A contextual advert is chosen by what surrounds it, not by who’s it. When an advert slot masses inside a phrase puzzle app, the system reads indicators about that second: the app’s class, the language of the content material, the machine, and the nation. It picks an advert that matches these indicators. It doesn’t seek the advice of a saved profile of the individual holding the telephone.
That single distinction attracts the road between the two important approaches in cell promoting:
- Contextual concentrating on reads the location. The query it solutions is “the place is that this advert displaying, and what’s round it proper now?”
- Behavioral concentrating on reads the individual. The query it solutions is “what has this consumer performed earlier than, throughout apps and periods, and what does that predict?”
Behavioral concentrating on wants a steady identifier to tie these previous actions collectively: on iOS that was the IDFA, on Android the GAID. Contextual concentrating on wants none of that. The match occurs from knowledge in regards to the content material and the request, which is current whether or not or not a consumer has agreed to monitoring.
A concrete instance
Image a free racing sport in Germany. A participant opens it on an iPhone, and an interstitial slot turns into obtainable between ranges. The bid request that goes out to patrons carries fields like app class (Video games), key phrases pulled from the app’s retailer itemizing and content material, machine sort (cell, iOS), language (de), and a rough geo (Germany). A purchaser working a marketing campaign for a brand new racing title can match on these fields alone. The artistic proven strains up with the context: a racing advert inside a racing sport, in German, on the proper machine.
No a part of that call required figuring out the participant’s age, previous purchases, or which apps they used yesterday. The relevance comes from the environment, not from a historical past.
Behavioral concentrating on would method the identical slot otherwise. It might lookup the machine ID, pull what that ID has performed, a consumer who put in 3 racing video games final month and spent in 1 of them, and bid based mostly on that sample. When the sign is out there, it may be extra exact. The price is that it will depend on the consumer having an ID that patrons can learn, and on that ID being shared throughout the apps in query.
Contextual isn’t a watered-down model of behavioral. It solutions a distinct query. A racing advert positioned subsequent to racing content material can carry out properly as a result of intent is seen within the second: somebody taking part in a racing sport is, proper then, all in favour of racing. The sign is present relatively than historic.
The sections that observe present which fields carry these indicators in an actual bid request, why contextual moved from a fallback to a major technique after the ID modifications of 2021 onward, and the place it suits towards behavioral concentrating on in a working UA stack.
How contextual concentrating on works underneath the hood
The match runs inside an public sale that finishes in underneath 100 milliseconds. When an advert slot masses, the writer’s SSP packages what it is aware of in regards to the placement right into a bid request and sends it to demand companions. Most of cell programmatic runs on the OpenRTB spec from the IAB, so the fields in that request are standardized. Consumers learn them, resolve whether or not their marketing campaign suits, and return a value. The entire trade occurs earlier than the interstitial or banner finishes loading.
Contextual concentrating on works as a result of the request already carries sufficient in regards to the placement to decide with out a consumer ID. Listed below are the indicators that matter most and the place they reside within the request.
The fields that carry context
- App class. In OpenRTB that is the app.cat discipline, populated with IAB content material taxonomy codes (for instance IAB9-30 for video and pc video games). A purchaser can filter to sport classes and skip every little thing else.
- Content material key phrases. The app.key phrases and app.content material.key phrases fields maintain phrases pulled from the shop itemizing, the app title, and generally the display screen’s content material. A racing sport would possibly carry “racing”, “vehicles”, “drift”, “multiplayer”.
- Gadget sort and OS. The machine object experiences make, mannequin, OS, and OS model. A marketing campaign that solely has an iOS artistic filters on machine.os = iOS; a low-end Android marketing campaign can filter on mannequin.
- Language. The machine.language discipline (and generally content material.language) provides an ISO code like de or tr. This decides whether or not a German artistic or an English one will get served.
- Geo. The machine.geo object experiences nation, and sometimes area and a rough lat/lengthy. With out consent, that is normally restricted to nation and area stage relatively than exact coordinates.
