First, confirm the sales are really missing
Start with the counting, because it is the cheapest to rule out. Click a link in a seed copy of the email that went out and look at the address that loads: the campaign parameters should arrive intact on the final page. Google's help centre on URL builders describes the parameters a tagged link carries and how they populate campaign data, so a missing or altered value means the visit, and any sale, is reported under another source. Then check whether total orders fell over the same days. If the business sold as much as usual while email-credited orders dropped, the sales moved in the reports, not in the world.
Next, check the rule. Write down the attribution window and model each report uses, and compare them with how long customers normally take to decide. The email metrics and testing guide asks for the window and model to be agreed, printed on the report and kept stable. Finally, count unique clickers against orders rather than total clicks: one keen reader opening every link can make a send look busy without a single extra buyer. If the tags survive, the rule fits and unique clickers truly rose while orders did not, the sales are missing, and the path causes below apply.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| Email orders fell, total orders did not, direct or referral rose | Follow a live email link and read the parameters on the final page | Links untagged or tags stripped, so the sales are credited elsewhere | Tag every link and make redirects pass parameters through |
| Email sales look weak on high-value or considered products only | The report's window and model against the time customers take to buy | An attribution window or model too short for the purchase cycle | Agree a window that fits the cycle and keep it stable |
| Visitors from the email leave the first page quickly | The landing page's headline, product and price against the email's promise | A landing page that does not continue the email's offer | Send each link to a page that shows exactly what the email offered |
| Clicks concentrate on one link, then checkout errors or complaints arrive | The code, price and destination in the email that went out | A code, price or link broken in the email that went out | Fix it at the destination first, then decide on a correction |
| Phone visitors from email convert far below desktop | Load and layout of the email landing pages on a real phone | A mobile landing page that is slow or awkward to use | Lighten the page and rebuild it for one-handed use |
| Clicks come mostly from people who have never bought | Clickers and buyers by lifecycle and engagement segment | Offers sent to subscribers who read but do not buy | Send offers to buyers and near-buyers; give readers content |
Fix, verify and prevent each cause
Tags missing or stripped on the way
An untagged link, or one passed through a shortener or tracking redirect that drops its parameters, lands as direct or referral traffic, and the order goes with it. The naming conventions and UTMs guide warns that shortened links, vanity domains and tracking redirects must pass the parameters through, which is a thing to test rather than assume, and that values are read exactly as written, so mixed case creates separate mediums.
Fix: tag every link with consistent UTMURL tags identifying traffic source. values from one builder, and replace any redirect that loses them.
Verify: a click from a seed copy reaches the page with every parameter intact, and the email channel's sessions and orders rise while direct falls by a similar amount.
Prevent: make "open one link from the seed copy and read the address" a step of every send's final check.
A counting window shorter than the decision
A hamper, a sofa or an annual plan is rarely bought in the same sitting as the click. If the report only credits email for orders within a short window, considered purchases are undercounted; if someone changed the model or the window recently, the drop is an artefact. The metrics guide explains both effects and why the rule must be stable, so this page does not repeat it.
Fix: agree with whoever owns analytics a window that fits the product's usual time to purchase, write it on the report and apply it to every period compared.
Verify: re-reading the earlier weeks under the agreed rule shows whether the drop survives.
Prevent: log every change to attribution settings with a date, so a step in the chart can be matched to a decision before anyone blames the email.
A page that drops the email's promise
The email made an offer: a particular product, a price, a bundle, a free extra. If the link opens the homepage or a general category, the reader has to find the offer again, and many will not. The message match guide explains the mechanism for ads, a person deciding in moments whether the page is the one the click promised, and it holds for email clicks just as well.
Fix: link each offer to the page that shows it, with the email's words and image repeated in the first screen.
Verify: email visitors' bounce falls on the new destinations and their CVRShare of visits or clicks that convert. rises against the same send type before the change.
Prevent: brief the landing page with the email, not after it, and name the destination in the email's copy document.
