Before You Set the Discount: Underrated Tactics to Run Before BFCM

By September, most of us have read the same Black Friday article multiple times.

Build the calendar early, warm the list, don't forget SMS. It’s all good advice, but at this point we all know the drill.

The site, support, and lifecycle work done in October is what actually protects margin during BFCM.

It's structural, unglamorous, and easy to defer, because none of it feels urgent until the week it becomes impossible.

So we asked four partners we work with all year (Kustomer, Orita, Digioh, and rediem) for the tactics they watch brands skip.

Bear's Take: The Discount Is the Last Decision You Should Make

Here's the pattern we see every year. A brand discounts deeply, revenue arrives, and the next November it takes a deeper cut to hit the same number.

We've written about this before, and the thesis still holds: price and promotions are levers to pull, not a strategy to base your brand growth and value on.

The more useful question is what your discount is being asked to cover. Of the shoppers who converted at 40% off, how many landed on a page that gave them no other reason to say yes?

So before you settle on a discount percentage, audit where that traffic is going to land.

Awe Inspired hired us last year for a Conversation Rate Optimization (CRO) audit ahead of BFCM. Once we got under the hood, we found a better use for that time than a round of A/B tests. We rebuilt the site architecture, merchandising, and post-purchase flow into one clear user journey, and it returned a 23% YoY conversion rate lift during BFCM.

How to pull this off: take your five highest-traffic landing pages and your ten best-selling product pages (PDPs), and ask three things of each:

1. Can a shopper tell what else is available from here?

2. Can they tell why this costs what it costs?

3. Can they predict where a click will take them?

Every "no" is a place your discount is about to do work it shouldn't have to.

Kustomer: Narrow What Your AI Is Allowed to Say

Anna Fisher, Chief Marketing Officer, Kustomer

Support is where a shaky BFCM shows up first, usually about six hours before anyone in marketing notices. Anna's answers are about the 30-day fixes for that: what your AI agent is allowed to say, and what October's ticket queue is already telling you.

What's the single change a brand should make to its help center AI agent in the next 30 days that pays off on Cyber Monday?

Instead of training your AI agent on your entire knowledge base, scope it down to one locked, finalized section per topic instead. Most brands treat "get the AI ready" as a content problem: write more articles, cover more edge cases. But the change that actually pays off is narrower than that:

1. Take your highest-volume topic (likely “Where is my Order” (WISMO) or returns)

2. Finalize the policy that governs it

3. Train the agent on only that section

Give it one area it can answer with total confidence, instead of broad access to a knowledge base that's still shifting week to week through Q4.

Just as important, build an exit ramp. The moment a conversation touches something outside that locked scope (a damaged item, a missing receipt, a policy question it wasn't trained on) it should hand off to a human immediately instead of guessing.

What WISMO deflection setup actually works, and how do you know by October?

The setups that hold up share a specific shape:

They trigger off message content, matching on phrases like "where is my order," "tracking," or "when will it arrive”

They have a second condition that suppresses the reply once a conversation has already been auto-answered, so customers don't get looped into the same canned response twice.

The part teams underestimate is that the auto-reply is only as good as what's behind it. Without a live feed of tracking or order status, that trigger just prompts the customer for an order number and hands them right back into the queue, frustrated. And the keyword list isn't a one-time setup. Customers stop saying "where is my order" and start saying "my package is stuck" the moment the first version stops landing, so it needs to be refreshed against real conversation data, not written once in September and left alone.

By October, know your actual deflection rate: conversations that specific workflow resolved and marked done, against total WISMO volume.

And, check customer satisfaction (CSAT) on that deflected set specifically, not your team-wide score, because a conversation that closed automatically but left someone annoyed isn't a win.

What should you pull from October's support tickets before the sale starts?

Support conversations in October are the most honest research you'll get all year, because they capture what customers are actually struggling with, which often stands in stark contrast to your team’s pre-conceived ideas of what to prioritize.

1. Start with what's slipping past whatever automation or self-service you already have: Every WISMO or return question your current setup fails to catch is a customer using language you didn't anticipate, and that gap is exactly what should expand your FAQ and keyword coverage before volume triples, not after.

