THE CURRICULUM / CH 2 — Shopify Essentials

2.1Shopify's Must-Have Apps

A decision procedure for software: the job, the candidate test, and when the answer is no app.

6 min read

It is a Sunday and you have installed nine apps and sold nothing. A badge row, a countdown timer, a bundle builder, and two review apps, because a video recommended one and a list recommended the other. On a phone, your price now arrives after your buy button does.

THE RULE, AND ITS ONE EXCEPTION

What installing one app actually costs, drawn as five lines. The monthly fee is the only one that arrives as a bill. Underneath it, four costs that never appear on an invoice: page weight, because scripts load on every page rather than only where the feature is used; script conflicts, because two apps touching the cart will eventually disagree; theme residue, because uninstalling does not remove what the app wrote in; and attention, because every app is one more panel to learn and one more suspect when something breaks. The four are drawn equal on purpose and no proportion is claimed.
The fee is the only line you agreed to.
  • Everything here acts on traffic you already have, so it can wait until that traffic exists.
  • The exception is email and SMS capture. Contacts you fail to collect this month are not recoverable later.

JOBS DO NOT EXPIRE. NAMES DO.

apps get bought, renamed, repriced and abandoned

THE SIX JOBS

WHAT EACH CATEGORY IS ACTUALLY FOR

The jobWhat it has to achieveWhen to fill it
Reviews and social proofEvidence a real person received this and it matched the page.Once you have customers.
Upsell and cart valueIncrease what an already-convinced buyer takes. The word is already.After the page converts.
Shipping and trackingTell the customer where the order is before they ask.Before the first order ships.
Analytics and attributionWhich ad, page and product produced the order.When platform and admin cannot answer between them.
Page buildingLay out a page the theme cannot produce.Last, and reluctantly. Hardest to reverse.
Email and SMS captureOwn a channel that does not charge you per view.Early. The one exception.

A third tool that also disagrees about attribution is not clarity.

How a stranger's trust responds to reviews, drawn as two shapes rather than as numbers. The first is a curve of trust against how many reviews there are: it climbs steeply out of nothing, because the first few are the difference between nobody having said anything and somebody having, and then it flattens, because each one after that adds less. The second compares the shape of the ratings themselves. A wall of perfect ones reads as chosen rather than collected, and leaves nothing to weigh it against. A few with a spread across them reads as collected rather than chosen, and the unflattering one is what makes the rest readable. No counts, ratings or star figures are shown, and both the curve and the two spread shapes are illustrative.
The first few do the most work. A flawless wall reads as chosen.

WHAT AN APP COSTS BEYOND THE FEE

What makes a store page slow, drawn as five weights stacked on the page beneath them. Images and video are the heaviest, and the heaviest thing loads before anything can be read. Apps and their scripts come next, because every page pays for a feature that one page uses. Then tracking and pixels, because measurement is not free and loads too. Then chat, popups and widgets, which arrive before they are wanted. Then fonts and theme extras, because a second typeface is a second download. Underneath all of it is the page itself: a photo, a price and a button, carrying the lot. The thickness of each weight is a rank, not a measurement, and no timing figure appears.
Scripts are the second heaviest thing on the page.
  • Script conflicts. Two apps that both modify the cart will disagree eventually, and the symptom is a checkout that fails for some customers and works every time you test it.
  • Residue. Uninstalling does not reliably remove code an app wrote into your theme, so stores load dead snippets from software deleted a year ago.
The path from add-to-cart to paid, with the specific reason a stranger leaves at each step. Add to cart: the button has scrolled out of reach. The cart: a total that does not include shipping yet. Checkout opens: it looks like a different company. Details: forced account creation and too many fields. Shipping: the cost and the delivery date both seen for the first time here. Payment: no method they recognise, or a currency to convert. Every exit is a reason, not a rate.
A cart app operates here, invisibly from your admin.

THE FEE IS THE CHEAPEST THING AN APP COSTS YOU.

the other three lines never appear on a bill

Statement poster: "Your store is not slow. Your offer is."

THE CANDIDATE TEST

  1. Write the job in one sentence, before the app store is open.
  2. Check whether your theme already does it. A theme section adds no script.
  3. Sort reviews by most recent. Old praise plus recent complaints means it changed hands.
  4. Check the last update date and whether the developer replies.
  5. Read the permissions. An app asking for customer data to do a job that does not involve it is telling you how it makes money.
  6. Install onto a duplicate theme, then load your product page on a real phone before and after.
  7. Set a review date as you install. If you cannot point at what it did, remove it.
The above-the-fold stack of a product page on a phone: product name a stranger can parse, one image that reads at thumb size, the price plus whether tax and shipping sit on top, one line saying what it does or who it is for, and an add-to-cart reached without stretching that does not move. Below the fold, the remaining questions in the order they arrive: is it for me, what does it look like in real life, what exactly do I get, who else bought it, what if it is wrong, when does it arrive, and how do I pay. No load time, price or rate is shown.
What you load before and after an install.

READING YOUR OWN APP LIST HONESTLY

SignalHealthyAverageConcerning
Effect on page loadNo delay you can feelWeight you can feel on a slow connectionThe page stalls or reflows while it loads
What breaks if you remove itNothing you could not rebuild in the themeOne workflow goes manualYou do not know, and will not find out

Monthly. The right-hand column is a decision postponed, not a tool used.

An offer taken apart into the four things it is made of, drawn as four equal quarters of one bar. Promise answers what am I actually getting, and is about what changes for the buyer rather than what the thing is. Proof answers why would I believe you, and has to be something the buyer can check without trusting you. Price answers is this worth it to me, and is the only quarter most people ever touch. Risk answers what if it is wrong, and covers returns, warranty and who pays the postage. A reader who changes only the price has changed one quarter of the offer and left the other three exactly as they were. The quarters are drawn equal and no weighting is claimed.
Promise, proof, price, risk. Four quarters of an offer.
What a stranger checks before entering a card, the signal that answers each doubt, and where on the page that signal lives. Is this a real shop: a contact route with a real address, in the footer of every page. Will it actually arrive: a delivery window in one sentence, on the page above the reviews. Will it look like the photo: customer photographs, with the reviews. Is it worth this: a rating summary with the moderate ones left in, beside the price. What if it is wrong: the return window and who pays postage, on the page rather than only the footer. Am I safe typing a card: the same name and branding as the page before, in the checkout header and on the card statement. Below, four fabrications that reverse all of it, struck out.
What a stranger checks before entering a card.

TAKEAWAYS

  • Choose by job, not by list. Write the job before opening the app store.
  • Email and SMS capture is the exception: uncollected contacts are not recoverable later.
  • Anything that fabricates urgency or proof is a processor risk, not a tactic.
  • Audit monthly. A short app list is a store you can still debug.

CHECK YOURSELF

You read a guide recommending a specific bundle of apps for new stores. What is the best first move?

Two installed apps both modify the cart. What is the most likely symptom?

DO THIS BEFORE THE NEXT MODULE

  1. 01List every installed app and write, beside each, the job it does in one sentence.
  2. 02Load your own product page on a mid-range phone, on mobile data, cache cleared.
  3. 03Put the conversion date of every active free trial into your calendar.
Enrol free