A code, price or link that fails
A discount code that does not apply, a price that differs at checkout, or a link to a sold-out or missing product turns a ready buyer into a complaint. The tutorial on fixing a mistake in a sent email shows that most fixes belong at the destination: redirect the link, correct the page, make the code work on the advertised terms where the business decides to honour them.
Fix: repair the destination first, then decide whether readers need a correction at all.
Verify: follow the email's own link from a seed address and complete a test order with the code.
Prevent: test every code and every link from the final proof, and check that codes outlive any flow that carries them.
A mobile page that fights the reader
Most email is read on a phone, so most clicks arrive on one. A heavy page, a layout that shifts as it loads, a buy button out of thumb reach or a checkout that needs zooming loses the reader who was ready a moment earlier. The speed and mobile guide sets out the rules: the primary action reachable, large and separated tap targets, a stable layout and a first screen that holds the offer and the request.
Fix: measure the email landing pages on a real phone and on the mobile connection most readers use, remove what slows them and rebuild the first screen for one hand.
Verify: the gap between phone and desktop conversion for email visitors narrows.
Prevent: check every new email landing page on a phone before the send is scheduled.
Offers sent to people who read but never buy
A list holds readers who enjoy the emails and buyers who act on them, and the two groups overlap less than reports suggest. A strong CTRShare of impressions that result in a click. from people who have never ordered can lift engagement with no effect on sales. The segmentation guide places each subscriber on the path from first contact to repeat customer, with different messages for each stage.
Fix: send offers to buyers and recent browsers, and give long-standing readers content that earns their first purchase rather than another discount.
Verify: clicks per send may fall while orders per send hold or rise.
Prevent: report clicks and orders by lifecycle segment as standard, so a rise in curiosity is never mistaken for a rise in demand.
A prevention routine from click to checkout
Put the whole route on one page and check it for every send. Before scheduling, open a seed copy on a phone, follow every link to its final address, confirm the campaign parameters are intact, confirm the page shows the offer the email describes, and complete a test purchase with any code the email carries. These checks take minutes and catch four of the six causes before a single subscriber sees them.
After each send, read the figures in a fixed order: unique clickers, email visitors in analytics, the conversion rate of those visitors by device, and orders under the agreed attribution rule, all beside total orders for the business over the same days. A gap between unique clickers and analytics visits points at tagging; a gap between visits and orders points at the page, the offer or the device; a fall in email orders with total orders unchanged points back at counting. Keeping this order turns a vague complaint into a stage within a day.
Monthly, look at who clicks. If the share of clicks from people who have never bought keeps rising, the programme is serving readers well and buyers less, which is a segmentation decision rather than a copy problem. Keep a dated log of changes to tagging, redirects, attribution settings, landing templates and segments, because every cause on this page can appear the day something in that list changes. The email troubleshooting section lists the neighbouring symptoms, such as a sudden fall in opens, that sometimes arrive at the same time.
Fudgewhisk Gourmet Hampers follows its clicks to the checkout
Fudgewhisk Gourmet Hampers, an invented hamper business, sees its weekly newsletter hold at about 2,300 unique clickers through October while email-credited orders fall from 118 to 41. Total orders for the business fall only from 690 to 662, and direct traffic rises by roughly 1,900 sessions a week. Following a link from a seed copy shows the reason: a new link shortener added in late September drops every campaign parameter. Restoring direct, tagged links brings email-credited orders back to 97 the next week. The remaining gap sits on the page. The autumn hamper email linked to the homepage, where the featured hamper was not shown; email visitors there converted at 1.1% against 3.4% for a send that linked to the hamper itself. Pointing each offer at its own page lifts email visitors' conversion to 3.0% over the next three sends, and the team adds a seed-copy link check to the send checklist.
Common mistakes
Rewriting subject lines when the clicks are fine. Blaming the offer before reading the link's final address. Changing the attribution window and calling the result a recovery. Sending every link to the homepage. Adding a discount to fix a broken code.
