RekazAds & RekazRocket Learnings

Rekaz Ads began as a simplified interface over Google Ads. It ended up teaching us that advertising is not the advertisement — it is the complete system connecting attention, trust, inquiry, follow-up, booking, operations, and repeat engagement.

The journey ran through five distinct stages. First, a simplified Google Ads interface where merchants still built their own campaigns while we silently applied better defaults. Second, the addition of Meta Ads, which exposed a problem the interface could not fix: merchants uploaded creatives that could not persuade/hook anyone. Third, Marvel Salon — a fictional, controlled salon built to test roughly 45 creative concepts and find out what actually generates demand. Fourth, the Sales Package, which converted those creative learnings into a productized service where we launch the campaign for the merchant. Fifth, Rekaz Rocket — running the full acquisition system for real merchants, which is where the hardest lessons surfaced: demand generation is easy to waste when the merchant replies slowly, follows up weakly, or can't handle the volume.

The experiments, in the order they were run. Several lessons in this document only exist because of the later entries — they were not known at the start.

What merchants ultimately want is simple: sales and more customers. But Rekaz is not a marketplace — we cannot hand merchants a stream of customers ourselves. That gap needed to be closed, and advertising was the way to close it.

The founding observation of Rekaz Ads was equally simple: platforms like Google Ads and Meta Ads are too difficult for most SMB owners to operate.

Merchants generally do not understand — and do not want to manage:

• Complex campaign structures
• Bidding strategies
• Keyword configuration
• Negative keywords
• Ad-set configuration
• Audience targeting
• Campaign objectives
• Optimization settings
• Attribution
• Creative strategy

The purpose of Rekaz Ads was to put a simpler surface between the merchant and that complexity, so campaigns could be launched without touching it directly.

Merchants do not necessarily need more control. They need fewer and easier decisions.

The first version supported Google Ads only. Importantly, we did not yet create or manage everything. The merchant still:

• Uploaded their own creatives or ad material
• Selected or entered their own targeting
• Chose the relevant campaign information
• Used a simpler Rekaz interface instead of the raw Google Ads console

We simplified the experience, but the merchant was still actively building the campaign. What the merchant never saw was the layer of assumptions we applied silently on their behalf.

The negative-keyword example

In Google Search campaigns, we automatically excluded low-intent or irrelevant searches, such as:

• وظائف — jobs
• مجانا — free
• jobs
• free
• DIY

…and similar low-intent terms. The purpose was to stop merchants from paying for clicks from job-seekers, freebie-hunters, and do-it-yourselfers. The merchant never needed to understand or configure any of it. Negative keywords are exactly the mechanism Google itself documents for excluding searches you don't serve — Google's own example is an optometrist excluding “drinking glasses”^[1] — and exactly the kind of mechanism an SMB owner will never configure on their own.

Hide complexity when we can make a better default decision than the merchant.

Rekaz Ads initially supported Google Display Ads too. They were later removed — and not because they were technically broken.

Display advertising is structurally suited to brand awareness, reach, repeated exposure, and familiarity. Our merchants, however, expected immediate sales and bookings, and evaluated every campaign as if it were direct response. The industry data explains why the gap is so wide: WordStream's benchmark across 14,000+ Google Ads accounts measured an average click-through rate of 3.17% on Search versus 0.46% on Display — search captures existing intent; display has to interrupt someone doing something else, which is why Google positions it for awareness and consideration^[2].

The mismatch played out the same way every time:

Rather than keep exposing a campaign type merchants consistently misread, we removed the option entirely. A product-simplification decision, not a technical one.

A feature can be valid in theory and still be wrong for the product if customers consistently use it with the wrong expectations.

After Google Ads, we added Meta Ads — and inherited a new class of problem.

On Google Search, existing search intent carries part of the campaign: the customer already typed what they want. On Meta, nobody is searching. The advertisement has to create the interest itself, which makes creative quality the dominant variable.

When merchants uploaded their own creatives, many used weak images that could not convert or hook anyone. The canonical example:

• An image of nails only
• No salon name
• No branch or location
• No phone number
• No offer
• No hook
• No compelling value proposition
• No reason to contact the salon
• No clear call to action

The campaign could be configured perfectly and still fail, because the advertisement itself persuaded no one. Simplifying campaign publishing was not enough — we needed to solve the creative problem.

What types of creatives, hooks, offers, and formats actually generate demand?