2. Then look at the product level: Pull interactions where the same sizing, fit, or "this isn't what I expected" complaint shows up more than once. That's a signal to fix the product page copy now, not to add another FAQ entry once the sale has already started.

3. Apply the same lens to shipping: If customers keep asking something your policy page technically already answers, the answer is in the wrong place or the wrong words.

Treat every recurring theme, in tickets your team handled and tickets your automation missed, as a vote for what to change before the sale starts.

Orita: Know Who Wants to Hear From You Before You Send

Aaron Schwartz, Co-Founder and Co-CEO, Orita

The biggest senders are mailing fewer people than they used to and reaching more of them. Aaron's answers are about the three decisions behind that: who's worth waking up, where your SMS budget is leaking, and how far you can reach into your list without going in cold.

How should you test which win-back window actually predicts a comeback?

A flat rule like "90 days quiet, send" treats every dormant customer the same, and that's the problem: some people are ready to come back in 40 days, others take a year, and a single window misses both.

Before locking in a win-back flow for BFCM, test multiple windows against each other instead of guessing one.

1. Run small slices of your list across different quiet-periods

2. Watch which ones actually convert, and shift volume toward the windows that show real lift.

3. Continue sending small amounts to the lower-performing tiers, so you keep getting data.

Do this early, not the week before Black Friday. You want the flow tuned before your highest-stakes send, not learning in real time during it.

How should you test which SMS format earns its keep by segment?

RCS and emojis cost more per message, so "does it look nicer" isn't the question. "Does it convert enough to justify the cost" is.

Test it the same way you'd test any spend:

1. Pick a few engagement tiers

2. Send the richer version to some and plain text to others

3. Compare revenue per recipient, not just click rate.

You'll likely find RCS earns its cost with your most engaged customers and doesn't with the rest. Use it selectively instead of applying it everywhere by default.

The same logic applies to SMS credits generally: full-list texts that include your least-engaged segment burn budget on the group least likely to convert, while starving your best customers of the volume that would actually pay off.

What should you watch before emailing people you don't normally message?

Before you decide to email people you don't normally message, know two things:

1. Who's a past purchaser

2. Who's a cold prospect

Purchasers are a lower-risk send, they've already converted once, and even a purchaser who's gone quiet deserves at least one touch during peak week.

Prospects are a different, more selective call, lean on engagement data to decide who's worth it rather than blanket-including the whole list.

Then, when you do reach further than usual, don't send cold.

1. Send to your most-engaged segment first.

2. Then, send the same campaign to everyone else roughly 60 to 90 minutes later.

Inbox providers weight early engagement heavily when deciding how to treat the rest of a send, so that first wave earns a better reception for the riskier one behind it. Too short a gap and the signal hasn't registered yet. Too long and you've lost the moment that justified reaching that far in the first place.

Digioh: November Traffic Is the Biggest Data Window You Get All Year

Matt Teed-Arthur, Head of Partnerships, Digioh

Every brand gets a traffic spike in November, and most of them spend it collecting email addresses. Matt's answers are about what else you could be capturing while those people are already on the page, and how that changes what you're able to do in January.

What should a brand be capturing on-site in October & November that changes what they're able to send in Q1?

In this season you should be trying to learn three things:

1. Gift-givers vs. Self-shoppers

2. What their style or product match looks like

3. Whether you can push them toward loyalty.

A short welcome quiz pop-up or product recommendation journey does this really effectively.

Keep it to a quick three-tap flow, triggered on-site via UTM, exit intent, or a session timer, and run it through October and November.

Ask three things:

1. Gift or for you

2. Pick your style or category

3. A quick-click loyalty enrollment.

Then drop them on the category page right away. Three taps max, or you risk losing them.

The important part is that it writes to the profile record at capture rather than sitting as an anonymous form fill. That's the ID layer doing its job, making the data usable downstream in ad platforms and your ESP or SMS tool.

Now you have a flywheel. Gift-givers get retargeted with occasion-based creative — new arrivals, "still shopping for someone?", "how'd they like the gift?", or a follow-up next gifting season. Self-shoppers get restock and loyalty-tier creative, or complementary add-ons.