- Connection and provider. The machine.connectiontype discipline separates Wi-Fi from mobile. A marketing campaign with a 30 MB playable would possibly solely bid on Wi-Fi to keep away from losing installs on customers who abandon the obtain.
None of those fields require the IDFA or GAID. They describe the location and the machine, not a saved historical past of the individual. That’s the structural cause contextual survives consent loss: the info it reads is current within the request whether or not or not the consumer opted into monitoring.
How a match will get made
Take a marketing campaign for a brand new racing title. The customer units up concentrating on on the DSP: class in video games, key phrases embody “racing” or “vehicles”, OS is iOS, geo is Germany, language is German. When the bid request for the racing sport slot arrives, the DSP checks every rule towards the fields:
- app.cat comprises a video games code. Move.
- app.key phrases comprises “racing”. Move.
- machine.os is iOS. Move.
- machine.geo.nation is DEU. Move.
- machine.language is de. Move.
All guidelines cross, so the DSP returns a bid with the German racing artistic hooked up. The SSP runs its public sale throughout all of the bids it received again, picks the winner by value, and the artistic renders. The choice used 5 fields and nil identification knowledge.
The identical equipment handles a no-match cleanly. If the request got here from a finance app in France with French because the language, the key phrase rule and the geo rule each fail, so the racing marketing campaign doesn’t bid. The slot goes to whichever marketing campaign does match. That is why contextual attain is huge: any purchaser whose guidelines match the location can compete, together with for impressions the place the consumer declined monitoring.
Key phrase high quality decides relevance
The weakest a part of contextual is key phrase knowledge, and it varies by supply. Retailer-listing key phrases are dependable however coarse, they describe the entire app, not the present display screen. Some SSPs add page- or screen-level content material indicators by means of SDKs that learn what’s on show, which sharpens the match however provides integration work. A puzzle app whose retailer itemizing says “mind video games” and “logic” provides a purchaser one thing to focus on; an app that passes solely a class code and no key phrases provides a lot much less.
Consumers deal with this by combining indicators relatively than counting on one. Class plus language plus geo already narrows a marketing campaign to a usable viewers. Key phrases add precision on high. The estimated impact is {that a} marketing campaign matching on class, geo, and language reaches a broad set of placements, whereas including key phrase guidelines trims that set to essentially the most related slots at the price of quantity.
What the request doesn’t carry
Underneath contextual, the request doesn’t embody a tool ID a purchaser can use to lookup previous conduct, a listing of different apps put in, or a frequency historical past tied to the individual. It carries the second and the environment. If a purchaser must know that this machine put in 3 racing video games final month, contextual can not inform them. That restrict is the topic of the behavioral comparability later. For the public sale itself, the fields above are sufficient to position a related advert and return a value within the time it takes the slot to load.
Why contextual received greater after IDFA modifications
For a lot of the final decade, behavioral concentrating on was the default in cell. Consumers learn a tool ID, regarded up what that ID had performed, and bid on the sample. That mannequin assumed the ID was there to learn. In 2021 that assumption stopped holding.
What ATT modified on iOS
Apple launched App Monitoring Transparency with iOS 14.5 in April 2021. From that time, an app has to point out a immediate earlier than it might entry the IDFA, and the consumer has to faucet “Permit.” If they refuse, the IDFA returns a string of zeros, and the machine is successfully unaddressable for cross-app monitoring.
The opt-in price landed decrease than most patrons anticipated. Measurement companies that monitor the immediate have reported app opt-in charges clustering within the 20 to 30 p.c vary, with Regulate and AppsFlyer figures over 2021 to 2022 placing the typical close to 25 p.c for a lot of app classes. Which means for roughly 3 out of 4 iOS customers, there is no such thing as a IDFA to learn. A concentrating on technique that will depend on that ID loses most of its viewers on iOS earlier than the public sale even begins.
