Skip to main content
Back to Blog

How to Calculate True ROAS for Taboola Campaigns

Calculate true ROAS for Taboola campaigns: Realize conversion value and spend, backend orders tied to tblci, refunds, break-even ROAS and worked examples.

21 min read
How to Calculate True ROAS for Taboola Campaigns

Key Takeaways

  • •Taboola ROAS in Realize is tracked conversion value divided by spent, and the conversion value is whatever your purchase event sends: an event with no revenue parameter reports a conversion value of zero and a ROAS that means nothing
  • •True ROAS for a Taboola campaign is the net revenue from backend orders tied to that campaign, each counted once, divided by the campaign's actual spend in the same currency and period. Platform ROAS and true ROAS differ because they use different revenue, different dates and different counting rules
  • •Realize attributes a conversion to the day of the click, not the day of the order, with a click window that defaults to 30 days and a view window of up to 24 hours, so a same-day comparison with your store will disagree even when tracking is perfect
  • •Break-even ROAS is 1 divided by your margin before ad spend: 2.0 at a 50% margin, 2.5 at 40% and 4.0 at 25%. A Taboola campaign below that number loses money on the first order unless repeat purchases repay it
  • •SignalBridge sends Taboola the revenue, currency and order ID from the order and syncs Taboola spend on the Pro plan and above, but it does not send quantity and it cannot correct Taboola revenue for a later refund, so judge Taboola on net backend revenue

How do you calculate true ROAS for Taboola campaigns?

True ROAS for a Taboola campaign is the net revenue from orders in your own backend that you can tie to that campaign, each order counted once, divided by what the campaign actually spent in the same period and currency. Taboola's own ROAS in Realize is simpler: tracked conversion value divided by spent. The two numbers answer different questions, and this guide shows how to build both, why they disagree, and which one to use for a budget decision.

Start with the formulas, because everything after this is about making the inputs honest:

Taboola platform ROAS = Conversion value reported to Taboola ÷ Spent in Realize
True Taboola ROAS     = Net backend revenue tied to the campaign ÷ Actual Taboola spend
True Taboola CPA      = Actual Taboola spend ÷ Backend orders tied to the campaign
Break-even ROAS       = 1 ÷ Margin before ad spend

ROAS is expressed either as a ratio or a percentage depending on the report. A ratio of 3.0 and 300% are the same number: three dollars of revenue for one dollar of spend. Pick one format for your own reporting and keep it, because mixing the two in one spreadsheet is how a 3.0 becomes a 300 in a budget meeting.

We build SignalBridge, which sends server-side conversions to Taboola and can sync its spend. The calculation in this guide does not need SignalBridge or any other tool. It needs three things you already have: your orders, a way to tie an order to a Taboola click, and the campaign's spend. Where SignalBridge changes an input, we say so plainly, including the inputs it cannot fix.

What does ROAS mean inside Taboola Realize?

Before you calculate a truer version, read the platform version correctly. Realize reports these fields per campaign, and the definitions decide what the number can and cannot tell you:

Realize fieldWhat Taboola says it isWhat to check before you trust it
SpentTotal spend for the campaign in the selected periodWhether the period and currency match your backend comparison
ConversionsActions attributed to Taboola under your conversion definitionsWhether the purchase event is the only event in the total you are reading, or one of several funnel events
Conversion ValueThe sum of the values your conversion events sent. If the value parameter is empty, Taboola records zeroWhether every purchase sends a dynamic revenue value, and whether that value includes tax and shipping
CPATotal spent divided by the number of conversionsWhether the conversion in the denominator is a purchase or an earlier funnel step
CVRConversions divided by clicks, expressed as a percentage in Taboola's campaign reportingWhether you are comparing it with a click-based or session-based rate in your analytics. They are not the same denominator
ROASConversion value divided by spent, available in campaign and custom reportingWhether conversion value is complete, deduplicated and in the same currency as spent

Two Realize behaviours explain most arguments between a media buyer and a store owner:

  • Conversions are dated by the click. Taboola's campaign reporting attributes a conversion to when the person clicked the campaign item, not when they bought. A person who clicks on Tuesday and buys on Friday appears in Tuesday's row. Your store reports the order on Friday. Neither is wrong, and a day-by-day comparison will never reconcile.
  • View-through credit is included under its own rules. A conversion can be credited when a person saw a viewable impression, at least 50% in view for over one second, and converted within the view window without clicking. Your backend, which ties orders to a stored click ID, will not see that order as Taboola's unless the same person also clicked.

