The Data-Driven Store — Using Analytics to Pick Products, Kill Losers & Scale Winners in Algeria (2026)
📊 "I Think This Will Sell"
That phrase. Five words. Said in every wilaya, by every merchant, before every product launch.
It has killed more Algerian e-commerce stores than COD rejection, shipping delays, and supplier fraud combined.
Here is why.
When you pick products based on intuition, you are not making a decision. You are making a bet. A bet that your gut feeling about what Algerians want to buy is more accurate than actual data about what Algerians are already buying.
That bet loses 70% of the time.
Up to 70% of new e-commerce sellers fail within their first year — and the most common cause is not bad marketing, not poor customer service, not even pricing mistakes. It is sourcing products without validating demand, competition, or margins first.
The merchant who says "I think this will sell" is the merchant who, six months later, is sitting on 300 units of something nobody wanted, wondering where the profit went.
The merchant who says "the data says this will sell" is the one still in business.
This blog is about becoming the second merchant.
You will learn how to read external data (Facebook Ad Library, Google Trends, competitor intelligence) to find products before they saturate. How to read your own store data to know which products to feed and which to starve. The exact thresholds for when to kill a product — not based on feelings, based on numbers. And how to build a product portfolio where every SKU earns its place.
No gut feelings. No "I think." Just data.
🔍 External Data — What the Market Is Already Telling You
Before you spend a single dinar on inventory, the market has already told you what sells. You just need to know where to look.
Facebook Ad Library — The Free Product Validator
The Meta Ad Library is a free, public, searchable database of every active ad running on Facebook and Instagram. Access it at facebook.com/ads/library. No account required.
Here is what makes it the single most underused product research tool in Algeria.
The start date tells you everything.
| Ad Age | What It Means | What to Do |
|---|---|---|
| 60+ days running | The product is profitable. No one keeps spending on a losing ad for two months. | Highest-priority signal. Study the product, the creative, the landing page. |
| 30–60 days | Survived the test phase. Gaining traction. | Monitor. If it survives to 60+, validate for your store. |
| 7–30 days | In testing. Most ads die here. | Watch but don't act. Most won't survive. |
| 1–7 days | Brand new test. | Note for awareness. Come back in two weeks. |
Research across 47,000+ ads found that only about 11% survive past 60 days. When you find one that has, you have found a validated product. The market has already paid to test it for you.
How to use it for the Algerian market:
- Go to
facebook.com/ads/library, set country to Algeria - Search product keywords in French and Arabic: "écouteurs sans fil," "montre connectée," "سماعات بلوتوث," "produits beauté"
- Filter by Active status, then sort by impression volume
- For every ad running 60+ days, document the product, price point, creative style, and landing page
- Build a watchlist of 8–12 competitors. Check it weekly.
The cross-brand signal. When three or more unrelated Algerian brands are all running long-lived ads for the same product category — wireless earbuds, hair care tools, kitchen gadgets — that is not coincidence. That is a market-level signal that the category converts.
The Longevity Trap — What the Ad Library Cannot Tell You
The Ad Library shows you what is running. It does not show you click-through rates, conversion rates, or ROAS. It does not show you how much they are spending or whether they are profitable.
A product can run for 90 days on a 2× ROAS that loses money on every sale. Or a product can run for 30 days on a 0.5× ROAS funded by venture capital. You cannot see the difference in the Ad Library.
The fix: Combine Ad Library signals with marketplace validation. Search the product on Ouedkniss. Check if anyone is selling it on Facebook Marketplace. Look at the comments on the ads — are people tagging friends with "we need this" or are they complaining about price and quality?
The Ad Library tells you what is being advertised. The marketplace tells you what is being bought. You need both.
Google Trends — Demand Direction in 30 Seconds
Google Trends is free. It takes 30 seconds to check. And almost nobody in Algerian e-commerce uses it.
Here is how to read it.
| Curve Shape | What It Means | Decision |
|---|---|---|
| Steady rise over 3+ months | Demand is building. Not a fad. | ✅ Proceed with validation |
| Spike then crash | Viral moment. Already fading. | ❌ Avoid — unless you can source and ship within days |
| Flat but stable | Evergreen demand. Not growing, not shrinking. | ✅ Proceed if saturation and margin are favorable |
| Seasonal peaks, predictable pattern | Ramadan, back-to-school, winter coats. | ✅ Time entry 3–4 months before the peak |
| Declining over 6+ months | Market is shrinking. | ❌ Skip. You are entering a dying category. |
Three Algeria-specific searches to run right now:
- Set region to Algeria, time range to Past 12 months, and search your product category. Is the line going up, down, or sideways?
