Chapter 12: Digital, Social and Search Marketing

Jess Patel, Lumen Cycles' digital and content manager, tries an experiment before a planning meeting. She opens an AI assistant and types the question she imagines a parent in Chicago asking: "What's the best electric cargo bike for carrying two kids?" The answer is fluent and confident. It names several brands, explains what to look for in a family cargo bike, and mentions battery safety. Lumen isn't in it.

A few years ago, Jess's job was mostly about ranking well on a page of search results and getting attention in social feeds. Those jobs haven't gone away. But some customers now get their first shortlist of brands from an AI-written answer, before they ever see a list of links or an ad. If Lumen isn't part of that answer, the best ad campaign in the world may never get the chance to persuade them.

This chapter covers the channels where most marketing attention, and a large share of most marketing budgets, now goes: websites, email, mobile, social media, content, influencers and search. It builds on the strategy from Chapter 11. Integrated Marketing Communications, where these channels were one part of the promotion mix. Here you'll learn how each one works, how to plan social and content strategy, how search is changing as AI answers questions directly, and how to measure what digital marketing actually achieves.

12.1 Digital and direct marketing

Most of the promotion tools in the previous chapter talk to many people at once. Direct and digital marketing tries to reach specific people, often one at a time, and get a measurable response.

For buyers, direct and digital marketing is convenient: they can research, compare and buy anytime, from anywhere. For sellers, it's targeted, flexible and measurable. A marketer can change an email or a search ad in minutes and see within days whether it worked. That measurability is both the great strength of digital marketing and, as you'll see later in the chapter, a trap.

12.1.1 The website as the hub

For a direct-to-consumer brand, the website is the center of everything. About 55% of Lumen's sales come through its own site, and almost every other channel, from a podcast ad to a dealer conversation, sends people there to learn more.

A company website has to inform (specs, comparisons, answers), persuade (photos, video, reviews, reasons to believe), convert (buy, book a test ride, find a dealer) and serve (order tracking, service, support). Lumen's AI chat assistant, which answers product and service questions and hands warranty claims to a person, is typical of how AI has entered websites: helping visitors find what they need faster.

Good websites are designed around what visitors are trying to do. A parent researching the Cargo at 10 p.m. wants to know whether two kids fit, how far the battery goes, whether it handles hills, what it costs and how to try one. A site organized around the company's org chart, rather than those questions, loses that parent. The share of visitors who complete a desired action is the site's conversion rate, defined formally later in this chapter.

12.1.2 Email marketing

Email is one of the oldest digital channels and still one of the most valuable, for a simple reason: the company owns the list. Social platforms can change their algorithms and search engines can change their rankings, but a customer who has agreed to receive Lumen's emails can be reached directly.

Email works best when it's relevant, and relevance comes from segmentation and timing. Common types include:

• Newsletters and promotions sent to segments of the list, such as Cargo owners, Commute owners and people who have browsed but not bought.
• Triggered emails sent automatically when someone does something: a welcome series after sign-up, a reminder when someone leaves a bike in their online cart, a confirmation and riding tips after purchase.
• Lifecycle emails timed to the customer's relationship: a service reminder before the first tune-up, a Lumen Care renewal notice, an invitation to a local owner ride.

In the United States, commercial email must follow the CAN-SPAM Act, which among other things requires honest sender and subject information and a working way to unsubscribe. Other countries have stricter consent rules. Beyond legal compliance, the practical rule is that every email should be one the recipient is glad to get, because unwanted email leads to unsubscribes and spam complaints that hurt delivery for everyone on the list.

12.1.3 Mobile marketing

Most people now carry their main internet device in a pocket. Mobile marketing reaches them there.

• Mobile-friendly websites are the baseline. A parent who taps a social ad on a phone and lands on a slow, hard-to-read page leaves.
• Apps give a brand a permanent spot on the phone. Lumen's free app tracks rides, battery health and service needs, which makes it useful every day rather than only at purchase time. That usefulness is what earns permission to send the occasional reminder or offer.
• Text messages reach people quickly and are usually read, which is exactly why they must be used sparingly. In the United States, marketing texts generally require the recipient's prior consent.
• Push notifications from an app can deliver timely, useful messages, such as "Your battery health check is due," or annoying ones, such as daily promotions. Annoying ones get the app's notifications turned off, or the app deleted.
• Location-based marketing uses a phone's location, with permission, to deliver relevant messages, such as letting an app user know about a test-ride event at a nearby dealer.