The click window defaults to 30 days and the view window defaults to 24 hours in Taboola's documented setup. If your account was configured with different windows, use yours and write them down, because a ROAS built on a 30-day click window is not comparable with one built on a 7-day window.

Our native ad conversion tracking guide covers how the tblci click ID, event names and deduplication rules work across Taboola, Outbrain, MGID and Revcontent. This guide assumes that plumbing exists and concentrates on the money.

Step-by-step: how to calculate true ROAS for a Taboola campaign

These steps work at campaign level. Run them for one campaign first, not for the whole account, because the fixes are usually campaign-specific: one landing page that drops tblci, one event with a fixed value, one currency mismatch.

Step 1: Pull the campaign's actual spend

Take spent for the campaign from Realize for the period you are judging, in the account currency, and use the same figure Taboola billed on. Do not rebuild spend from clicks multiplied by an average CPC, because actual CPC is spend divided by clicks after the fact, and rounding in the average becomes a real error at volume.

If you sync spend into another system, check that the synced figure matches Realize for one full day before you calculate anything from it. SignalBridge, for example, reads daily Taboola spend through the Backstage API on the Pro plan and above. A sync is a convenience for reporting; the platform's own spent remains the source you reconcile against.

Step 2: Decide what counts as revenue, and write it down

Choose one revenue definition and use it on both sides of every comparison:

  • Order total as paid, including tax and shipping
  • Order total excluding tax and shipping
  • Net revenue: the chosen total minus refunds, cancellations and chargebacks in the period

For a true ROAS you will use to set budgets, net revenue is the honest numerator, because Taboola keeps counting an order after it is refunded. SignalBridge illustrates the limit: it sends the revenue from the original order to Taboola and does not send refunds or cancellations to Taboola, so the platform's conversion value stays gross. Google Ads is handled differently in SignalBridge, which corrects the original conversion for a full or partial refund, but that correction does not reach Taboola. If refund rates differ between campaigns, a supplement subscription campaign and a one-off purchase campaign for example, gross ROAS will flatter the high-refund campaign. Net both of them.

Decide the tax and shipping question by matching what your purchase event sends. If your event sends the full paid total and your backend comparison uses revenue without shipping, the gap you find is a definition, not a tracking failure.

Step 3: Tie backend orders to the campaign

An order counts toward a Taboola campaign's true ROAS when your own records tie it to that campaign. In practice that tie is one of three strengths:

TieHow you get itWhat it proves
Stored tblci on the orderYour tracking captured the click ID on landing and kept it with the visitor through to the orderThis order began on a Taboola click you can identify
Campaign UTMs on the orderThe landing URL carried utm_source=taboola and a campaign value you maintain, stored with the orderThis order began on a session you labelled as this campaign
Platform-reported conversion onlyRealize says a conversion happened, with no matching backend keyNothing you can audit. Use it for platform optimisation, not for true ROAS

Keep the stored click ID as the primary key and UTMs as the fallback. UTMs are typed by people and copied between campaigns; click IDs are issued by Taboola. Our click ID explainer covers how each platform's ID is captured and where funnels drop it, which matters for Taboola more than most, because native campaigns often run through an advertorial on a different domain before the store.

An order with neither a stored tblci nor Taboola UTMs does not enter this campaign's true ROAS. That makes true ROAS a floor for Taboola's influence, not a complete measure of it. A person who read the advertorial, searched your brand two days later and bought from an organic result was influenced by the campaign and is invisible to this calculation. Only a holdout or lift test measures that effect, which is why true ROAS governs budget shifts between campaigns rather than proving whether Taboola works at all.