7.1 — Why it was created

Marvel Salon belongs to the Rekaz Rocket story, not just the creative one. Rocket was its own experiment with its own mission: could we increase sales for salons — especially صوالين الحي, the small neighborhood salons that nobody knows, often struggling businesses? (Darnita, later in the timeline, was the exception — large compared to every other salon we worked with.)

Marvel Salon was the controlled starting point of that experiment. Rather than gambling a real merchant's money and reputation on untested ideas, we built a controlled, fictional salon we fully owned and used it to test many creative concepts directly — informed by what we had seen Meta merchants upload.

Owning a fictional salon gave us control over every variable merchants normally control badly:

• The salon identity
• Branding
• Visual positioning
• Instagram presence
• Offers
• Creative concepts
• Campaign structure
• Customer-facing messaging

7.2 — Identity and location

Marvel Salon was a fake salon. It did not physically exist. It was presented as located at:

حي الروابي – شارع عنيزة، الرياض

Because there was no real shop, the Google Maps listing pointed generally to the street rather than to a specific building. The Instagram page — @marvelsalon.sa, still up today — was designed to look premium, fancy, modern, high-end, credible, and established — and approximately 1,000 fake followers were purchased for the account, on the assumption that follower count could manufacture perceived reputation and social proof. That assumption was part of the experiment, not a proven mechanic.

7.3 — Campaign structure: CBO

Marvel was tested through multiple campaigns. Some campaigns carried multiple ad sets, and roughly 45 creative concepts were tested in total.

The structure leaned heavily on CBO — Campaign Budget Optimization. Budget is assigned at the campaign level, and Meta distributes spend across ad sets based on the performance signals it detects. The intent was to let the platform “do its own thing” and push spend toward whatever it expected to perform.

7.4 — The creative tests

Among the ~45 concepts tested:

• Free Matcha with a manicure / nail appointment
• Bring your daughter — her service is free
• Bring your friend — her service is free
• Free coffee
• Free hot chocolate
• Free frozen yogurt
• Service bundles
• Gift-based offers
• Companion-based offers
• Strange / unconventional promotional ideas

Most creatives did not perform. The Matcha concept was the strongest performer and became the main winning reference creative of the whole experiment.

7.5 — Results

Marvel Salon generated a very large number of inquiries through WhatsApp and Instagram DMs. But the strongest evidence of success was physical:

Some customers travelled to شارع عنيزة looking for the salon. Because the map listing covered the street rather than a storefront, they arrived in the area, searched around — and then contacted the account angrily, saying they were already on the street and could not find the salon.

That anger was the proof. The campaign had generated more than clicks or curiosity. Customers trusted the brand enough to:

1. Contact the salon.
2. Consider booking.
3. Travel to the location.
4. Physically search for the business.

The salon was fake, but the customer intent was real.

7.6 — Why Marvel may have succeeded

The same success did not always transfer to other salons — notably Orvi Nail Spa. One major hypothesis is that Marvel's Instagram page and overall brand did much of the work:

1. The account had ~1,000 (purchased) followers.
2. Follower count may have created perceived reputation and social proof.
3. Marvel looked premium and high-end.
4. The branding suggested a fashionable, luxurious experience.
5. The page may have built trust before the customer ever messaged.
6. Customers may buy the overall image and experience, not just the technical service.

Orvi, by comparison, felt like صالون الحي — the neighborhood salon. The working hypothesis is that salon customers don't only buy the service; they may also buy the feeling of luxury, status, the environment, a fashionable experience, something photographable, something they can show off socially.

The Sales Package came after Marvel in the timeline — but the two were not directly related. Marvel was not run in order to build a package; it was its own experiment. When the Sales Package was later created, we simply reflected what we had already learned — the creative patterns, hooks, and formats that had proven themselves — into how the package works. The sequence:

1. Rekaz Ads launched with Google Ads.
2. Meta Ads was added.
3. Merchants uploaded poor creatives.
4. Marvel Salon ran as its own experiment.
5. Creative concepts and formats were tested.
6. We identified useful creative patterns.
7. The Sales Package was built later, and those learnings were reflected into it.

With the Sales Package, the merchant no longer built anything alone. The merchant would purchase the package, connect their advertising account, and rely on business information already inside Rekaz from onboarding. We would then launch the campaign for them.

