in Google Analytics has been something wise webmasters have done for years to optimize their site for desired behavior flow — as a shop owner, using more ecommerce-focused tools can help achieve the same result.

Wrapping up

Not to keep hammering at the same point (but really, what are wrap ups for?), your goal as a dropshipper is to spend less money to sell more items for the highest profit you can generate from those sales. Product Optimization — the process we’ve outlined where you find the best way to sell your inventory (especially new or trending inventory) for the most money to the most people — is essential to meeting that goal as a store owner. While there are many approaches that might work, our recommendation is that you utilize a machine learning program for large data experiments while saving your money and elbow grease for more refined conclusions that become evident after your broader experimentation.

While all of this work might seem like a lot of labor to locate a sweet spot for a single product, I’ll offer one bit of wisdom I’ve gleaned from working with dozens of dropshippers — one killer product, properly optimized, can make a store. It can bring it from a side hustle to a full-time gig. It can change monthly revenues from hundreds to tens of thousands. It can pivot an entire business strategy. But without the intentional, procedural approach to discovering and refining that product outlined in this article, there’s no telling whether or not your store is that one perfectly optimized product away from becoming that kind of enterprise.

Good luck and happy dropshipping!

Recommendations from this article:

Oberlo — Dropshipping inventory sourcing and fulfillment
(free with paid plans at volume)

DAR.WIN — Product Optimization with Machine Learning
(free with paid plans for professional tools and longer data history)

Product Genie — Inventory Discovery
(free three day trial with paid plans)

Learning Resources:
Oberlo Newsletter — General dropshipping wisdom and trending items

AliExpress Dropship Club — Bite-sized videos on different dropshipping topics

Dropship Lifestyle — Frequently updated blog with great info on niche discovery

A Crash Course on A/B Testing Facebook Ad Campaigns from ConversionXL

Multivariate Testing Defined

Machine Learning 101 from a Googler

Just For Fun:
Universal Paperclips Game— become a blackbelt in understanding supply chain and pricing while losing hours of sleep.


(NB: these definitions can be very expansive, and their description here is to clarify their use in this article rather than to provide a comprehensive outline of the broader topics each invokes):

Stands for Average Order Value. This is the flat average of the monetary worth of what’s been ordered in consideration of how much it costs to fulfill that order. This is typically calculated by sales revenue (your bottom line not counting expenses) / # of orders. A tighter metric for experimental store runnings is sales profit (your bottom line after operational and marketing costs have been deducted) / # of orders. The value in this case comes from profit and sustainability, rather than the number of zeros on the price tag.

Machine Learning
In our context, Machine Learning is the use of computational algorithms to produce actionable datasets. In other words, having a computer crunch large numbers to provide information to a user that would be difficult to discern otherwise. Some examples include averaging traffic patterns of all shoppers, parsing patterns within subsets of shoppers, and creating data-driven product recommendation sets from those patterns.

Essentially, the opposite of ‘gut’ or ‘instinct’ as a means by which business owners make decisions. Data-driven operations merely show the averages of observed interactions across a large sample set — watching how all shoppers behave on your checkout page, for instance, rather than getting a small number of poll answers from volunteers. Often, data-driven conclusions butt up against gut feeling — odd product pairings or strange inventory pivots. Neither approach can be empirically put at odds with the other, but data-driven hypothesis can usually be tested more conclusively than human assumptions.

A/B Testing
Traditionally, the discipline of running two ads with a single variable to optimize performance. These variables are things like images, copy, pricing, bundling, demographic targeting, and more. The more tightly the variability is defined, the better the conclusions will be drawn at the end of the trial. The goal of A/B testing is to spend less money on your ad buy in general by only promoting your most proven value propositions.

Multivariate Testing
A newer model of ad testing (evolving beyond A/B testing) where multiple variables can be tested at once, drawing more concrete observations from smaller data sets. Multivariate testing is typically supported by machine learning services and is super valuable to read up on, but is better approached once you’re comfortable with ad testing at a more hands-on level.

A specific population of consumers likely to purchase your inventory whom you can target with ads. Broad niches require more overhead and are more difficult to make concise assumptions about (since they include people with different goals and preferences), while overly specific niches may not be populated well enough to support ongoing sales efforts. Dropshippers typically research a niche based on how saturated that niche is with options already, how much money they can make from inventory targeting that niche, and how well they can speak to members of that niche from a place of authority. Your task as a dropshipper is to balance the size of your selected niche so it’s big enough to sustain operations but concise enough to speak to with a degree of generality.

A segment is a sub-niche — in other words, distinct cohorts of shoppers in your selected niche. For example, your store’s niche market might be new parents. One segment in that niche is dads, and another is moms. Extending that, you may also have baby shower buyers, non-native speakers, first-time parents vs. parents of several children, and more. Segmentation of your niche audience allows you to spend less money and time marketing products to more highly attuned buyers. Segmentation can be done through a variety of means, either through self-identification of the shopper (like in a questionnaire at checkout or on an email signup form), or can be done programmatically by a machine learning app like DAR.WIN.

Organic vs. Inorganic
In our context, organic refers to traffic you didn’t directly pay for (people finding you on a search engine or social media by searching for terms related to your shop and finding you in non-ad results) and inorganic traffic refers to traffic you pay for (when visitors click through ads and promoted posts to get to your site). Obviously, organic results are the result of SEO (search engine optimization) labor, which while it may not have a cost per click associated, is certainly expensive in terms of manhours and, if you choose, consultancy.

Search Engine Optimization. The various means by which you can have your site (or individual pages on your site) appear in SERPs (search engine results pages), and how to appear more prominently for certain valuable search terms.

Value Proposition
The framing of why your product is worth purchasing or your store is worth patronizing. These can be concrete (like pricing, shipping time, personalization) as well as emotional (ethical product sourcing, support for artisans, partial donation of proceeds). A value proposition is meant to increase your profit — whether that is long-term or short-term is a matter of customer acquisition and loyalty strategies.