The Lumen app shows the opportunity and the risk in mobile marketing. Lumen is considering using app ride data to personalize offers. A rider who commutes 15 miles a day might be a good prospect for Lumen Care, and a rider whose battery health is declining might welcome a replacement offer. But customers who downloaded the app to track rides may not expect their riding patterns to be used for marketing. That's a question of consent and trust that Chapter 13. Ethics, Responsible AI and Sustainability takes up in depth. For now, the marketing lesson is that data customers share for one purpose shouldn't be quietly reused for another.

12.1.4 Traditional direct marketing

Some older direct channels still work. Direct mail is expensive per contact but tangible, targetable by address and less crowded than inboxes; a postcard inviting households near a new dealer to a Cargo test-ride day is a reasonable use. Telemarketing is restricted (in the United States, the National Do Not Call Registry lets consumers opt out of most sales calls) and is now used mostly with business customers. Direct-response television, which asks viewers to act right away, lives on in streaming ads viewers can act on from a phone.

12.1.5 First-party data

Much of digital marketing used to depend on tracking people across websites and apps with third-party cookies and device identifiers. Over the past several years, web browsers and phone operating systems have restricted that tracking, and privacy laws have tightened. The result is a shift toward data that companies collect directly, with permission.

Lumen is well placed here. Its direct website sales, email list, app and service records give it first-party data that a brand selling only through retailers wouldn't have. The flip side is responsibility: customers trust Lumen with that data, and using it in ways they don't expect, or exposing it to tools that aren't secure, puts that trust at risk. The protect data rule from Chapter 1. Marketing in an AI Era applies to companies, not just students. Customer data doesn't belong in a public AI tool.

12.2 Social media and content marketing

Social media is where people spend a large share of their time online, and it's where brands can both advertise and take part in conversations. But brands are guests there. Social platforms decide who sees what, and their rules change often.

12.2.1 Choosing platforms

A brand doesn't need to be on every platform. It needs to be where its audience is, in a form that fits each platform's culture. The descriptions below are general and change over time, so check current audience data before committing a budget.

Platform type	Examples	Typical strengths	Possible Lumen use
Short-form video	TikTok, Instagram Reels, YouTube Shorts	Discovery by people who don't follow you; entertainment; trends	Quick clips of real school runs and grocery hauls
Long-form video	YouTube	Search-driven; how-to and review content; long life	Cargo walk-throughs, battery safety explainers, maintenance tips
Visual and lifestyle	Instagram, Pinterest	Inspiration and planning; strong visuals	Family ride photos; "car-light family" ideas and routes
Community and conversation	Facebook groups, Reddit	Detailed discussion; local communities; honest opinions	Listening to what parents ask; answering questions openly
Professional	LinkedIn	B2B audiences; company news	The fleet program; hiring; corporate sustainability stories

12.2.2 Organic and paid social

Brands use social media in two ways.

• Organic social is the content a brand posts on its own accounts for free. Its reach depends on the platform's algorithm, which usually favors content that people engage with, and on whether people share it.
• Paid social is advertising on social platforms. It lets a brand target audiences by interests, demographics, location and behavior, and pay to reach people who would never see its organic posts.

Most brands need both: organic content builds community and voice, and paid social reaches new people and drives actions such as test-ride bookings. Organic reach for brand accounts is often limited, so don't assume posting alone will reach the audience.

12.2.3 Content marketing

The idea is old. Michelin started publishing its restaurant and travel guides in 1900 to encourage people to drive more, and so to wear out more tires. Red Bull now runs a media operation that produces sports films and events. The principle is the same: give the audience something it values, and the brand benefits from the attention and trust.

Good content strategy starts with the audience's questions, not the company's messages. For Lumen's Cargo, families are asking things like:

• Can an e-bike really replace a second car for us?
• How do I ride safely with kids?
• What routes work for school drop-off in my city?
• How do I charge and store an e-bike safely in an apartment or garage?
• What does it really cost to own one?