Automated targeting assumptions

Because the business activity was already known from onboarding, we could make targeting assumptions automatically. For a salon, for example:

These are examples of automated assumptions, not universal rules applied to every merchant. Radius-based local targeting of this kind is standard, documented practice: Meta officially supports radius targeting from 1 km up to 70 km around a point outside the US, and Google Ads supports the same pattern for local businesses^[3].

Creative-generation rules

The Sales Package encoded the creative lessons into an explicit generation system. Rather than “make a pretty image,” the generator builds promotional posters against hard rules:

• Arabic-language promotional design
• Right-to-left layout
• Premium, minimalist visual style
• Consistent visual hierarchy
• Clear logo / brand-name zone
• Large hero headline
• Single dominant emphasis word
• Controlled accent colors
• A clear offer or free gift
• Badge when relevant
• Short body copy
• City · street · place name
• Phone number
• Optional feature strips
• Business-specific visual styles
• Correct Arabic text rendering
• Clear composition — one coherent scene, not disconnected panels

Every rule is the inverse of the “nails only, no name, no phone, no offer” failure that started this chapter of the story.

Visual treatment changes by business category — beauty salons, barbershops, spas, gyms, sports academies, resorts, mobile services, car-care, clinics, photography studios, pet care, event services.

Rocket was the next step in involvement: not a tool, not a package, but us running the campaigns and testing the complete acquisition process with real merchants. During the early experiments, we often paid for the campaigns ourselves — the work was still experimental.

10.1 — Speed of response

Responding quickly made a major difference in conversion. A customer who sends an inquiry is interested at that moment. Wait too long and the customer contacts a competitor, loses interest, gets distracted — the inquiry goes cold.

The external research is emphatic on this point. The MIT / InsideSales.com Lead Response Management study found that contacting a lead within 5 minutes versus 30 minutes makes you ~100× more likely to reach them and 21× more likely to qualify them. Harvard Business Review's audit of 2,241 US companies found the average response time was 42 hours, only 37% responded within an hour, and 23% never responded at all — while firms responding within the hour were ~7× as likely to qualify the lead^[6]. Our salons were living examples of the slow majority — and the same-minute call workflow we built for Orvi's Snapchat leads (Case 02) was the counter-example: call immediately, and the customer would most likely book.

10.2 — Follow-up

When a customer inquired, got an answer, then went silent — we found that following up after roughly two to three hours could make the difference. Silence does not mean disinterest. The customer may have become busy, forgotten, needed time, or been comparing options. A structured follow-up process recovers leads that would otherwise be lost — which matches the broader sales research: industry data has 44% of salespeople giving up after a single follow-up, while most sales close on later contacts (the famous “80% of sales take five follow-ups” figure is industry folklore without a peer-reviewed origin, but the direction is well supported)^[7].

10.3 — Ads amplify operations

Good advertising exposes a business's operational weaknesses. More leads means the business must be capable of replying, following up, confirming appointments, managing availability, delivering the service, and handling demand. If it can't, better advertising simply manufactures more missed opportunities — Glowing Sisters and Orvi on one side of that ledger, Darnita on the other.

At Orvi, the Matcha creative and others performed well at first — then performance declined, with the dashboard showing high frequency: the ads were being shown repeatedly to the same or overlapping people.

Explanation one — creative fatigue

An audience that sees the same ad repeatedly becomes less responsive to it. This was the initial assumption, and it is a well-documented effect. Meta officially flags a creative as “fatigued” when its cost per result reaches 2× its historical baseline, and Meta's own data science team measured CTR decaying with repeated exposure — with conversion likelihood dropping roughly 45% by the fourth repeat exposure, and refreshed creative recovering ~8% on average^[8].

Explanation two — low-hanging fruit

The platform may initially find the people most likely to convert — the easiest, cheapest customers. As those high-propensity users get reached or converted, the remaining audience is harder and more expensive. Performance declines even if the creative is not exhausted. This is audience saturation / diminishing marginal returns — and notably, Meta's own analysis treats it as a separate, measurable phenomenon from creative fatigue (tracked via the first-time impression ratio), while practitioners describe the same economics as rising marginal CPA hidden behind a stable average CPA^[9]. Our “low-hanging fruit” intuition has an official name.

