CASE: Six Months of Meta Ads

How One Sri Lankan Adhesive Brand Engineered 4.05 Million Views from Meta Ads.

A Sri Lankan adhesive brand went from a dormant digital presence to 4.05 million views, 175,696 engagements, and a 653,648-person warm audience pool inside Meta’s algorithm in six months. Total ad spend was LKR 632,000. Overall cost per engagement was LKR 1.73. In the final month of the engagement, the gap between the best-performing creative and the worst-performing creative measured 86 times.

This article documents the full methodology behind that engagement. October 2025 to March 2026. Six phases, six permanent learnings, one creative formula confirmed across more than 180 ads. Every figure below comes from the audited final performance report delivered to the client. The brand is anonymised at our discretion. The data is not.

The strategic argument the case proves is this. Paid social in Sri Lanka does not fail because the platforms do not work. It fails because most accounts are run as activity rather than engineering. Treat paid media as engineering, measure the right thing on the right platform, hold creative discipline across hundreds of variants, and the unit economics shift dramatically.

What follows is how that was done.

The Diagnosis Before The Spend

In August 2025, the brand had a dormant Facebook page sitting at roughly 9,700 followers. Distribution was strong. Retail presence was nationwide. Trade relationships were intact. Consumer mindshare was eroding.

The market reality the brand sat inside is worth stating in numbers, because the strategic case for digital investment depends on it.

Sri Lanka has approximately 12 million internet users. Of those, around 8 million are active on mobile-first social platforms. Industry data on consumer purchase behaviour indicates that approximately 60 percent of shoppers begin their buying journey online, even for categories traditionally considered offline-first.

In the adhesive category specifically, several aggressive competitors were expanding their digital presence rapidly. Some were running humour-led creative that built distinctiveness. Others were investing in lifestyle positioning that built aspirational pull. The brand at the centre of this case study was visible in trade but invisible online. It occupied what the Ehrenberg-Bass Institute calls a “mainstream” position on the Centrality-Distinctiveness map. High centrality, low distinctiveness. Familiar but fading.

The diagnostic before any ad spend was committed produced three clear findings.

The brand was not suffering an awareness problem. It was suffering a distinctiveness problem. Customers knew the name. They could not articulate why this brand specifically rather than any other. In a category where competitors were aggressively building emotional and creative distinctiveness, this was a quiet erosion of preference that would not show up in distributor sell-through until it was too late to correct.

The right channel for solving this was paid social, not search. Adhesives are not a high-intent search category. Buyers do not Google for “best contact adhesive Sri Lanka” before walking into a hardware store. They buy what they recognise, what they trust, and what they remember from the last time they had a problem. Paid social was the channel that could rebuild that recognition.

The creative direction needed to inverse the category default. While competitors were investing in humour and exaggeration, this brand had an opportunity to occupy “trust and resolution” as its emotional territory. Calm, no-drama fixes. The hero is the user, not the product.

This diagnostic took two weeks. It produced a written strategy document. The client signed it off before any ad account was touched.

Phase 1, October 2025, The Awakening

The first month of paid activity is misunderstood by almost everyone outside the paid media discipline.

Most clients arrive expecting month one to deliver sales. Most agencies oblige this expectation by reporting impressive-looking impression and reach numbers that obscure the fact that month one is an investment, not a return.

Here is what month one actually does on a cold-start account. Meta’s machine learning algorithm has no prior data to optimise against. Every rupee spent in this phase is buying two things simultaneously. The first is reach. The second is data. The data is the more valuable purchase.

In October 2025, the brand spent LKR 86,249 across a deliberately wide-net awareness campaign covering the full product range. Sealants, super glue, epoxy, PUR, no more nails, weather sealant, gap filler. The strategic intent was not to sell any specific product. It was to identify which products and which audiences the algorithm could match efficiently.

The output of October looked, on the surface, modest.