Each question is a content opportunity. A practical way to organize them is content pillars, a few broad themes that all content fits into. Lumen might use four: family life on two wheels (real stories and rides), safety (riding with kids, battery care and certification), practical how-to (charging, storage, routes, maintenance) and the numbers (cost of ownership compared with driving, using clearly stated assumptions).

Another common framework, which YouTube popularized for creators and brands, sorts content into three levels:

• Hero content: occasional big pieces designed to reach a wide audience, such as a short documentary about a family that sold its second car.
• Hub content: regular episodes that give people a reason to come back, such as a monthly "school run" series filmed in different cities.
• Help content: evergreen answers to the questions people search for, such as "how to install a child seat on a cargo bike."

Help content is the least glamorous and often the most valuable, because it keeps working in search results, and in AI answers, long after it's published.

User-generated content. Some of the most persuasive content comes from customers: a parent's video of their kids laughing on the way to school, an owner's review of a winter commute. Brands encourage it with hashtags, owner communities and features of customer stories. Two rules apply. Get permission before reusing a customer's content in marketing, and never fake it.

The editorial calendar lists what will be published, where, when and by whom, and which pillar and objective each piece serves, giving the team (and any AI tools it uses) a clear plan to work from.

12.2.4 Listening and engagement

Social media is two-way. Social listening tracks what people say about the brand, competitors and the category; AI tools sort large volumes of posts by topic and sentiment, as you saw in Chapter 6. Marketing Research and Analytics, but someone still needs to read samples, because sentiment tools misread sarcasm and context. Engagement means responding: a complaint answered quickly and helpfully in public can win more trust than an ad, and a battery-safety question answered with a link to Lumen's certification turns a worry into a reason to believe.

12.2.5 Influencer marketing

Influencers are a message source, in the sense covered in the previous chapter. They can be persuasive because their audiences trust them and see them as people like themselves. Influencers are often grouped by audience size, from nano and micro influencers with small, close-knit followings to macro and mega influencers with very large ones. Exact thresholds vary by source, but the trade-off is consistent: larger influencers offer reach, and smaller ones often offer closer relationships with their audience, more specific niches and lower cost.

For the Cargo, a parent in Minneapolis with a modest following of local families who rides her kids to school year-round may be worth more than a celebrity with millions of followers who has never carried a child on a bike. Fit and credibility matter more than follower counts.

When choosing influencers, marketers check audience fit (the right people in the right places), authenticity (would this person plausibly use the product?), engagement quality (thoughtful comments, not bot-inflated numbers), values and history, and for Lumen, safety (helmets and proper child seats in every video).

Disclosure is required. In the United States, the Federal Trade Commission's guidance on endorsements says that when there's a material connection between a brand and an influencer, such as payment or free products, it must be clearly disclosed. A clear "#ad" or "Paid partnership with Lumen Cycles" label at the start of the post meets the spirit of the rule. A disclosure buried in a string of hashtags doesn't. Many other countries have similar rules. Brands are expected to tell influencers to disclose and to check that they do.

12.2.6 AI influencers and automated content

Two newer developments raise sharper questions.

Virtual influencers are computer-generated characters run as social media personalities. One of the best known, Lil Miquela, appeared on Instagram in 2016 and went on to partner with fashion and lifestyle brands. Generative AI has made such characters much easier and cheaper to create. Their appeal to brands is control: a virtual influencer never goes off-script, never has a scandal from years ago and never ages. Their drawbacks are just as clear:

• Credibility. An influencer's persuasive power comes from the audience's belief that a real person uses and likes the product. A virtual influencer can't ride a Cargo with real kids on a cold morning.
• Deception risk. If followers don't realize the character isn't real, the brand is misleading them about who is speaking. Platforms increasingly require labels on AI-generated content, and the endorsement disclosure rules still apply.
• Fit. Virtual influencers may suit fashion, gaming or entertainment, where audiences enjoy the fiction. They fit poorly where the product's value depends on real-world trust and safety.

Automated content is the broader trend. AI tools can now produce social posts, captions, images, short videos and replies to comments at a volume no human team could match. That helps a small team like Lumen's keep a steady presence, adapt content for different platforms, and test more ideas. It also creates risks:

• Sameness: generic AI content is easy to scroll past.
• Errors at scale: one wrong claim in a template can appear in hundreds of posts.
• Authenticity: content that feels mass-produced weakens the sense of community a brand is trying to build.
• Automated replies: a cheerful automated response to a serious safety complaint can do real damage.