- Run the same search for Past 5 years. Is the 12-month trend part of a larger rise or a temporary blip?
- Compare your product against a known winner. If you are considering "home coffee machines," plot it against "machine à café" and see which has stronger, more consistent demand.
The related queries goldmine. Scroll down to "Related queries" and filter by Rising. These are search terms growing fastest right now — often sub-niches or product variations that have not yet appeared on any "winning products" list. A rising query like "chargeur sans fil voiture" (wireless car charger) tells you something specific is catching fire before the broad market notices.
Competitor Scraping — The Ouedkniss and Facebook Marketplace Signal
Your competitors are running live experiments with real money every single day. Their listings are free data.
The Ouedkniss scan (weekly, 20 minutes):
- Search your product category on Ouedkniss
- Sort by "Plus récentes" (most recent)
- Count how many new listings appear per week. Increasing velocity = growing category.
- Note which listings get marked as "Vendu" (sold) within 48 hours. Fast sales at asking price = strong demand.
- Track the price range. If the bottom of the range is dropping, the market is racing to zero. If the top holds steady while new entrants appear at lower prices, there is still premium positioning available.
The Facebook Marketplace scan (weekly, 15 minutes):
Same process. But also: check how many sellers are listing the same product. If you see the identical stock photo on 15 different Marketplace listings, the product has hit saturation. Margins are gone. Move on.
📈 Internal Data — What Your Own Store Is Already Telling You
External data tells you what to sell. Internal data tells you what to keep selling, what to fix, and what to kill.
Most Algerian merchants cannot answer these three questions about their own store:
- Which product made you the most profit last month — not revenue, profit?
- Which product has the highest COD rejection rate?
- Which product has not sold a single unit in 60 days?
If you cannot answer all three, you are flying blind. Here is how to fix that.
The Only Eight Metrics That Matter
| # | Metric | How to Calculate | Why It Matters |
|---|---|---|---|
| 1 | Net profit per unit | Selling price − (COGS + shipping + packaging + COD fee + return allocation) | Revenue is vanity. This is survival. |
| 2 | Sell-through rate | Units sold ÷ units available (over 4 weeks) | How fast inventory turns into cash. Below 30% = problem. |
| 3 | COD rejection rate by product | Rejected deliveries ÷ total deliveries (per SKU) | Some products attract serial rejecters. Find them. |
| 4 | Gross margin % | (Selling price − COGS) ÷ selling price | Below 40% leaves no room for ads, returns, or mistakes. |
| 5 | Customer acquisition cost (CAC) | Total marketing spend ÷ new customers acquired | If CAC > first-order profit, you are buying customers you cannot afford. |
| 6 | Repeat purchase rate | Customers with 2+ orders ÷ total customers | The difference between a business and a series of one-time transactions. |
| 7 | Days of inventory outstanding | (Average inventory value ÷ COGS) × 30 | How many days your cash is tied up in unsold product. Rising = danger. |
| 8 | GMROI (Gross Margin Return on Investment) | Gross profit ÷ average inventory cost | If this is near or below 1.0, your inventory is not earning its keep. |
If you track nothing else, track these eight. Every product decision in the rest of this blog flows from them.
The Product Segmentation Matrix — Where Every SKU Lives
Not all products are equal. Treating them as if they are is how you starve winners and feed losers.