• Total reach: 571,000 unique users
• Tagged warm audience built inside Meta: 653,648 users
• Facebook engagement rate: 1.21 percent
• Instagram engagement rate: 4.22 percent

A traditional vanity-metric report would have framed October as inefficient. The Facebook engagement rate of 1.21 percent was below industry benchmarks for category averages. The cost per engagement appeared high relative to what a mature account would deliver.

The correct way to read October was different. The question was not “why was October inefficient.” The question was “what did October buy that November could not survive without.”

The answer was infrastructure. The 653,648 tagged users in Meta’s system were not eyeballs that had been rented. They were a data asset. Meta now knew, with high confidence, who was likely to interact with this brand’s content. The next month, the algorithm would no longer be guessing.

This is the single most misunderstood principle in Sri Lankan paid social. Month one is not where you measure return. Month one is where you build the audience pool that months two through twelve will exploit.

The creative testing in October was equally important. Twenty-two distinct ads ran in parallel. Different copy directions, different visual styles, different language combinations. By week three, a clear creative signal had emerged. Focused Sinhala product posts on Facebook were delivering 9.3 to 13.1 percent click-through rates at low cost. Wide-net English awareness ads were delivering 0.18 to 0.22 percent. The performance gap was not subtle. It was directional.

The implication for month two was set. Stop testing widely. Scale what is winning.

Phase 2, November 2025, The Surge

November was where October’s investment paid back, and the magnitude of the payback is the strongest argument in the entire case for disciplined cold-start spending.

The strategic shift entering November was twofold. The audience targeting tightened from broad awareness to retargeting and lookalike audiences built on October’s warm pool. The creative direction narrowed from product range to focused, problem-first storytelling in Sinhala.

The results spoke directly to whether the diagnostic was correct.

• Facebook views jumped from 1.13 million in October to 1.54 million in November. A 35.6 percent month-over-month increase.
• Facebook engagement rate rose from 1.21 percent to 7.1 percent. A 487 percent jump.
• Instagram engagement rate rose from 4.22 percent to 27.2 percent. A 544 percent jump.
• Facebook interactions rose 168 percent.

The single highest-performing post of November illustrates the creative formula that won. A Sinhala-language ad for weather sealant carried the headline “සීල් කරා, අමතක කරා,” which translates to “seal it and forget it.” The supporting line spoke directly to the seasonal pain point of monsoon rain seeping through walls. The post delivered 75,768 views, 58,790 unique reach, and 3,239 interactions. Critically, 88 percent of the people who saw the post were not followers of the page. This was new audience discovery at scale, not content recycling to existing fans.

The creative pattern that emerged across November’s top posts was the formula that would be confirmed across the next four months and 180 plus ads. Four parts.

First, a Sinhala headline that stated the problem in the audience’s own language. Not a translated English headline. A headline written in Sinhala first, by a writer thinking in Sinhala. The cultural and emotional register of the language matters. Most agencies write in English and translate. The difference is the difference between content that lands and content that scrolls past.

Second, an emotional before-moment. The panic, the embarrassment, the inconvenience, the worry, shown before the product enters the frame. Generic product demos skip this step. Skipping it kills shareability.

Third, the product as the effortless resolution. No drama, no complex instructions, no aspirational over-promise. The hero of the ad is the user, not the product. The product just makes the user look competent.

Fourth, a seasonal or cultural hook tied to a real Sri Lankan moment in time. Monsoon prep. Awurudu home repairs. Festival cleaning. School holiday DIY projects. Timing creative to real life moments multiplies organic reach at zero additional cost.

This four-part structure won, without exception, across every month of the engagement.

Phase 3, December 2025, The Consistency Lesson

December is the most expensive advertising month of the year on Meta, anywhere in the world. The auction floods with Black Friday campaigns, Christmas retail offers, and year-end promotions from every brand category. CPM, the cost per thousand impressions, rises substantially across the board. The same budget reaches fewer people. Every rupee buys less.

This is a known constraint of the calendar, not a failure of the platform. Mature paid media accounts plan for it. They build December content stockpiles in November. They protect feed presence with paid amplification. They treat December as a defensive month, not a growth month.