Step 4: Count each order once, even when two networks claim it

One buyer can click an Outbrain ad on Monday, a Taboola ad on Wednesday and buy on Friday, with both click IDs stored and both networks inside their windows. Both platforms can then claim the order, and adding platform reports together counts it twice.

Your backend calculation needs one rule. The common choices are:

  • Last paid click before the order. The most recent stored click ID takes the revenue. Simple, and it matches how most platforms argue.
  • First paid click. The campaign that started the journey takes it. Kinder to native and advertorial funnels, which often introduce a product rather than close it.
  • Split credit. Divide the revenue by a fixed rule. More faithful and more work, and rarely worth it below a large spend.

The rule matters less than using the same rule for every channel in the same report. If Taboola gets last-click credit in your spreadsheet and Meta gets its own platform-reported number in the next column, you have not compared two channels; you have compared two definitions. Our true ROAS guide sets out the same discipline across platforms, and the Taboola vs Facebook tracking comparison walks through the overlap between these two networks specifically.

Step 5: Convert currency before you divide

Taboola reads postback revenue in the currency you send and converts it to your account currency. A store selling in euros and pounds into a dollar account therefore has two conversion points: Taboola's and your backend's. If they use different rates or different days, the ROAS gap is exchange arithmetic.

For true ROAS, convert backend revenue into the spend currency with one rate source and one rule, such as the rate on the order date, and keep it for every campaign in the report. SignalBridge sends the order's own currency code with the revenue and leaves the conversion to Taboola; it does not send quantity, so a multi-item order's value has to be the order total you want credited, not a per-item figure your reporting later multiplies.

Step 6: Divide, then calculate true CPA next to it

True ROAS = Net backend revenue tied to the campaign ÷ Actual campaign spend
True CPA  = Actual campaign spend ÷ Backend orders tied to the campaign

Always read the two together. ROAS alone hides scale and order value. A campaign can post the best ROAS in the account on eleven orders while another carries the revenue at a lower ratio. True CPA tells you what an audited order cost; ROAS tells you what an audited dollar returned. Budget decisions need both, plus the order count, so a ratio built on three orders is not treated like one built on three thousand.

A worked example: one Taboola campaign, three ROAS numbers

These are illustrative figures for one campaign over one full week, chosen to show the arithmetic rather than to describe any real account:

MeasureRevenue usedSpendResult
Realize, as reported$18,400 conversion value$5,000ROAS 3.68, CPA $41.67 on 120 conversions
Backend orders tied to the campaign (tblci or UTM)$15,900 gross$5,000True ROAS (gross) 3.18
The same orders after refunds and cancellations$14,700 net$5,000True ROAS (net) 2.94, true CPA $46.30 on 108 orders

Read the gaps in order. Realize counts 120 conversions; the backend ties 108 orders to the campaign. The missing 12 are the usual mixture: view-through credit your click-based backend cannot see, orders another channel took under your last-click rule, and any conversion your deduplication or consent rules held back. The revenue gap is larger than the count gap because Realize holds gross values, including two refunded orders and the tax and shipping your backend definition excludes.

None of the three numbers is lying. Realize's 3.68 is what Taboola's bidding worked from. The net 2.94 is what you should compare with your break-even ROAS before you raise the budget. If this campaign's margin before ad spend is 40%, break-even is 2.5, so at a true ROAS of 2.94 the campaign clears its first-order break-even with a small margin. At a 30% margin, break-even is 3.33, and the same campaign loses money on first orders whatever Realize displays.

What is a good ROAS for Taboola?

