Newsletter Ad Make-Goods: What You Owe When It Fails

A make-good is what a publisher owes an advertiser when a placement failed to deliver what was sold. Whether you owe one depends entirely on what you guaranteed. If you sold a flat placement with no stated delivery, you owe a make-good only when the ad did not run as agreed. If you guaranteed opens or clicks, you owe the shortfall, in the same unit you sold, once it passes whatever tolerance the contract allows.

Worked through: a $1,200 placement that guaranteed 16,000 opens and delivered 13,600 is 15% short, which is worth $180. It is not worth a free re-run of the whole issue, which is what most publishers offer by reflex and which in this example hands over about $1,020 of inventory to settle a $180 debt.

That gap between what is owed and what gets given is the subject of this post. So are the two guarantees that have become hard to honour in email, opens after Apple Mail Privacy Protection and clicks after corporate link scanners, and a clause you can put in your next insertion order so the question is settled before anything goes wrong.

What is a make-good in newsletter advertising?

The term comes from broadcast. A television network that sold a campaign against a ratings guarantee and fell short would run additional spots, free, until the audience delivered matched what was bought. Those extra spots made good on the guarantee. The cash never changed hands; the inventory did.

Newsletter advertising inherited the word but not the discipline. Most newsletter sponsorships are sold on a flat fee with an implied audience ("40,000 subscribers, 40% open rate") and no written guarantee. When something goes wrong, the publisher and advertiser negotiate from scratch, usually by email, usually after the advertiser has already decided they are unhappy. The outcome depends more on the relationship than on the contract, because there often is no contract worth the name.

That works while volumes are small. It stops working the first time a larger buyer, or an agency buying on their behalf, sends an insertion order with standard terms attached, because those terms have a view on make-goods and it may not be yours.

The guarantee decides whether you owe anything

Before discussing what a make-good should look like, settle what was promised. There are five common bases for a newsletter placement, and they carry very different obligations.

What was sold              Make-good owed when...
─────────────────────────  ──────────────────────────────────────
Flat fee, nothing stated   The ad did not run as agreed
Sends / list size          Fewer emails sent than stated
Opens / impressions        Opens delivered below guarantee
Clicks                     Clicks delivered below guarantee
Outcomes (sales, signups)  Rarely a make-good; usually CPA pricing

The further down the table you go, the more of the result depends on things you do not control. You control whether the email goes out. You partly control how many people open it. You barely control how many click, because that depends on the advertiser's creative and offer. You do not control whether they buy.

This is why the flat fee with a stated audience remains the most common newsletter deal, and why negotiation advice tends to warn publishers away from performance guarantees. A guarantee lets you charge more. It also converts every factor you do not control into a liability you carry.

Five ways a placement fails, and which ones earn a make-good

Most disputes start because the two sides are describing different failures. It helps to name them.

1. It did not run, or ran on the wrong date

The clearest case. The issue was skipped, the ad was left out, or it ran a week late after the advertiser's launch. You owe a make-good whatever the contract says about guarantees, because the thing sold did not happen. If the date mattered (a launch, an event, a sale ending), a later placement may be worth nothing to them, and a refund is the honest answer.

2. It ran, but broken

A dead link, a missing image, the previous advertiser's creative, or a promo code that did not work. This is the most expensive failure per incident because the clicks it cost cannot be recovered: the readers who clicked and hit a 404 are not coming back for a second look. If the break is on your side, you owe. If the advertiser supplied the broken URL or code and you ran it as supplied, you do not, though pointing that out is rarely the best way to keep the account.

3. It ran to fewer inboxes than stated

The list is smaller than the media kit said, or a deliverability incident sent a large share of the issue to spam. Gmail's bulk-sender rules set a spam complaint threshold of 0.3%, with 0.1% as the level to stay under, and a sender who crosses it can see placement collapse for days. The advertiser bought an audience and received part of one. Whether that is a make-good depends on whether you stated the audience. The broader cost of this failure is covered in why deliverability is an ad revenue problem.

4. It ran fine, but missed a performance guarantee

You guaranteed 16,000 opens or 500 clicks and did not reach it. If the guarantee is written, the make-good is mechanical, and the next section shows the arithmetic. If it was only implied ("our readers usually open at 40%"), you are in a negotiation, and your position is weaker than you would like.