The case study brand entered December in a strong position. November’s surge had built algorithmic momentum. The audience pool was warm. The creative formula was confirmed.

Then a 15-day publishing gap broke that momentum.

The cause was a content approval bottleneck unrelated to the platform or the agency. Publishing paused. The Meta algorithm interpreted the silence as a signal to deprioritise the page in feed delivery. Audience attention, once redirected to other content sources, did not return automatically.

The data tells the story.

• Facebook views dropped from 1.54 million in November to 212,000 in December. A 42.8 percent decline.
• Total engagement dropped 39.5 percent.
• The drop was disproportionately costly because it happened in the most expensive month of the year. Recovering ground in December cost more per impression than it would have in any other month.

The strategic lesson the case proves is not specific to one client. It is universal to any business running paid social.

Algorithmic momentum is unforgiving. Meta and Instagram both run delivery prioritisation models that reward consistent publishing and penalise gaps. The penalty is not a small adjustment. In observed cases including this one, a single 15-day gap can erase several months of compounding algorithmic trust. The audience does not leave. The platform stops showing them your content.

The cost of recovery is asymmetric to the cost of consistency. Maintaining a daily or near-daily publishing cadence is a fixed effort. Recovering from a publishing gap requires heavy paid amplification, refreshed creative, and weeks of feed pressure. The math always favours maintenance.

Consistency is therefore not a content goal. It is infrastructure. A missed week is not a missed post. It is a system reset.

This lesson is now hardcoded into how the agency sets up every client workflow. Approval processes are mapped before the first ad goes live. Content stockpiles are built three weeks ahead. Backup creative is approved in advance. The December dip in this engagement is what made these protocols non-negotiable.

Phase 4, January 2026, The Recovery

The recovery in January answered a question that mattered more than the December dip itself. Once algorithmic momentum is broken, can it be rebuilt at all?

The answer the data delivered was yes, but the method matters. Organic alone would not have recovered the position in time. The recovery required deliberate paid amplification across the warm audience pool to restore feed pressure quickly.

The strategy was straightforward. Engagement boost campaigns were activated across the proven creative formats from October and November. Retargeting and lookalike audiences from the warm pool were the primary delivery targets. The budget was front-loaded to push feed presence aggressively in the first ten days of January.

The market response was immediate.

• Facebook views recovered from 212,000 in December to 674,400 in January. A 157.3 percent month-over-month increase.
• Instagram views grew from 53,500 in December to 195,300 in January. A 264.4 percent increase.
• Instagram reach grew 405 percent.
• Three-second video views grew 1,308 percent, the largest single-metric jump of the entire engagement.

The three-second view metric is significant because it indicates retention. Reels and short-form video succeed or fail in the first three seconds. A 1,308 percent jump in that metric indicates that the creative was holding attention immediately on view, which is the single most important quality of a feed-native video ad.

Of January’s 674,400 Facebook views, approximately 623,000 came from paid amplification. Roughly 50,000 came from organic discovery. This 92 percent paid contribution is not a failure of organic. It is a reflection of what was needed to recover from a momentum gap during the most expensive advertising window of the year. Paid amplification did the heavy work of restoring feed pressure. Once that pressure was restored, organic engagement returned in February.

Phase 5, February 2026, The Peak

February confirmed that the recovery was structural, not just a paid amplification artifact.

• Facebook views reached 1.15 million, a 150 percent month-over-month increase.
• The content mix shifted permanently. Reels jumped from 4.5 percent of total content in December to 35.3 percent of total content in February.
• Reels drove 67.4 percent of all new follower discovery in the month.

The Reels shift is the most strategically important content finding of the engagement. The brand’s audience, predominantly DIY contractors and homeowners aged 25 to 44, had been assumed to prefer static images with clear product information. The data showed otherwise. Short-form video was driving the majority of discovery and follower acquisition by February.

Cost per engagement on video was, on average, five to twelve times lower than cost per engagement on static images. The best video CPE recorded in the engagement was LKR 0.18 in March. The best static CPE was approximately LKR 2.00. This is not a small difference. It is an order of magnitude difference in unit economics.