There is no Taboola-specific good ROAS. There is only your break-even ROAS and the margin you need above it. Break-even depends solely on the margin before ad spend, which is what remains from revenue after cost of goods, shipping, packaging, payment fees and a refund allowance:

Margin before ad spendBreak-even ROAS
25%4.00
30%3.33
40%2.50
50%2.00
60%1.67

If you want a 10% profit margin on a product with a 40% margin before ads, your target ROAS is 1 divided by (0.40 minus 0.10), which is 3.33. A campaign can also run below first-order break-even on purpose when repeat purchases repay the acquisition cost, through a subscription or a short replenishment cycle. That is a lifetime value decision, and it should be made with the repeat revenue written into the calculation, not by pointing at a platform ROAS that happens to look higher. The MER, CAC and LTV guide covers that blended view, and our guide to reducing CPA covers the cost side of the same equation.

Compare Taboola campaigns with each other before you compare Taboola with other channels. Two Taboola campaigns share a window, a counting rule and a platform; a Taboola campaign and a Meta campaign share none of those by default. Channel-level decisions belong to blended measures that need no attribution agreement at all, such as blended ROAS and MER.

Why does Taboola ROAS read too high or too low?

When platform ROAS and your backend calculation disagree by more than reporting delay and window effects explain, work through these causes in this order:

SymptomLikely causeFix
Conversion value is zero or ROAS is zero with conversions presentThe purchase event sends no revenue parameter, or a fixed value instead of the order totalSend the dynamic order value on every purchase and confirm the conversion is included in total conversion value
Platform ROAS higher than backendRefunded and cancelled orders still counted at gross valueJudge the campaign on net backend revenue and track the refund rate per campaign
Platform ROAS higher than backendThe same order counted by pixel and postback with different valuesKeep order ID, revenue and currency identical on both routes. Taboola keeps the server copy only when the values match
Platform ROAS higher than backendAnother network also claims the same orders in your comparisonAssign each backend order to one channel under one rule in your own reporting
Platform ROAS lower than backendPurchases never reached Taboola because tblci was dropped by a redirect, advertorial domain or link shortenerCapture the click ID on the first page, store it server-side and forward the query string at every hop
Platform ROAS lower than backendConsent held conversions back, or bot filtering removed events you expectedTest one order with consent granted and check what your filter classified before you change it
Numbers disagree only on daily rowsTaboola dates conversions by click day; your store dates them by order dayCompare full weeks, and compare a mature period that gives the 30-day click window time to fill

The last row deserves emphasis during any sale or launch. A campaign launched this week has had no time to accumulate conversions inside a 30-day window, so its early platform ROAS is structurally understated against a mature campaign. Judge a new campaign on click-tied backend orders for its first weeks, and let platform ROAS take over only when the window has filled.

How do you improve the inputs, instead of arguing about the output?

True ROAS gets more useful when the revenue Taboola receives matches the revenue you calculate from. Four inputs decide that:

  • Send a dynamic value with every purchase. A fixed conversion value makes every order worth the same amount to Realize, so its ROAS, its CPA optimisation and any value-based bidding all work from fiction. Send the same order total you use in your backend, with its currency and order ID. Taboola's own guidance makes the same point: a conversion contributes to conversion value and ROAS only when the value is sent and the conversion is included in the total.
  • Run the pixel and the postback as one system. Taboola recommends both and deduplicates the pair when the click, event and values agree. Two routes with different values are not redundancy; they are two different claims. Our Taboola server-side tracking guide covers the postback mechanics and Taboola Conversions API setup covers the no-code version.
  • Protect the click ID across the funnel. Store tblci on your server at landing rather than only in a script-written cookie, which Safari can cap at 24 hours after a click-ID landing and clear after 7 days without a visit. A 30-day click window is only useful if your copy of the ID lives that long.
  • Keep fake conversions out. Taboola's bidding learns from the conversions you report, so a bot order or a test order left in the data is a training signal. Filter non-human sessions before the postback, and exclude your own test orders from both sides of the calculation.