Segment every product in your store into one of eight categories based on traffic volume × add-to-cart rate:
| Category | Traffic | Add-to-Cart Rate | What It Means | What to Do |
|---|---|---|---|---|
| ⭐ Stars | High | Above average | Proven winners. Customers find them and buy them. | Feature prominently. Never discount. Protect margin. |
| 📦 Essentials | High | Average | Reliable workhorses. | Small incentives, bundles, subscription offers. |
| 🔧 Bottlenecks | High | Below average | People are interested but something on the page is killing conversion. | Fix the page — images, reviews, description, load speed. Do NOT discount. Discounting a Bottleneck fixes the wrong problem. |
| 💎 Gems | Low | Above average | Hidden winners. High conversion but low visibility. | Best growth opportunity. Increase traffic immediately — ads, homepage placement, email features. |
| 🌱 Prospects | Low | Average | Could go either way. | Test with more traffic. Give them 30 days. |
| ⏳ Underperformers | Low | Below average | Weak on both dimensions. | Deprioritize. Move to bottom of category page. |
| 👻 Invisibles | Very low | Any | No one sees them. | Drive traffic or archive. A product no one views is dead inventory. |
| 🛑 Stoppers | 100+ views | Zero add-to-carts | Something is actively wrong. | Diagnose immediately — is the price absurd? Are the images broken? Is the description in the wrong language? Fix within 48 hours or remove. |
The rule most merchants break: Never discount a Star. Never discount a Bottleneck. Stars don't need discounts — they sell at full price. Bottlenecks don't need discounts — they need better product pages. Discounting either is leaving money on the table for no reason.
☠️ The Kill Criteria — When to Stop Selling a Product
This is the hardest section in this blog. Not because the math is complicated. Because killing a product feels like admitting failure.
It is not. It is admitting you have better places to put your money.
Every product you keep that should be dead is consuming cash, storage space, and — most importantly — your attention. The product that generates 2% of your revenue but consumes 15% of your mental energy is not an asset. It is a parasite.
Here are the exact thresholds. When a product crosses one, you act.
The Five Kill Signals
| Kill Signal | Threshold | Why This Number | Grace Period |
|---|---|---|---|
| Sell-through rate collapse | Below 30% for 8 consecutive weeks | Below 30% means your inventory turns less than 4× per year. Cash is trapped. | 8 weeks. If it does not recover with a fix attempt, cut. |
| Days of inventory > 120 | Stock on hand ÷ daily sales rate > 120 days | Your money has been sitting in a box for four months. That money could be in a product that sells. | None. Liquidate immediately. |
| GMROI below 1.0 | Gross profit ÷ average inventory cost < 1.0 | Every dinar tied up in this product generates less than one dinar in gross profit. You would make more money doing nothing. | Two measurement cycles (8 weeks). If it stays below 1.0 after a fix attempt, cut. |
| COD rejection rate > 25% | More than 1 in 4 deliveries rejected | This product attracts buyers who change their mind. Each rejection costs you two-way shipping + COD fee + lost opportunity. The product itself might be fine — but the buyer profile it attracts is killing your unit economics. | 4 weeks to test: change product photos to be more accurate, add size/dimension clarity, adjust description to reduce expectation gap. If rejection stays above 25%, kill. |
| Negative profit after all costs | Net margin < 0% for 3 consecutive months | You are paying customers to take this product. | One month. You cannot fix a product that loses money on every sale unless it is a deliberate loss leader driving bundle purchases. |
What "Kill" Actually Means
Killing a product does not always mean throwing it away. It means stopping the bleed.
| Product Type | Disposal Strategy | Recovery |
|---|---|---|
| Sellable, just slow | Bundle with a Star product. "Buy X, get Y at 50% off." | Recovers ~40–60% of cost |
| Sellable, wrong audience | Flash sale to your WhatsApp broadcast list. "Last chance — 48 hours only." | Recovers ~50–70% of cost |
| Unsellable at any price | Donate to charity. Document it. Use the receipt for tax purposes. | Tax deduction |
| Niche but not dead | List on Ouedkniss at cost. Someone somewhere wants it. | Recovers ~60–80% of cost |
The goal is not to recover 100%. The goal is to free up the cash, the shelf space, and the mental bandwidth for products that actually make money.
🚀 The Scale Signal — When to Double Down on a Winner
Killing losers is defensive. Scaling winners is offensive. Most merchants are bad at both — but they are worse at scaling.
Here is why: they do not recognize a winner when it is staring at them.
A product that sells well feels like luck. So they treat it gently. They don't want to "jinx it." They keep inventory lean, spend cautiously on ads, and cross their fingers that it keeps selling.
That is not strategy. That is fear dressed as prudence.