The observed decline may have been a combination of:

• Creative fatigue
• Audience saturation
• High frequency
• Exhaustion of easy converters
• Diminishing marginal returns
• A limited local audience

These frameworks were learned after the experiments, and they retroactively explain much of what was observed — especially Orvi's 15–20-day delayed conversions. The funnel model divides audiences by how close they are to buying, and prescribes different creative goals for each stage — awareness content at the top, comparison and trust content in the middle, offers and urgency at the bottom^[10].

TOFU — awareness

Who: people who don't know the business, may not know they need the service, aren't ready to book. Creative goals: attention, awareness, education, entertainment, problem recognition, brand introduction.

Concepts for our verticals: nail-care educational videos · “three mistakes that damage your nails” · behind-the-scenes salon content · transformations · trend Reels · lifestyle and brand-story content · satisfying service videos · educational car-care content · “what happens when you never deep-clean your car interior?”

TOFU should not always force an immediate discount or booking — the goal is to make the customer aware and interested.

MOFU — consideration

Who: people who know the brand, have interacted, are comparing, need more trust or information. Creative goals: trust, explanation, objection handling, differentiation, demonstration, social proof.

Concepts: testimonials · before-and-after · hygiene procedures · technician introductions · service comparisons (gel vs. acrylic) · how the mobile service works · what's inside a detailing package · FAQs · reviews · Instagram profile content · retargeting people who watched a video or opened the booking page.

BOFU — conversion

Who: people who understand the business and are actively considering booking — they need a final reason. Creative goals: booking, closing, urgency, reducing hesitation, a clear offer.

Concepts: limited-time package · the 299 SAR multi-service offer · free Matcha with a paid appointment · “book today” · limited appointments · retargeting past inquirers · WhatsApp follow-up · appointment reminders · review + clear CTA · “complete your booking.”

The funnel is not perfectly linear — customers move between stages and meet content out of order. It remains useful as a messaging framework.

Multiple touchpoints before conversion

The old marketing “Rule of 7” — traced to 1930s Hollywood promotion practice and formalized in the 1980s by Dr. Jeffrey Lant — holds that customers need repeated exposure before they trust a business enough to buy. Its modern descendants say 7–12+ touches; RAIN Group's study of B2B buyers found an average of 8 touches just to get a first meeting. None of these numbers is a scientific constant (the number 7 has no peer-reviewed proof, and the widely repeated “Google 7-11-4 rule” has no locatable Google publication behind it)^[11]. It is a heuristic. The real lesson:

Conversion is often the result of accumulated trust rather than a single advertisement.

A plausible journey for one of our salon customers:

1. Sees a Meta ad.
2. Opens the Instagram profile.
3. Reviews the follower count.
4. Looks at previous posts.
5. Sees another Story ad.
6. Searches for the salon on Google.
7. Checks the map.
8. Messages on WhatsApp.
9. Delays the decision.
10. Sees a retargeting ad.
11. Receives a follow-up.
12. Books.

This connects directly to the Orvi observation: conversions surfacing 15–20 days after first contact, helped by retargeting. Industry data agrees that the overwhelming majority of visitors — the commonly cited figure is 97–98% — don't convert on first exposure, and that retargeted display ads earn roughly 10× the CTR of standard display; these numbers trace to vendor data (Criteo, AdRoll) rather than academic studies, so treat them as directional, but the mechanism matched what we saw at Orvi^[12].

Owned audience & customer data

The WhatsApp shift generalizes: platforms control access to audiences, and a merchant can lose that access through account restrictions, policy changes, rising costs, algorithm shifts, or bans. A phone number is durable. It enables follow-up, retargeting, reactivation, future offers, CRM history, and repeat-business measurement. WhatsApp doesn't automatically convert better — it builds a more valuable long-term asset.

Independent sources that affirm — or contextualize — what the experiments taught. Internal observations stand on their own; these show the industry arrived at the same conclusions.