A reasonable approach for most brands is to use AI to assist content creation, with people choosing what to publish and checking every factual claim, and to reserve the most trust-dependent content, such as customer stories, safety information and responses to complaints, for real people. The same disclosure test from the previous chapter applies: would the audience be misled about something that matters if they didn't know AI was involved?

12.3 Search in the AI era

Search is where customers tell you what they want, in their own words. Someone who types "best cargo e-bike for two kids" is far closer to buying than someone scrolling a feed, which is why changes in how search works matter so much.

12.3.1 Search intent

Marketers usually sort searches by the intent behind them:

• Informational: the person wants to learn ("how to ride an e-bike with kids").
• Navigational: the person wants a specific site ("Lumen Cycles dealer locator").
• Commercial investigation: the person is comparing options ("Lumen Cargo vs [competitor]," "best family e-bike").
• Transactional: the person is ready to act ("buy Lumen Cargo," "e-bike test ride near me").

A search strategy maps content to intent (help content for informational searches, comparisons and reviews for commercial investigation, product pages and dealer locators for transactional ones) across the buying journey from Chapter 4. Consumer Behavior.

12.3.2 Search engine optimization

Search engines don't publish exactly how they rank pages, and the details change constantly. The broad principles have been stable for years:

• Technical SEO. The site loads quickly, works well on phones, and can be crawled and understood by search engines. Clear page structure and structured data (code that labels things like product names, prices and reviews in a standard format) help machines understand the content.
• On-page SEO. Each page focuses on a clear topic, uses the words searchers actually use, and answers their questions thoroughly. Titles and headings say what the page is about.
• Off-page SEO. Other reputable sites link to and mention the brand. A battery-safety story in a respected news outlet, which is a PR win, is also an SEO win.
• Content quality. Google's public guidance for evaluating search quality emphasizes content that shows experience, expertise, authoritativeness and trustworthiness, often abbreviated E-E-A-T. A guide to riding with kids written by a Lumen staff member who actually does it, with real photos, is the kind of content those principles reward.

Local search matters especially for Lumen's dealer network. Someone searching "e-bike shop near me" sees local listings and maps. Making sure each of Lumen's roughly 60 dealers shows up correctly in local search, with Lumen named as a brand they carry, helps the push side of the strategy.

SEO is slow. Results build over months, and they're never guaranteed. Its advantage is that a strong position, once earned, keeps sending visitors without paying for each click.

12.3.3 Paid search

In simplified terms, the search engine's auction weighs each advertiser's bid together with the expected quality and relevance of the ad and its landing page, so a highly relevant ad can beat a higher bid for a less relevant one.

Key decisions in paid search include:

• Which keywords. Lumen might bid on its own brand name, on category terms like "cargo e-bike," on problem terms like "bike to carry kids to school," and possibly on competitor names (a common but contentious practice).
• Brand versus non-brand. Bidding on "Lumen Cargo" captures people already looking for Lumen. It's usually cheap and converts well, but many of those buyers might have found Lumen anyway. Non-brand terms like "family e-bike" reach new customers but cost more and convert less. Keep this distinction in mind for the attribution discussion later in the chapter.
• Where and when. Ads can be limited to specific cities (such as the new markets for the Cargo expansion), times of day and devices.
• Landing pages. The ad's job is to get the click. The landing page's job is to convert. Sending a click on "cargo e-bike for two kids" to Lumen's home page instead of the Cargo page wastes it.

Paid search works quickly and can be adjusted daily, so it covers the gaps while organic positions are built.

12.3.4 AI answer engines

The biggest change in search in decades is that search no longer always ends in a list of links. Search engines now often show AI-generated summaries above the results, and many people ask AI assistants directly, the way Jess did in this chapter's opening.

This changes the marketer's problem in several ways.