The implication for budget allocation is direct. Resources that were originally weighted toward static photography production needed to shift toward short-form video. By February, the production split had moved from 95 percent static to 65 percent static, with video continuing to grow share.

The Reels finding also revealed something about how the buyer actually interacts with content in this category. The data on attribution sources showed that 56.2 percent of new follower discovery came from Reels, 19.2 percent from the Facebook feed, and the remainder from a mix of profile visits and direct shares. The buyer watches video. The buyer does not click through to external sites. The buyer reacts and saves. This profile of behaviour required a different KPI framework, which is covered in Section 8 below.

Phase 6, March 2026, The Close

March was the final month of the engagement. The campaign closed not because performance had declined, but because of a client business decision made mid-momentum.

The month delivered the strongest unit economics of the entire six months.

• Total ad spend in March: LKR 93,700.
• Total engagements: 54,146.
• Unique reach: 419,781.
• Overall cost per engagement: LKR 1.73.
• 19 active ads under systematic naming and tracking.

The most strategically important finding from March is the one that should reshape how every business thinks about creative testing in paid media.

The gap between the best-performing ad in March and the worst-performing ad in March measured 86 times.

The best ad delivered engagement at LKR 0.18 per engagement. The worst delivered engagement at approximately LKR 15.50 per engagement. Both ads ran in the same account, in the same month, against overlapping target audiences. The only meaningful variable between them was creative.

This is the entire argument for treating paid media as engineering rather than activity. Most agencies and most in-house teams run their full ad set at the average. They do not isolate top performers. They do not aggressively kill bottom performers. They report on the aggregate, which masks the fact that 80 percent of results are coming from 20 percent of creative.

The discipline that produced March’s unit economics required three operational habits.

First, weekly creative review. Every ad’s cost per engagement was reviewed against the cohort. Bottom-quartile ads were paused or rewritten. Top-quartile ads received reallocated budget.

Second, structured naming conventions. Every ad in the engagement was named according to a system that allowed creative attributes to be filtered and compared. This sounds administrative. It is, in fact, the foundation of any reliable creative analysis. Without it, the 86 times gap could not have been identified.

Third, willingness to act on the data. Killing a creative that the client likes but the data does not support is professionally uncomfortable. It is also the discipline that separates engineered paid media from cosmetic paid media.

The Six Permanent Learnings

The numbers above describe a single engagement. The learnings that follow are now applied to every account the agency runs.

Learning 1: Sinhala problem-first copy outperforms English product copy, without exception.

Across all six months and more than 180 ads, every month showed the same pattern. Sinhala copy that stated the problem in the audience’s own emotional language outperformed English product demos. The performance gap was not marginal. It was directional and consistent. The implication for any consumer brand operating in Sri Lanka is that the default English-first content production model is leaving substantial performance on the table.

Learning 2: Consistency is infrastructure, not content.

December proved that a single 15-day publishing gap can erase months of algorithmic trust. The lesson is operational, not creative. Publishing pipelines must be designed to be resilient against approval delays, holiday seasons, key person dependencies, and content production bottlenecks. A missed week is a system reset, not a missed post.

Learning 3: Video cost per engagement is five to twelve times more efficient than static image cost per engagement.

This finding shifts budget allocation logic permanently. The best video CPE recorded in this engagement was LKR 0.18. The best static CPE was approximately LKR 2.00. Reels also drove 67.4 percent of new follower discovery in February. Any brand that has not yet shifted production weight toward short-form video is paying a substantial efficiency tax for static-first content production.

Learning 4: Facebook and Instagram serve different strategic roles and require different KPIs.

Facebook in this engagement functioned as a click-through and intent engine. Higher click-through rates, higher cost per click, users actively clicking external links. Instagram functioned as an awareness and discovery engine. Lower click-through rates, dramatically lower cost per engagement, users watching and saving rather than clicking. Measuring Instagram by Facebook’s KPIs declares Instagram a failure. Measuring it correctly reveals it as the most efficient awareness channel in the mix. Distinct KPIs per platform are not a refinement. They are a category requirement.