Better segmentation, better personalization, one campaign.

What does a good preference center look like going into BFCM?

Treat your preference center as a BFCM prep and targeting tactic, and it does far more than defend against unsubscribes.

Aaron's answer on engagement-based sending is a strong defensive play — it drives your highest-intent traffic at the right times. The complementary proactive play is giving people a way to state their preferences for more contextual targeting in the first place.

A good preference center, or zero-party data capture flow, going into BFCM asks three things:

1. Frequency tolerance ("how often do you want to hear from us during sale season")

2. Their core interest (advance access, discounts, back in stock)

3. Category interest

Surface it right after or during opt-in for the highest impact, and a pre-BFCM prompt to confirm can work wonders.

Combining both signals makes your suppression logic more specific and protects deliverability with real data instead of inferences off open rates. That's inbox placement and personal communication, which is the real margin protection that should be talked about more.

What capture or quiz flow have you seen convert during a heavy discount period without giving away margin?

Product-match quizzes beat deeper discounts. The flows that convert during a heavy discount window without eating margin are the ones that upsell through tailored recommendations.

A "build your bundle" or "match your style" quiz on a landing page or PDP, where completing it unlocks a smaller incentive like free shipping, a bonus item, or loyalty points, works better than stacking on more discount codes.

Two things happen.

1. AOV goes up, because the quiz steers toward bundles and add-ons instead of a single discounted SKU.

2. You capture zero-party data — skin type, gift occasion, budget tier, whatever's relevant — at the exact moment intent is highest.

That data is what lets you remarket more effectively in Q1 without another discount, because you can put their purchase and preferences into better context. Adding a subscription or loyalty-incentive layer takes that even further with an actionable LTV motion.

rediem: Give the November Cohort a Reason to Come Back in January

Steve Yang, Head of Growth, rediem

The BFCM cohort is the one brands complain about in February. They bought once, at the deepest discount of the year, and then went quiet. Steve's answers are about the setup work that changes that outcome.

What should you set up in your loyalty program before Black Friday?

There's a behavioral economics concept called endowed progress: people who start partway toward a goal finish at much higher rates than people starting from zero. In the classic car wash study, customers given a blank 8-punch card finished at 19%. Customers given a 10-punch card with two already stamped finished at 34%. Identical effort, nearly double the completion.

Before Black Friday, set your loyalty program up so a first engagement lands the customer visibly close to their first reward, and make that reward redeemable with a purchase in Q1. The shopper who came for the discount now has an almost-finished goal sitting in their account.

Another tip if you can pull off a quick turnaround: host challenges in the weeks leading up to BFCM so customers earn status and perks they can redeem during the sale itself (like a members-only drop, first pick of limited inventory). This runs on effort justification — we value what we worked for, even when the work was 30 seconds of snapping a photo.

Does VIP early access hold up against a competitor's 30% off sale?

Early access works in exactly one situation: your product actually sells out, and your customers know it. Timing becomes valuable when the popular size is gone by 9am. If you're not Supreme, early access usually falls flat.

Stacking steeper discounts to manufacture excitement is the other trap. That's a slippery slope with a margin problem at the bottom, and every year the number it takes to feel exciting gets bigger.

Instead of eroding margin, get weird.

1. A colorway only members can buy.

2. A vote on what you make next.

3. Points that convert into a donation to a cause your community picked.

And quit sending everyone the same offer. By November, your program has been collecting zero-party data all fall — quiz answers, challenge entries, the flavors people told you they love. Use it to build cohorts and design personalized incentives. A generic sitewide promo hands your most loyal customer the same deal as a stranger.

A smarter strategy unlocks something a deeper discount never will: a reason for your community to belong.

What loyalty metric tells you whether your BFCM cohort was worth acquiring?

Did BFCM buy you an audience, or a community? An audience might listen. A community participates, and participation is measurable long before repeat purchase rate.

The metric we watch is community engagement rate: the share of your BFCM cohort doing anything with your brand besides buying. Reviews, photo challenges, referrals, points put toward a donation. Repeat purchase rate makes you wait until February to learn what participation tells you within days of the first order.