The Five Scale Signals
When a product hits three or more of these, you do not "monitor" it. You pour fuel on it.
| Scale Signal | Threshold | What It Tells You |
|---|---|---|
| Sell-through rate > 70% | More than 70% of stocked units sell within 4 weeks | Demand is outrunning supply. You are leaving sales on the table every day you are understocked. |
| 3+ consecutive months of growth | Units sold increasing month-over-month for 3+ months | This is not a spike. This is a trend. |
| Repeat purchase rate > 25% | More than 1 in 4 buyers comes back for the same product | The product delivers on its promise. Customers want more. This is the strongest signal in e-commerce. |
| CAC < 30% of first-order profit | You recoup your ad spend in the first purchase and then some | You can scale ad spend without destroying unit economics. This is rare. When you find it, push. |
| COD rejection rate < 10% | Fewer than 1 in 10 deliveries rejected | The product matches expectations. The buyer profile is high-intent. Low rejection = low friction = high scalability. |
What "Pouring Fuel" Looks Like
| Action | Timeline | Expected Impact |
|---|---|---|
| Double inventory order | This week | Stop stockouts. Every day out of stock is a day of zero revenue on a proven winner. |
| Increase ad budget by 50% | This week, monitor daily for 5 days | If CAC holds under 40% of first-order profit, increase again. |
| Add to homepage featured section | Today | Give your winner the most valuable real estate in your store — free traffic, zero ad cost. |
| Create a bundle with a slower product | This week | Winner sells the bundle. Slower product gets adopted. Both win. |
| Order samples for 2–3 variants | This month | Colors, sizes, complementary products. Winners spawn product lines. |
| Negotiate volume pricing with supplier | This month | If you are doubling orders, your supplier should give you a better price. Ask. |
The One Rule of Scaling
Never scale a product you have not personally tested.
Order your own product. Open the box. Use it. Check the quality. If you would not be happy receiving it as a customer, do not spend a single dinar on ads. Ads accelerate everything — including disappointment. A product with a 5% return rate at 10 orders a day becomes a customer service disaster at 100 orders a day.
🧩 Building a Product Portfolio — Not Just a Product List
A product list is a collection of things you sell. A product portfolio is a collection of things that work together.
The difference is the difference between a store that survives and a store that scales.
The ABC Portfolio Framework
Rank every product by profit contribution. Then classify.
| Tier | % of Products | % of Total Profit | What They Are | How to Manage |
|---|---|---|---|---|
| A — The Core | Top 20% | 70–80% | Your Stars and high-volume Essentials. The products your store is known for. | Intensive management. Weekly performance review. Never out of stock. Protect margins. Deep supplier relationships. |
| B — The Support | Middle 30% | 15–20% | Reliable performers. Gems you are growing. Prospects with potential. | Monthly review. Keep stocked. Test traffic increases on Gems. Watch for B→A migration signals. |
| C — The Tail | Bottom 50% | 5–10% | Underperformers, Invisibles, slow-movers, seasonal items off-peak. | Quarterly review. Kill anything below kill thresholds. Consolidate variants. Consider if the operational complexity is worth the contribution. |
Here is the uncomfortable truth about your C-tier.
If the bottom 50% of your products generate only 5–10% of your profit, and each one consumes roughly the same amount of your attention as an A-tier product — you are spending half your mental energy on 10% of your outcome.
Cut the bottom 20% of your C-tier today. Not next month. Not "when I have time." Today. Pick the five worst-performing products in your store and remove them. You will feel lighter within an hour.
The Portfolio Health Check (Monthly, 30 Minutes)
| Check | What to Look For | Action |
|---|---|---|
| A-tier count | At least 3–5 products generating consistent profit | If fewer than 3, your business is fragile. Find more A-tier products. |
| Concentration risk | No single product > 40% of total profit | If one product dominates, one supplier problem or trend shift kills your business. Diversify. |
| Gems migration | At least 1–2 products moving from B to A per quarter | If nothing is ascending, your product pipeline is stagnant. |
| Kill list | Products below kill thresholds | Cut them. This month. Every month. |
| New entries | At least 1–2 new products tested per month | If you are not testing, you are not growing. You are just maintaining. |
The Product Lifecycle — Every Product Dies Eventually
Every product in your store is somewhere on this curve.