• Fewer clicks. When the answer appears on the results page, many people don't click through to any website. Traffic from search can fall even when a brand is well regarded.
• Being named matters more than being ranked. In a list of ten links, being fifth still gets some attention. In a generated answer that names three brands, being left out means being invisible for that question.
• The answer may be wrong. AI answers can state outdated prices, wrong specifications or invented claims about a brand, or confuse it with another company. The brand doesn't control the wording.
• Measurement is harder. It's difficult to know how often a brand is mentioned in AI answers, or how many purchases started there.
• AI agents may shop on the customer's behalf. As Chapter 10. Channels, Retail and the Customer Journey discussed, some AI assistants can compare products and even complete purchases, which makes clear, accurate product information even more important.

The practice of improving a brand's presence in AI-generated answers is sometimes called generative engine optimization or answer engine optimization. It's new, and nobody outside the companies that build these systems knows exactly how they choose sources. Be skeptical of anyone who promises guaranteed results. The practices that make sense are mostly extensions of good SEO and good PR:

Practice	Why it helps	Lumen example
Publish clear, factual, well-organized answers to common questions	AI systems draw on content that directly answers questions	A Cargo FAQ: capacity, range, hills, charging, storage, price
Keep facts consistent everywhere	Conflicting information confuses both people and machines	The same specs and prices on the Lumen site, dealer sites and retailer listings
Use structured data	Helps machines identify products, prices and reviews correctly	Product markup on every model page
Earn credible third-party coverage	AI answers often draw on reviews, news and expert sources	Battery-safety coverage; reviews in family and cycling media
Encourage genuine reviews	Reviews are a source of evidence for answers about quality	Post-purchase emails inviting honest reviews
Monitor AI answers regularly	You can't fix what you don't see	Monthly checks of what major AI assistants say about Lumen and the Cargo

There's no shortcut around being genuinely useful and credible: content written only to manipulate machines tends to work poorly and damages the brand when people read it. What stays a human judgment is also clear. An AI tool can draft FAQ content, suggest keywords and summarize what AI assistants say about the brand. People must check that every published fact is accurate, decide which questions matter for the strategy, and decide how to respond when an AI answer gets the brand wrong.

12.4 Measuring digital performance

Digital marketing can measure almost everything. That's its great advantage, and its trap: it's easy to measure what's convenient instead of what matters.

12.4.1 The core metrics

You met CPM, the cost per thousand impressions, in Chapter 11. Integrated Marketing Communications. Digital campaigns add metrics for each step from seeing an ad to buying.

Two related measures are cost per click (CPC), which is ad spend ÷ clicks, and engagement rate on social media, usually interactions (likes, comments, shares, saves) divided by impressions or followers. Platforms define engagement differently, so always check the definition before comparing numbers.

12.4.2 Worked example: social versus search

Lumen ran two one-month campaigns for the Cargo in its new markets: a paid social video campaign and a paid search campaign on non-brand terms such as "cargo e-bike." The figures below are illustrative.

Metric	Paid social	Paid search
Ad spend	$8,000	$6,000
Impressions	400,000	40,000
Clicks	3,200	2,000
Test-ride bookings	120	60
Online purchases	16	20

Now calculate the metrics, step by step.

CPM.

• Social: $8,000 ÷ 400,000 × 1,000 = $20.00
• Search: $6,000 ÷ 40,000 × 1,000 = $150.00

CTR.

• Social: 3,200 ÷ 400,000 = 0.008 = 0.8%
• Search: 2,000 ÷ 40,000 = 0.05 = 5.0%

CPC.

• Social: $8,000 ÷ 3,200 = $2.50
• Search: $6,000 ÷ 2,000 = $3.00

Purchase conversion rate (purchases ÷ clicks).

• Social: 16 ÷ 3,200 = 0.005 = 0.5%
• Search: 20 ÷ 2,000 = 0.01 = 1.0%

CPA for a purchase.

• Social: $8,000 ÷ 16 = $500
• Search: $6,000 ÷ 20 = $300

Cost per test-ride booking.

• Social: $8,000 ÷ 120 = $66.67
• Search: $6,000 ÷ 60 = $100.00

ROAS (each purchase is one Cargo at $2,999).