5. It ran fine and the advertiser is disappointed

Delivery matched what was sold, and the advertiser got fewer signups than they hoped. This is not a make-good. It is a campaign that did not work for them, and the conversation it calls for is about creative, offer and fit, not compensation. Treating it as a make-good teaches the advertiser that your guarantees are negotiable after the fact. Treating it as a performance review, with their data on the table, is how a disappointed first campaign sometimes becomes a renewal. The renewal economics post covers that conversation.

How to calculate a make-good

The formula is short. Shortfall as a percentage of the guarantee, multiplied by the value of the placement, delivered in the same unit you sold.

Placement fee                   $1,200
List size                       40,000
Guaranteed opens (40%)          16,000     → effective $75 CPM
Delivered opens (34%)           13,600

Shortfall                        2,400 opens  (15.0% of guarantee)
Make-good value                   $180        (2,400 × $75 / 1,000)

Two refinements matter in practice.

Tolerance. Delivery always varies a little, and most contracts that include a guarantee also include a band inside which no make-good is owed. The question is what happens when the band is exceeded. Under a threshold reading, once the shortfall passes 10% you owe all of it: $180 here. Under a deductible reading, you owe only the part beyond 10%: 5% of the guarantee, or 800 opens, worth $60. Both are defensible. Pick one and write it down, because a 3x difference in the settlement is a big thing to discover you disagree about after the fact.

The unit. A make-good is owed in the currency of the guarantee. A shortfall in opens is made good in opens; a shortfall in clicks is made good in clicks, or in the opens needed to produce them at the placement's click-through rate. Converting to a cash refund is a fallback when neither side wants more inventory, not the default.

Why a free re-run is usually the wrong make-good

The reflex when a placement goes wrong is to offer to run the ad again in the next issue. It feels generous, it is easy to say, and it ends the uncomfortable email. It is also usually the wrong size.

In the example above, a full re-run delivers roughly another 13,600 opens. At the effective $75 CPM of the original deal, that is about $1,020 of inventory given to settle a $180 shortfall, roughly 5.7 times what was owed. If that slot would otherwise have gone unsold, the cost is small. In a sold-out month it is a real placement you are giving away, and in the fourth quarter, when newsletter inventory is at its tightest, it can mean turning away a paying advertiser to honour a make-good that did not need to be that large. The Q4 pricing post covers why those slots are worth more than your rate card says.

Better-sized options, roughly in order of preference:

1. Deliver exactly the shortfall. If your ad stack can cap a placement at a number of impressions, run the make-good as a line item with an impression goal of 2,400 and let it stop when it reaches it. This is the broadcast model applied properly.
2. A secondary slot. A lower-position or text-only placement in the next issue, priced on your card at roughly the owed value.
3. A credit on the next booking. $180 off their next placement. This has the advantage of pointing the conversation towards a next booking.
4. A cash refund. Clean and final. Appropriate when the date mattered and later inventory is useless to them.

The full re-run still has a place: when the failure was total (the ad never ran), when it was broken in a way that cost all the clicks, or when the relationship is worth the overpayment. Just make it a decision rather than a reflex.

Open guarantees after Apple Mail Privacy Protection

The largest single change to newsletter make-goods in recent years came from a feature that has nothing to do with advertising. Apple Mail Privacy Protection, introduced in iOS 15 in 2021, has Apple's servers fetch the remote images in an email when it is delivered, before anyone opens it. Tracking pixels fire. Opens are recorded. No human has looked at anything.

For an open guarantee this cuts both ways. A list with a large share of Apple Mail users will report open rates well above the rate at which people actually read the email, so the guarantee is easy to meet on paper. The advertiser, who sees sessions and signups rather than opens, may conclude the audience is much less engaged than the open count suggested. A guarantee met with machine opens is a guarantee met in a way that damages the renewal. The Gmail and Apple MPP tracking guide covers the mechanics in detail, and the glossary entry on Apple MPP has the short version.

The fix is in the contract, not the measurement. An open guarantee written today should say one of three things:

• MPP-attributed opens count toward the guarantee (easiest for the publisher, weakest for the advertiser);
• MPP-attributed opens are excluded, and the guarantee is set lower to reflect that;
• Both figures are reported, and the guarantee applies to one named figure.

The third is the one I would sign on either side, because it keeps both numbers visible. Which leads to a decision I had to make in the MailAdx ad server.

How we decided to count Apple prefetch

MailAdx serves ads at open time: the ad in the email is an image URL, and the ad server decides what to return when the image is requested. With Apple MPP, that request comes from Apple's infrastructure at delivery, so the ad is chosen and fetched whether or not the subscriber ever opens the email. We had to decide whether that fetch is a billable impression.