Use actions specific to your brand:

1. "Show us your furry friend with our product"

2. "Tell us how this fits into your daily routine"

3. "Dream up a favorite flavor combo"

Collecting that UGC and those reviews on-site is also critical for SEO, GEO, and AEO.

This kind of non-transactional participation converts. When one of our brands launched a donation incentive, 55% of the members who donated were back on site for a purchase within five weeks.

By mid-December, split the cohort into participants and spectators. Participants are new active members of your brand's ecosystem. Spectators bought because of the discount, and your next mission is to turn them into participants.

What We'd Actually Do With the Next 30 Days

Four different disciplines, one shared idea: BFCM pays out on the work you did before it.

• Audit the store before you audit the offer: Architecture, navigation, and how value reads on the page are all still fixable in September and frozen by mid-November.

• Lock one support topic, completely: Finalize your highest-volume policy, scope your AI agent to that section alone, and write the handoff rule for everything else.

• Tune your win-back windows now: Test a few quiet periods against each other so the flow is calibrated before the highest-stakes send of the year.

• Earn the reach before you use it: Send to your most-engaged segment first and let the rest follow 60 to 90 minutes behind.

• Give the November cohort a head start: Set loyalty up so a first purchase lands them visibly close to a reward they can only claim in Q1.

• Capture more than an email address: Three taps — gift or self, style or category, loyalty opt-in — written to the profile at the moment of capture.

You're still going to run a promotion (everyone is). The work above decides how much that promotion has to carry.

Thanks to Kustomer, Orita, Digioh, and rediem for contributing!

Frequently Asked Questions

How far ahead of Black Friday should you actually audit your landing pages and PDPs?

Finish it in September, not October. Site architecture, navigation, and how well a landing page explains its own value are all still fixable in September and effectively frozen by mid-November, once dev bandwidth and inventory get locked in for the sale. Bear ran exactly this kind of landing page and PDP audit for Awe Inspired ahead of BFCM last year, and it returned a 23% year-over-year conversion lift during the sale itself — proof the audit pays off before the discount ever gets set.

What is the endowed progress effect, and how does it apply to a Black Friday loyalty program?

It's the finding that people who start partway toward a goal finish it at far higher rates than people starting from zero, even when the remaining effort is identical. In the original study, customers given a blank eight-punch loyalty card finished at a 19% rate, while customers given a 10-punch card with two stamps already applied finished at 34%. Applied to BFCM, it means setting up your loyalty program so a first purchase visibly lands a customer close to a reward they can only redeem in Q1, instead of starting their counter at zero.

If you can only fix one thing before BFCM, what should you prioritize?

Audit the store before you audit the offer. Site architecture, navigation, and how clearly a landing page or PDP explains its value are the pieces that go stationary once November hits, while support ticket triage, email and SMS testing, and loyalty setup can still flex through most of October. An AI agent's scope, a win-back window, or an on-site quiz is far easier to adjust after Black Friday than a page that's already live and pulling in paid traffic.

Do you need a big dev lift to run an on-site quiz for zero-party data before BFCM?

No — the version worth building is deliberately small. A three-tap flow (gift or self, style or category, a loyalty opt-in) triggered by a UTM, exit intent, or a session timer, dropping shoppers onto a category page right after. The only requirement that actually matters technically is that each answer writes straight to the customer's profile record instead of sitting as an anonymous form fill, so it's usable downstream in your ad platforms and your ESP or SMS tool. Add a fourth tap and completion starts to drop.

What's the real risk of scoping your support AI down to one policy instead of your whole knowledge base?

The bigger risk actually runs the other direction. An AI agent trained broadly across a knowledge base that's still shifting through Q4 will answer confidently on things it shouldn't, because nothing tells it where its own knowledge gets shaky. Locking it to one finalized policy — WISMO or returns, whichever gets the most volume — with a hard handoff to a human the moment a conversation drifts outside that scope heads off that failure mode. That's the real payoff of going narrow: the agent always knows exactly where its edge is.