The impact was not even throughout the stack. Behavioral retargeting, which wants an ID to acknowledge a returning consumer, took the toughest hit. Campaigns that relied on constructing lookalike audiences from previous conduct misplaced the uncooked materials these fashions ran on.
What is going on on Android
Android stored the GAID addressable for longer, so the strain there was slower however shifting in the identical course. Google added a setting in Android 12 that lets customers reset the promoting ID to zeros, much like the iOS opt-out. Google’s Privateness Sandbox for Android, introduced in 2022, is the bigger shift: it replaces cross-app identifiers with on-device APIs like Subjects and Protected Viewers, which expose curiosity classes relatively than a steady per-device ID. The timeline has moved greater than as soon as, however the course is about. Consumers mustn’t plan on a sturdy cross-app ID on Android both.
Why contextual was the sign left standing
Contextual concentrating on reads the location, not the individual, so none of those modifications contact it. The fields that carry context, app class, content material key phrases, language, machine sort, and coarse geo, sit within the bid request whether or not or not the consumer granted monitoring consent. A racing sport in Germany sends the identical app.cat and language fields to patrons no matter how that participant answered the ATT immediate. The match nonetheless works.
That property turned contextual from a fallback right into a major technique. Earlier than 2021 it was typically the factor you fell again on when an ID was lacking. After ATT, the lacking ID turned the frequent case relatively than the exception, so the tactic that didn’t want an ID moved to the entrance.
The shift reveals up in spend. The IAB and different trade trackers reported rising contextual budgets by means of 2021 and 2022 as patrons redirected cash away from ID-dependent ways. The precise figures range by supply, so deal with any single share as estimated, however the course is constant throughout experiences: contextual share went up as addressable stock went down.
There’s a second cause it held up. A contextual sign reads intent within the second. Somebody taking part in a racing sport proper now’s all in favour of racing proper now, and that sign doesn’t depend upon a consent immediate or a saved historical past. When the ID disappeared, the present sign stayed readable, and patrons leaned on it.
Contextual vs behavioral concentrating on: the tradeoffs
Neither technique is strictly higher. They commerce the identical 3 issues towards one another: privateness publicity, attain, and precision. The place a marketing campaign lands on these axes decides which one to run.
The place contextual wins
Privateness and consent. Contextual reads the location, so it really works with out an opt-in. On iOS, that issues as a result of the App Monitoring Transparency immediate is the gate for studying the IDFA, and opt-in charges have sat close to 25 p.c throughout client apps since ATT launched (AppsFlyer and Regulate each reported figures in that vary by means of 2022 to 2023). Behavioral concentrating on can solely handle the consenting minority. Contextual addresses each impression.
Attain. As a result of the bid request carries app class, key phrases, language, and geo whether or not or not the consumer agreed to monitoring, contextual patrons can bid on the complete stock pool. Behavioral patrons bid on the slice that carries a readable ID, which on iOS is roughly 1 in 4 impressions. That hole in addressable quantity is the only largest sensible distinction between the two approaches right this moment.
No identification infrastructure. Contextual wants no ID graph, no cross-app linking, and no profile retailer. The match is computed from fields already within the request, so there’s much less to take care of and fewer regulatory floor underneath GDPR and related guidelines.
The place behavioral wins
Retargeting. If you wish to attain a consumer who put in your app, churned, and by no means got here again, you have to determine that particular machine once more. Contextual can not do that. It has no reminiscence of who noticed the advert earlier than. Re-engagement campaigns, lapsed-payer winbacks, and frequency capping throughout apps all depend upon a steady identifier.
Predicting worth from historical past. Behavioral concentrating on can favor a tool that put in 3 video games in a style and spent in 1 of them. That sample predicts a better likelihood of one other paid set up. Contextual sees solely the present placement, so it can not separate a possible spender from a possible non-spender inside the identical app. When the aim is return on advert spend relatively than uncooked installs, that historic sign carries actual weight.