There were three realistic options. Refuse to bill it, which under-counts every human who does later open the email in Apple Mail, since the image is served from Apple's cache and the ad server never sees a second request. Bill it silently, which over-counts and hides the problem. Or bill it and label it.

We chose the third. An Apple prefetch is recorded as a valid impression, because an ad slot was rendered and the ad served is the one any later human view will display. But the event carries a distinct device label, AppleMPP, so it can be separated in reporting. We also deliberately do not treat the prefetch as proof of a human open when judging whether a later click is genuine, because a prefetch at delivery says nothing about whether anyone read the email.

The consequence for make-goods is that a publisher can hand an advertiser both numbers, total impressions and impressions excluding Apple prefetch, from the same report, and the contract can name which one the guarantee applies to. That is what the reporting is for. The same logic appears in our help centre answer on why ad server numbers differ from ESP open rates: compare like with like, and label the machines.

Click guarantees and the scanner problem

Click guarantees are popular with advertisers and increasingly risky for publishers, for a reason specific to email.

Many corporate mail systems run security scanners that follow every link in an incoming message to check it before delivery. Each of those checks registers as a click. On a list with a meaningful share of work addresses, that machine traffic is significant: our analysis of bot and scanner clicks on newsletter ads put the inflation somewhere between 10% and 35% of recorded click-through, depending on the audience.

Now apply that to a guarantee. Suppose you promised 500 clicks and your ESP recorded 560.

Guaranteed clicks                 500
Recorded clicks                   560    → guarantee met on the raw count
Scanner share (assumed 20%)       112
Human clicks                      448    → 10.4% short on the filtered count

You have met the guarantee on the raw count and missed it on the filtered one. The advertiser's analytics, which record sessions rather than clicks, will show something close to 448 at best, fewer if some visitors bounce before the analytics tag loads. Report the raw number and you are claiming a delivery their data contradicts. Report the filtered one and you owe a make-good on a campaign that, by your ESP's count, succeeded.

The only stable position is to guarantee against filtered clicks from the start, price accordingly, and report both figures. In MailAdx, clicks pass through the ad server's redirect, where each click is classified against the invalid traffic policy and flagged clicks are recorded with their category rather than dropped, so the gap between raw and valid is visible to both sides. If a click guarantee is non-negotiable for the advertiser, the creative guidance on writing ads that get clicks is worth sending them before the campaign, since the creative drives the click rate far more than you do.

Whose count settles it?

Every make-good dispute eventually becomes a dispute about measurement. Your ESP says one thing, your ad server another, the advertiser's analytics a third. They will never agree exactly, and a contract that does not say which one wins leaves you arguing about the ruler instead of the shortfall.

Display advertising solved this years ago, and the solution transfers well. The IAB/4A's Standard Terms and Conditions for Internet Advertising (version 3.0), which most US display insertion orders descend from, names one count as the controlling measurement. If another party's count differs from it by more than 10% over the invoice period and the controlling measurement is lower, the parties reconcile. A shortfall that survives reconciliation is treated as under-delivery, and the parties either agree a make-good flight or settle on the controlling count with a 10% upward adjustment (IAB/4A's Standard Terms and Conditions 3.0).

For a newsletter, three things carry over directly: name the controlling measurement in writing, set a discrepancy threshold, and agree what happens above it. For most newsletter deals the sensible controlling measurement is the ad server or ESP count, with the advertiser entitled to see the underlying report. If you want a template for what that report should contain, the free sponsor delivery report tool produces one, and newsletter ad reporting beyond opens and clicks covers what advertisers look for in it.

A make-good clause you can use

Adapt this to your own deal. It is written for a flat-fee placement with an open guarantee; swap the unit for clicks or sends as needed. It is not legal advice, and if a deal is large enough to matter, have it reviewed.

DELIVERY AND MAKE-GOODS

1. Guarantee. Publisher guarantees [16,000] ad impressions
   in the [date] issue, excluding impressions attributed to
   Apple Mail Privacy Protection prefetch.

2. Measurement. Publisher's ad server report is the controlling
   measurement. Advertiser may request the underlying report.

3. Tolerance. No make-good is owed if delivery is within 10%
   of the guarantee. Beyond 10%, Publisher owes the full
   shortfall.

4. Form. Make-goods are delivered as additional impressions
   within 60 days, capped at the shortfall. If Advertiser
   declines, Publisher refunds the pro-rata value in cash.