• Social revenue: 16 × $2,999 = $47,984. ROAS = $47,984 ÷ $8,000 ≈ 6.0
• Search revenue: 20 × $2,999 = $59,980. ROAS = $59,980 ÷ $6,000 ≈ 10.0

Metric	Paid social	Paid search
CPM	$20.00	$150.00
CTR	0.8%	5.0%
CPC	$2.50	$3.00
Purchase conversion rate	0.5%	1.0%
CPA (purchase)	$500	$300
Cost per test-ride booking	$66.67	$100.00
ROAS	≈ 6.0	≈ 10.0

What does this tell Dana? On online purchases, search looks clearly better: a lower CPA and a higher ROAS. That makes sense, because people searching for "cargo e-bike" are already in the market, while social ads interrupt people who weren't looking. But social produced twice as many test-ride bookings at two-thirds the cost per booking. If many of those test riders later buy at a dealer, social may be doing more of the work than the purchase numbers show. The online purchase columns can't see those dealer sales at all.

Is a ROAS of 6 good? ROAS measures revenue, not profit. To judge it, you need the margin. Suppose, for illustration, that Lumen keeps about 40% of the Cargo's price as gross margin after the cost of making and delivering the bike. Then each sale contributes about $2,999 × 0.40 = $1,199.60 toward advertising and other costs.

• Break-even CPA is $1,199.60. A campaign that spends more than that per sale loses money on the first purchase.
• Break-even ROAS is 1 ÷ 0.40 = 2.5. Below 2.5, the ads cost more than the gross margin they bring in.

Both campaigns clear the bar. Social's CPA of $500 is well under $1,199.60, and search's $300 is further under. A rule of thumb follows: always compare ROAS with break-even ROAS (1 ÷ margin), not with 1. A ROAS of 2 sounds like doubling your money, but at a 40% margin, it loses money.

12.4.3 Attribution

Real customers rarely see one ad and buy. A typical Cargo buyer might see a short video on social media, hear a podcast ad, read a review, search for "Lumen Cargo," click a search ad and buy. Which of those deserves the credit?

Common attribution models include:

Model	How credit is assigned	What it favors
Last-click	All credit to the final touchpoint before the conversion	Channels that capture demand, such as brand search
First-click	All credit to the first touchpoint	Channels that create awareness, such as social video
Linear	Equal credit to every touchpoint	Every channel equally, regardless of role
Time-decay	More credit to touchpoints closer to the conversion	Late-stage channels
Data-driven	Credit estimated by a statistical model from many journeys	Whatever the model finds, which depends on the data it can see

Worked example. One buyer's journey has four touchpoints before a $2,999 online purchase: a social video ad, a podcast ad, an email, and a click on a brand search ad.

• Last-click: brand search gets all $2,999.
• First-click: social video gets all $2,999.
• Linear: each touchpoint gets $2,999 ÷ 4 = $749.75.

The same sale makes very different channels look like heroes. Under last-click, a marketer might cut social and podcast spending because they "don't convert," then watch brand searches fall a few months later because fewer people had heard of Lumen.

The deeper problem is that attribution models show correlation, not cause. A brand search ad gets clicked by someone who would probably have bought anyway. The question that matters is incrementality: how many sales happened because of the campaign that wouldn't have happened otherwise? The most reliable way to answer it is an experiment, such as a holdout test, where a campaign runs for some audiences or regions and not for comparable others. That's the same logic as the market experiments and marketing mix modeling in Chapter 11. Integrated Marketing Communications. Attribution also has blind spots. It can't follow a customer who sees an ad on a phone and buys at a dealer, and privacy restrictions mean platforms increasingly estimate, rather than observe, some conversions.

12.4.4 AI in digital measurement and buying

AI now runs much of the day-to-day work of digital advertising. Ad platforms offer automated bidding that adjusts bids for each auction, and AI-driven campaign types that choose audiences, placements and even creative combinations across a platform's properties with little manual control. These systems can outperform manual management, especially for a small team. They also share the weaknesses you saw with programmatic buying in the previous chapter, plus two more:

• They optimize exactly what you tell them to. Tell a system to maximize test-ride bookings and it will find bookings, including people who book and never show up. Tell it to maximize ROAS on online sales and it may shift spending toward brand search, which looks efficient but adds few new customers. Choosing the goal is a strategic decision.
• Platforms grade their own homework. The platform selling the ads also reports how well they worked, often using its own attribution model. Independent checks, such as holdout tests and comparing platform reports with actual sales in Lumen's own systems, keep the numbers honest.