| Stage | Signal | Strategy |
|---|---|---|
| 🌱 Launch | First 30 days. Low traffic, unknown conversion. | Test with small inventory. Validate demand before committing. Most products die here. That is normal. |
| 📈 Growth | Rising sales, improving rank, repeat purchases appearing. | Increase inventory. Scale ads. Capture market share before competitors notice. |
| ⭐ Maturity | Peak sales, stable margins, repeat purchase rate plateaus. | Protect margins. Bundle with newer products. Consider variants. |
| 📉 Decline | Sales falling, CAC rising, competitors undercutting, supplier raising prices. | Reduce inventory orders. Stop ad spend. Prepare exit strategy. |
| ⚰️ End of Life | Below kill thresholds. | Kill clean. Liquidate inventory. Remove from store. Move on. |
The mistake most merchants make is holding on through Decline, hoping for a reversal that never comes. Products do not resurrect. When the data says decline, listen.
🔄 The Weekly Data Routine — 45 Minutes That Replace Gut Feelings
You do not need to live in your analytics. You need a repeatable weekly routine that surfaces the signals before they become problems.
Here is the routine.
Monday Morning — 45 Minutes
| Time | Task | Tool | What You Are Looking For |
|---|---|---|---|
| 0–10 min | Check top 5 products by profit, top 5 by revenue, bottom 5 by profit | DZBuild Analytics or spreadsheet | Any surprises? A product you thought was a winner actually losing money after returns? A quiet product suddenly profitable? |
| 10–20 min | Scan sell-through rates. Flag anything below 30%. | Inventory report | Products approaching kill thresholds. Restock alerts for products under 2 weeks of supply. |
| 20–30 min | COD rejection rate by product. Flag anything above 20%. | Order + delivery report | Products attracting bad-fit buyers. Pattern: same product, same wilaya, high rejection = fix the listing or kill the product. |
| 30–40 min | Facebook Ad Library scan. 5–10 competitors. Note new long-runners. | facebook.com/ads/library | What are competitors betting on? Any validations you can piggyback on? |
| 40–45 min | Google Trends pulse check. 3 product categories. | trends.google.com | Any sharp movements? Rising queries you haven't seen before? |
Monthly Deep Dive — First Monday of the Month, 90 Minutes
| Time | Task |
|---|---|
| 0–30 min | Full ABC portfolio reclassification. Run the eight core metrics on every product. |
| 30–45 min | Kill list: products crossing kill thresholds. Decide: fix or cut. |
| 45–60 min | Scale list: products hitting 3+ scale signals. Decide: how much fuel. |
| 60–75 min | Competitor deep dive: what new products did competitors launch? What did they stop advertising? |
| 75–90 min | Pipeline planning: what products to test next month? What categories to enter? What to exit? |
That is it. 45 minutes a week. 90 minutes a month. Less time than you spend answering WhatsApp messages.
And it replaces every "I think this will sell" with "the data says this is selling."
🏁 The Store That Knows
The merchant who launches products based on intuition and the merchant who launches products based on data are playing two different games.
The first merchant is gambling. They might win. Some do — for a while. But they don't know why they won, so they cannot repeat it. And when they lose, they don't know why they lost, so they cannot fix it.
The second merchant is building a system. Every product decision — launch, scale, fix, kill — is backed by a number. When they win, they know which metric to credit. When they lose, they know which threshold was crossed and when.
Here is what the data-driven Algerian store looks like:
- It finds products by reading Facebook Ad Library signals and Google Trends curves — not by scrolling TikTok and hoping
- It knows the net profit of every SKU after shipping, COD fees, packaging, and returns
- It kills products at specific thresholds — not when the founder finally admits defeat
- It scales winners at specific signals — not when the founder feels brave
- It runs a 45-minute weekly routine that surfaces problems before they become crises
- It treats its product portfolio like a portfolio — not a junk drawer
You can build this store. You do not need a data science degree. You do not need expensive tools. You need discipline, the eight metrics, and the willingness to let numbers overrule ego.
Start with the weekly routine. 45 minutes next Monday morning. Open your store data. Find your bottom five products by profit. Ask yourself: if I killed these today, what would I lose? If the answer is "almost nothing" — kill them.
That is your first data-driven decision. It will not be your last.
The DZBuild Team We build the platform so you can build the business.