1. Google Ads Help (official) — “About negative keywords.” Negative keywords exclude search terms from campaigns so ads don't show for things you don't offer (Google's own example: an optometrist excluding “drinking glasses”). The mechanism behind our hidden وظائف / free / DIY exclusions.
   support.google.com/google-ads/answer/2453972
2. Google Ads Help (official) + WordStream benchmarks — Google positions Display for reaching people “as they browse,” i.e. awareness/consideration, not active intent. WordStream's study of 14,197 US accounts measured average CTR of 3.17% on Search vs 0.46% on Display; later benchmarks widened the gap further.
   support.google.com/google-ads/answer/2404190 · wordstream.com — Google Ads industry benchmarks
3. Meta Business Help Center + Google Ads Help (official) — Meta radius targeting: 1–70 km around a point outside the US, with “living in” vs “recently in” options. Google Ads supports radius targeting around a location (minimum 1 km). Basis for the ~10 km salon-radius assumption being standard practice.
   facebook.com/business/help — Use Location Targeting · support.google.com/google-ads/answer/1722043
4. SchedulingKit (industry aggregate) — Appointment-deposit statistics for salon/spa businesses: deposits are credited with cutting no-shows roughly 40–60%, with ~72% of consumers saying they're willing to pay one — directional industry data, not primary research, and it does not capture the conversion friction we measured at Orvi.
   schedulingkit.com — appointment deposit statistics
5. Think with Google × Boston Consulting Group — “Responsible Marketing with First-Party Data.” Brands using first-party data for key marketing functions achieved up to 2.9× revenue uplift and 1.5× cost savings — the industry-scale version of “a phone number is an asset; a DM is rented.”
   thinkwithgoogle.com — first-party data BCG report
6. Harvard Business Review (Oldroyd, McElheran, Elkington, 2011) + MIT / InsideSales.com Lead Response Management Study (2007) — HBR audit of 2,241 US companies: average lead response time 42 hours; 37% responded within an hour; 23% never responded; within-the-hour responders were ~7× as likely to qualify the lead. The MIT/InsideSales study: contacting within 5 minutes vs 30 makes contact ~100× more likely and qualification 21× more likely.
   hbr.org — The Short Life of Online Sales Leads · MIT/InsideSales.com study (PDF)
7. Invesp / IRC Sales Solutions — 44% of salespeople give up after one follow-up; most deals close on later contacts (breakdown: 22% quit after two, 14% after three, 12% after four). Caveat: the ubiquitous “80% of sales require 5 follow-ups” has no peer-reviewed primary source — treat as directional folklore.
   invespcro.com — sales follow-up statistics · ircsalessolutions.com — follow-up statistics
8. Meta Business Help Center + Analytics at Meta (official data-science blog, 2023) — Meta's delivery system flags “creative fatigue” when a creative's cost per result hits 2× its historical baseline. Meta's researchers measured CTR declining with repeat exposure (∝ (N+1)^−0.43), conversion likelihood dropping ~45% by the 4th repeat exposure, and refreshed creative improving conversion ~8% on average.
   facebook.com/business/help/1346816142327858 · Analytics at Meta — Creative fatigue
9. Analytics at Meta + practitioner analysis — Meta's fatigue research explicitly separates creative fatigue from audience saturation (measured via first-time impression ratio) — official confirmation that they are distinct mechanisms. Practitioner analyses describe the same economics as marginal CPA rising while average CPA hides it: the system reaches the cheapest, most-likely converters first. This is the “low-hanging fruit” explanation with a formal name.
   Analytics at Meta — fatigue vs saturation · get-ryze.ai — marginal vs average CPA
10. HubSpot / Dave Chaffey (Smart Insights) — Canonical TOFU/MOFU/BOFU definitions: educational awareness content at the top, comparison and trust-building content in the middle, case studies / offers / trials at the bottom.
    blog.hubspot.com — content for every funnel stage · davechaffey.com — TOFU vs MOFU vs BOFU
11. RAIN Group + Rule-of-7 historiography — RAIN Group's study of 488 B2B buyers / 489 sellers: average 8 touches to secure a first meeting (top performers: 5). The “Rule of 7” traces to 1930s Hollywood promotion and Dr. Jeffrey Lant's 1980s formulation (“7 exposures in 18 months”); no peer-reviewed proof of the number exists, and the “Google 7-11-4 rule” has no locatable Google publication.
    rainsalestraining.com — how many touchpoints · anartfulscience.com — Rule of 7 origins
12. Criteo / industry aggregates — Retargeting rationale: ~97–98% of visitors don't convert on first visit; retargeted display CTR ≈ 10× standard display (~0.7% vs ~0.07%); retargeted users ~43% more likely to convert. Caveat: vendor-attributed figures (Criteo, AdRoll, Wishpond) without published methodology — directional, but consistent with Orvi's 15–20-day delayed conversions.
    criteo.com — retargeting 101 (PDF) · cropink.com — retargeting statistics