12.4.5 Vanity metrics and decision metrics

A last test for any metric: would a change in this number change what you do? Follower counts, likes and impressions are sometimes called vanity metrics, because they can rise without any effect on the business. They aren't useless, since reach and engagement can be early signals, but they shouldn't be the goal. Decision metrics connect to objectives: cost per test-ride booking, test-ride-to-purchase conversion rate, CPA compared with break-even CPA, and incremental sales from holdout tests. Build your dashboard around decision metrics, using the approach to dashboards from Chapter 6. Marketing Research and Analytics.

12.5 Lumen Cycles: getting found by families

Back to Jess's experiment. The AI answer that left out Lumen isn't a disaster, but it's a signal. Lumen's digital presence was built mainly for people who already knew the brand in Denver, Seattle, Portland and Salt Lake City. The national Cargo expansion needs families who've never heard of Lumen to find it, trust it and try it.

The previous chapter set aside about $460,000 of the first-year Cargo budget for digital and social. Jess and Marcus draft a plan for it. All figures are illustrative.

Area	Main job	First-year budget
Paid social (short video, targeted to parents in new markets)	Awareness and test-ride bookings	$160,000
Paid search (non-brand and brand terms, new markets)	Capture active shoppers	$140,000
Influencer partnerships (real local parents, micro and nano)	Credibility and content	$60,000
Content production (help videos, FAQ, family stories)	Content for social, search and AI answers	$60,000
Website, email and app improvements	Conversion and retention	$40,000
Total		$460,000

Check the total: $160,000 + $140,000 + $60,000 + $60,000 + $40,000 = $460,000.

Key decisions in the draft:

• Website. Rebuild the Cargo pages around parents' questions (capacity, range, hills, charging, storage, cost), add structured product data, and put test-ride booking and the dealer locator on every Cargo page.
• Content pillars. Family life on two wheels, safety, practical how-to and the numbers, with a monthly "school run" video series in a different city each month.
• Influencers. Work with real parents in the new markets who ride with their kids, require clear disclosure and safe riding in every post, and don't use virtual influencers. For a brand whose promise is safety and trust, the team concludes, a synthetic parent is the wrong messenger.
• AI use. AI drafts captions, post variants and FAQ answers from approved claims. People check every fact, film the real family content, and personally answer any comment about safety or a complaint.
• Search and AI answers. Publish the Cargo FAQ, keep specs consistent across the Lumen site and dealer sites, support the battery-safety PR program, and run a monthly check of what AI assistants say about family cargo bikes.
• Email and app. Triggered emails for cart abandonment and post-test-ride follow-up. Personalizing offers from app ride data waits until the responsible marketing review.
• Measurement. The main decision metrics are cost per test-ride booking, test-ride-to-purchase conversion rate (including dealer sales, tracked with booking codes), CPA compared with break-even CPA, and a holdout test of paid social in one new market to estimate its incremental effect.

Your marketing plan. Draft the digital and social plan section of your marketing plan for the Lumen Cargo's national expansion (or your approved company). In about one to two pages, include:

1. Channel roles: which digital and direct channels you'll use (website, email, mobile, social, search, influencers) and the job of each, consistent with your communications plan.
2. Social and content strategy: platforms and why, three or four content pillars, and two or three example posts or content pieces.
3. Influencer approach: the type of influencers you'd use, how you'd choose them, and how you'll handle disclosure. State your position on AI influencers and automated content.
4. Search plan: key keyword groups by intent, SEO priorities, and how you'll improve and monitor your brand's presence in AI answers.
5. Budget: a table splitting your digital budget across channels, consistent with the budget line you reserved last week.
6. Metrics: three to five decision metrics with targets, at least one break-even calculation (CPA or ROAS), and how you'll test incrementality.
7. AI use: where AI will assist, what a person must check, and what you won't automate.

12.6 Try it with AI

In this exercise you'll use an AI assistant to draft a month of social media content for the Lumen Cargo, then evaluate the draft against Lumen's positioning and this chapter's metrics. The goal is to practice judging AI-generated content as a manager would, not to publish it.

Lumen is fictional, so you can share its case facts freely. If you use a real company instead, follow the protect data rule: share only public information.

Step 1: Draft the calendar. Start a new conversation and paste this prompt.