What SignalBridge sends Taboola, and what it cannot fix

SignalBridge captures tblci on landing, keeps it for the visitor, and on a purchase sends Taboola a server-to-server postback with the click ID, the event name, revenue, currency and order ID. Only an HTTP 204 response counts as delivered; temporary failures are retried from a durable queue, suspected bot events are filtered before delivery, and conversions are held back when marketing consent is denied or required and not granted. On the Pro plan and above it also reads daily Taboola spend through the Backstage API, so a dashboard ROAS uses synced spend rather than a typed figure. The Taboola integration page lists the event names and limits.

Three limits belong in any ROAS discussion. SignalBridge does not send quantity, so the revenue you send has to be the order total. It does not send refunds or cancellations to Taboola, so Realize revenue stays gross while your true ROAS should be net. And it cannot report an order with no stored tblci, so a funnel that drops the click ID before SignalBridge loads will understate Taboola no matter how clean the postback is. Taboola counts as one connected platform on plans that count platforms, and pricing starts at $29 a month for 20,000 events, with spend sync on Pro at $129 a month and above. Start your 14-day free trial today. No credit card required.

If you already send Taboola complete dynamic values through another route, keep it and use this guide's backend calculation as the audit. The spreadsheet is the product here, not the sender.

FAQ

How is Taboola ROAS calculated in Realize?

Realize calculates ROAS as tracked conversion value divided by spent for the selected period. Conversion value is the sum of the values your conversion events send, so a purchase event that sends no revenue parameter contributes zero and the ROAS is meaningless even when the conversion count is right. Taboola attributes the conversion to the day of the click under your click and view windows, not to the day the order was placed.

What is true ROAS for a Taboola campaign?

It is the net revenue from backend orders tied to that campaign, each counted once, divided by the campaign's actual spend in the same period and currency. Orders are tied by a stored tblci where you have one and by campaign UTMs as a fallback. Refunds and cancellations are removed, because Taboola continues to count the original order at its gross value.

Why is my Taboola ROAS higher than my store's numbers?

The common causes are refunded orders still counted at full value, a purchase counted by both the pixel and the postback because the values sent differed, view-through credit your click-based backend does not reproduce, and tax or shipping included in the platform value but excluded from your backend definition. Compare full weeks of net backend revenue tied to the campaign with Realize spend, rather than comparing daily rows.

Why is my Taboola ROAS lower than my store's numbers?

Purchases are reaching your store without reaching Taboola. Check whether tblci survives every redirect and domain in the funnel, whether the ID is stored beyond the short life of a script-written cookie in Safari, whether consent rules held the conversion back, and whether the event name in your postback matches the Event Name in Realize exactly, including case. A mismatch in any one of those removes the conversion while the order still appears in your backend.

What is a good ROAS for Taboola campaigns?

One that clears your break-even ROAS, which is 1 divided by your margin before ad spend: 2.0 at a 50% margin, 2.5 at 40% and 4.0 at 25%. Below break-even, a campaign loses money on first orders unless repeat purchases repay it. There is no Taboola-wide benchmark that can substitute for your own margin, because two stores with the same platform ROAS keep different amounts of every dollar.

Should I use platform ROAS or true ROAS to set Taboola budgets?

Use both, for different decisions. Platform ROAS is what Taboola's bidding sees, so watch it for delivery and optimisation changes inside the platform. Use true ROAS, built from net backend revenue and actual spend, to move budgets between campaigns or channels, because it applies one revenue definition and one counting rule to everything it compares. A steady gap between the two is a tracking problem to fix before it is a budget signal.

Does SignalBridge calculate Taboola ROAS for me?

SignalBridge sends Taboola the purchase postback with revenue, currency and order ID, and on the Pro plan and above it syncs daily Taboola spend through the Backstage API so cost and return reporting use actual spend. It does not send quantity and it does not correct Taboola for refunds, so a refund-heavy campaign should still be judged on the net backend calculation in this guide rather than on Realize revenue alone.

Sources

Ready to recover more conversions?

Start tracking what your pixels miss. Set up in 5 minutes, no credit card required.

Start Free Trial