Learning 5: The warm audience built in cold-start months is a paid asset that compounds.

The 653,648 tagged users built in October’s wide-net spend was the foundation that November’s 487 percent engagement jump rested on. Without October, November is mathematically impossible. This means the cold-start month should be evaluated as an audience-building investment, not as a sales month. Agencies that lose accounts in month one because results “are not visible yet” are losing them on the wrong KPI.

Learning 6: A defined buyer persona drives every creative decision.

The buyer in this engagement was identified through the audience data as a 25 to 34 year old urban Sri Lankan male, predominantly Colombo-based, with strong representation in Kandy and Galle. DIY-oriented, contractor-adjacent, time-poor, mobile-first. He discovers content via Reels. He reacts and saves but does not click through. He responds to Sinhala storytelling that respects his intelligence. Every creative produced from November onwards was designed for this person specifically. Generic aspirational lifestyle creative consistently underperformed because it was not aimed at him. Personas, when grounded in real audience data rather than imagined, are the most underused tool in Sri Lankan brand marketing.

What This Engagement Did Not Solve

Honest case studies disclose limits. This one had several.

The engagement built engagement and discovery infrastructure. It did not directly attribute revenue. The brand’s distribution model is distributor-led and largely offline, which means the link between digital engagement and sell-through is correlational rather than causal. Brands operating in similar offline-led distribution categories should expect the same limitation.

The engagement built a warm audience asset inside Meta’s system. The next contract holder of this account inherits that asset. If a future agency restarts the account from scratch, the LKR 86,249 of October cold-start spend will need to be repeated to rebuild what already exists. This is a contractual continuity issue that brands should consider before changing agencies mid-asset.

The engagement did not implement server-side event tracking via Meta CAPI. This means iOS 14 plus signal loss was not mitigated. Audience targeting accuracy would improve approximately 30 to 40 percent with CAPI implemented. This was scoped for phase two of the engagement, which did not activate.

The engagement did not extend to TikTok, Google Search, or Display, all of which are warranted next channels given the audience profile. Sri Lanka has approximately four million TikTok users in the brand’s core demographic. The Sinhala DIY content already proven on Meta would transfer to TikTok at near-zero CPM by relative comparison.

These are not failures. They are scope limits. A case study that does not disclose them is misleading by omission.

What This Means For Your Business

If you are a brand running paid social in Sri Lanka and your results do not look like the trajectory above, the diagnostic question is not “are we spending enough.” It is “are we engineering the spend or running it as activity.”

Most accounts the agency audits show the same structural failures. They use English-first content where Sinhala-first would convert better. They measure Instagram by Facebook’s KPIs. They allow publishing gaps that erase algorithmic momentum. They run wide ad sets at the average rather than ruthlessly killing the bottom quartile. They treat month one as a sales month rather than an audience-building investment. They report on impressions rather than on the audience asset being built.

The methodology that produced 4.05 million views in six months is not proprietary in its components. The Centrality-Distinctiveness Map is public marketing theory. Sinhala-first copy is a cultural observation any local writer can apply. The four-part creative formula is publishable, and you just read it. The Reels insight is now visible to anyone reading platform data.

What is hard to replicate is the discipline of running all of it together, week after week, for six months, while resisting the pressure to chase whatever the client meeting that week wanted to chase.

That discipline is what an agency relationship buys you. Or, if you have an internal team capable of it, that discipline is what produces the results above when you operate it yourself.

If you are a brand spending LKR 250,000 or more per month on marketing and your paid social results do not match the trajectory in this case, the diagnostic is usually visible within the first 30 minutes of a strategic call.

Book a strategic call

The call is 45 minutes. No pitch deck. No upsell. We review your current paid social if you have any, your distribution model, and your real buyer. By the end of it you will know whether paid media is your highest-use move right now, or whether something else in your funnel needs attention first.

If we are wrong for each other, we will say so. If we are right, we will tell you what the engagement would look like, what it would cost, and what it would not solve.