Step 2: Evaluate the calendar yourself. Before asking the AI to improve anything, score the draft. Make a copy of the calendar with these added columns, and fill them in:

• On-positioning? (yes/partly/no). Does the post express safety, practicality and the family car-replacement idea, or could it be for any e-bike brand?
• Claims check. List every factual claim. Mark each one in the case facts or not in the case facts. Any specific number (range, savings, weight capacity) or superlative that wasn't in your prompt must be verified or removed.
• Safety and representation. Does any described scene show riders without helmets, unsafe child seating, or a narrow stereotype of who a family is?
• Platform fit. Does the format suit the platform, or is it the same post copied across both?
• Metric. For each post, name the one metric you'd use to judge it, using the chapter's terms. An awareness post might be judged on reach or video completion; a test-ride post on CTR and cost per test-ride booking.

Then look at the calendar as a whole. Is it balanced across pillars? How many posts actually ask for a test ride? Would the month, taken together, move parents in a new city toward booking one?

Step 3: Ask the AI to critique and revise, then check its work.

Compare the revision with your own evaluation:

• Did the AI remove every claim you flagged, without adding new ones? Read every caption again.
• Do its metric choices match the post's goal? A common error is judging every post by likes or engagement. Awareness posts and test-ride posts need different metrics.
• Does its advice on AI-generated visuals match the chapter's reasoning about trust, authenticity and disclosure? If it suggested an AI-generated "real family" testimonial, that's a red flag.

What to submit. Your instructor may ask for your evaluated calendar from Step 2, the final calendar after your own edits, a short paragraph on where the AI helped and where your judgment changed its output, and a one-line AI-use note following the disclose it rule.

12.7 Summary

Direct and digital marketing reaches specific people and seeks a measurable response. For a direct-to-consumer brand, the website is the hub, supported by permission-based email, mobile channels such as apps and texts, and older direct channels such as direct mail. As tracking across sites becomes harder, first-party data grows more valuable, and so does the responsibility to use it the way customers expect. Social media lets brands advertise and join conversations, but brands are guests on platforms they don't control. Content marketing earns attention by answering the audience's real questions, organized by pillars and an editorial calendar. Influencers can be credible sources when they genuinely fit the brand, and paid relationships must be clearly disclosed. Virtual influencers and automated content offer control and scale but put credibility and authenticity at risk, especially for brands built on trust.

Search captures people at the moment they express a need. SEO earns unpaid visibility slowly, and paid search buys it quickly through keyword auctions. AI answer engines now answer many questions directly, which reduces clicks and makes being named in an answer more important than being ranked; the best response is clear, consistent, credible information plus regular monitoring. Digital metrics such as CTR, conversion rate, CPA and ROAS make performance measurable, but they must be read against margins (break-even CPA and ROAS) and objectives. Attribution models assign credit in very different ways, and only experiments such as holdout tests show what a campaign actually caused. AI now handles much of the bidding and optimization, which makes choosing the right goal and checking the results independently a central human job.

12.8 Review Questions

1. Why is an email list considered more valuable to a brand than the same number of social media followers? What makes email marketing permission-based, and why does that matter?
2. Lumen is considering using app ride data to personalize offers. Describe one use customers would probably welcome and one they might find intrusive, and explain the difference.
3. Explain the difference between hero, hub and help content, and give a Lumen Fold example of each. Which type is most likely to appear in AI-generated answers, and why?
4. A micro-influencer posts a video of her kids riding a Lumen Cargo, with "#family #ebike #lumenpartner #ad" at the end of a long caption. Does this meet the disclosure expectations described in the chapter? What would you ask her to change?
5. How do AI answer engines change the goals of search marketing? List three things a brand can do to improve its presence in AI answers, and one thing it cannot control.
6. A paid social campaign for the Lumen Commute ($1,899) costs $5,000, gets 250,000 impressions and 2,000 clicks, and produces 10 purchases. Calculate CTR, CPC, conversion rate, CPA and ROAS. If the Commute's gross margin is 35%, what are the break-even CPA and break-even ROAS, and is the campaign profitable on the first purchase? Show your work.
7. Explain why last-click attribution might lead Lumen to cut spending on social video, and why that could be a mistake. What kind of test would give a better answer?
8. Discussion: should Lumen ever use a virtual influencer or fully automated social content? Take a position and defend it using the chapter's ideas about credibility, disclosure and brand positioning.