Lookalike modeling. Constructing an viewers that resembles your present payers requires profiles to mannequin towards. With out IDs, the mannequin has nothing to be taught from on the consumer stage.
The circumstances that favor every
Run contextual when:
- You might be buying new customers at scale and intent is seen within the second, reminiscent of a racing advert inside a racing sport.
- Your stock is usually iOS, the place opt-in charges restrict behavioral attain.
- You use in markets with strict consent enforcement and wish to cut back compliance danger.
Run behavioral when:
- You might be retargeting recognized customers or capping frequency throughout apps.
- You optimize towards downstream income and want historic buy indicators.
- Your site visitors is Android with readable GAIDs, or iOS customers who opted in.
Most UA groups don’t choose one. They run behavioral the place IDs exist and consent is granted, they usually fall again to contextual for the bigger share of impressions that arrive with out an identifier. The sign you’ve obtainable at public sale time, greater than any choice, units which technique can bid.
The place contextual suits in a UA stack
Contextual placements don’t change rewarded and offerwall stock. They run beside them, filling slots that these codecs don’t cowl. A typical cell UA combine has 3 layers, and every solutions a distinct want.
- Rewarded video and offerwall. The consumer opts into an trade: watch an advert or full an motion, get forex or an merchandise again. Engagement is excessive as a result of the consumer selected it. These slots carry their very own context, the app class and the reward sort, so contextual indicators apply right here too.
- Interstitial and banner. These load between ranges or on the edges of the display screen with out an specific opt-in. Contextual concentrating on does a lot of the work right here, since many of those impressions arrive with out a usable machine ID.
- Contextual show throughout the open trade. Customary programmatic stock matched on app class, key phrases, language, and geo, the fields coated earlier.
The purpose of working them collectively: rewarded and offerwall offer you measured consumer intent, whereas contextual interstitials offer you attain into the identical audiences when no ID is current. On iOS, the place ATT opt-in sits close to 25 p.c by most public estimates, the no-ID portion of your stock is almost all. Contextual is what addresses it.
What you’ll be able to measure with out machine IDs
Dropping the IDFA doesn’t imply shedding measurement. It means altering what you depend and the way.
- SKAdNetwork and AdAttributionKit. Apple’s frameworks report installs and a rough conversion worth again to the advert community with out exposing the consumer. You get mixture counts per marketing campaign, not per-person paths.
- Combination and geo-level carry. Maintain out a area or a placement, evaluate set up and spend curves towards a matched group, and skim the distinction. This works with none consumer identifier.
- Placement-level metrics. Fill price, win price, eCPM, and viewability all connect to the slot, not the individual. They inform you whether or not a contextual purchase is reaching the stock you needed on the value you set.
- In-app occasion modeling. The place you’ve first-party consent, your personal SDK nonetheless logs purchases and retention. You tie these to campaigns by means of aggregated and modeled attribution relatively than a shared cross-app ID.
Setup selections that have an effect on efficiency
A couple of decisions at configuration time transfer outcomes greater than artistic does.
- Class and key phrase concentrating on width. Slim concentrating on raises relevance and lowers attain. Begin wider than you suppose, learn which app classes convert, then tighten. A racing marketing campaign that solely buys IAB sport codes might miss puzzle and informal apps the place racing followers additionally sit.
- Ground costs and bid pacing. Contextual stock clears at decrease CPMs than ID-based retargeting in lots of circumstances, so a flooring set for behavioral buys can starve a contextual line of quantity.
- Conversion worth mapping. SKAdNetwork provides you a restricted set of values. Determine early whether or not they encode an early motion (registration) or a income band. You’ll be able to’t simply change this mid-flight with out shedding comparability.
- Inventive tied to context. The match is just pretty much as good because the artistic behind it. A racing advert in a racing app earns its placement; a generic artistic wastes it.
Deal with contextual because the layer that retains spending environment friendly when the ID is lacking, measured by means of mixture frameworks and placement indicators, and tuned by means of concentrating on width and flooring relatively than consumer lookups.