5. Cap. Total make-goods never exceed the placement fee.

6. Exclusions. No make-good is owed for shortfalls caused by
   late or non-compliant creative, broken advertiser URLs or
   codes, or the performance of Advertiser's own site.

Point 5 is the one publishers most often forget. Without a cap, a series of disputes can in principle leave you owing more than you were paid. Point 6 protects you from the most common unfair claim: a campaign that under-performed because of the advertiser's own landing page. If you want the rest of the insertion order in similar shape, the direct-sold sponsorship guide covers the surrounding terms.

Fixing a broken placement after the send

One category of failure deserves a separate note, because it is the one where the tooling decides the outcome.

When a static ad is baked into an email at send time, a broken link or wrong creative is permanent. The email is in 40,000 inboxes, and every reader who opens it tomorrow, next week or next month sees the mistake. The only remedy is compensation.

With open-time serving, the ad in the email is a reference rather than a copy. MailAdx decides what image to return when the email is opened, and resolves where a click goes at the moment of the click, from the creative's registered destination. So if an advertiser's URL is wrong, fixing the creative fixes every click from that point on, including in emails already delivered. The same goes for a wrong or expired creative. Readers who opened before the fix still saw the broken version, so a make-good may still be owed, but it is owed for hours of exposure rather than for the life of the email. The open-time serving explainer covers the mechanism, and the ad server page the product.

The same stack makes the precisely sized make-good possible. A direct-sold line item in MailAdx can carry an impression goal, so a 2,400-impression make-good runs until it has delivered 2,400 and stops, instead of occupying a whole issue. For publishers running their own sponsors, that is what the publisher tools are built around.

Make-goods in a network

If you sell across several newsletters, as a network or an agency, the make-good problem compounds. The advertiser's contract is with you, not with the individual publisher whose send went to spam, so you owe the make-good and then have to decide whether to recover it. That needs its own clause in your publisher agreement, and the allowance for it belongs in your margin model. We treat it as a line in the network economics in the take-rate math of running a newsletter ad network.

Frequently asked questions

What is a make-good in advertising?

A make-good is compensation owed to an advertiser when a placement fails to deliver what was sold. The term comes from broadcast television, where networks that missed a ratings guarantee ran additional spots free until the promised audience was delivered. In newsletter advertising it usually takes the form of additional impressions, a secondary placement, a credit on a future booking, or a refund.

Do I owe a make-good if my newsletter open rate drops?

Only if you guaranteed opens. On a flat-fee placement with no stated delivery, a lower open rate is not a make-good obligation, though it may still be worth addressing to keep the advertiser. If you did guarantee opens, calculate the shortfall against the guarantee and compare it to any tolerance in the contract.

How do you calculate a make-good for a newsletter ad?

Divide the shortfall by the guarantee and multiply by the placement fee. A $1,200 placement guaranteeing 16,000 opens that delivered 13,600 is 2,400 opens, or 15%, short, which is worth $180. If the contract has a 10% tolerance read as a deductible, only the portion beyond 10% is owed, which is $60 in the same example.

Is a free re-run a fair make-good?

It is fair to the advertiser but often oversized. In the example above, a full re-run delivers around 13,600 opens, worth about $1,020 at the deal's $75 effective CPM, to settle a $180 shortfall. It makes sense when the ad never ran or was broken throughout. Otherwise, deliver the shortfall exactly, offer a secondary slot, or credit the next booking.

Should Apple MPP opens count toward an open guarantee?

The contract should say explicitly. Apple Mail Privacy Protection fetches email images at delivery, so it records opens that may not reflect a human reading the email. The safer approach is to report both total and MPP-excluded figures and name the one the guarantee applies to.

Why are click guarantees risky for newsletter publishers?

Because corporate security scanners click links to check them, inflating recorded clicks by an estimated 10% to 35% on lists with many work addresses. A guarantee met on raw clicks may be missed once scanner clicks are filtered, and the advertiser's analytics will reflect the filtered number. Guarantee against filtered clicks, price for it, and report both.

Can an advertiser claim a make-good because the campaign did not convert?

Not unless you guaranteed conversions, which newsletter publishers rarely should. A placement that delivered what was sold but produced fewer sales than hoped calls for a conversation about creative, offer and audience fit, not compensation. Writing exclusions for advertiser-side causes, such as late creative or a broken landing page, into the contract prevents the dispute.